blob: e5a249736ddbb8aabaf9ba814960ff2e918f52f8 [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 Tarreau9dc6b972019-06-16 21:49:47 +02005 version 2.1
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreaudb514072019-07-16 19:15:28 +02007 2019/07/16
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
Willy Tarreauc57f0e22009-05-10 13:12:33 +020055
564. Proxies
574.1. Proxy keywords matrix
584.2. Alphabetically sorted keywords reference
59
Davor Ocelice9ed2812017-12-25 17:49:28 +0100605. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200615.1. Bind options
625.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200635.3. Server DNS resolution
645.3.1. Global overview
655.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020066
Willy Tarreau74ca5042013-06-11 23:12:07 +0200677. Using ACLs and fetching samples
687.1. ACL basics
697.1.1. Matching booleans
707.1.2. Matching integers
717.1.3. Matching strings
727.1.4. Matching regular expressions (regexes)
737.1.5. Matching arbitrary data blocks
747.1.6. Matching IPv4 and IPv6 addresses
757.2. Using ACLs to form conditions
767.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200777.3.1. Converters
787.3.2. Fetching samples from internal states
797.3.3. Fetching samples at Layer 4
807.3.4. Fetching samples at Layer 5
817.3.5. Fetching samples from buffer contents (Layer 6)
827.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +0200837.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020084
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200856. Cache
866.1. Limitation
876.2. Setup
886.2.1. Cache section
896.2.2. Proxy section
90
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200918. Logging
928.1. Log levels
938.2. Log formats
948.2.1. Default log format
958.2.2. TCP log format
968.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +0100978.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +0100988.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200998.3. Advanced logging options
1008.3.1. Disabling logging of external tests
1018.3.2. Logging before waiting for the session to terminate
1028.3.3. Raising log level upon errors
1038.3.4. Disabling logging of successful connections
1048.4. Timing events
1058.5. Session state at disconnection
1068.6. Non-printable characters
1078.7. Capturing HTTP cookies
1088.8. Capturing HTTP headers
1098.9. Examples of logs
110
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001119. Supported filters
1129.1. Trace
1139.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001149.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001159.4. Cache
Christopher Fauletb30b3102019-09-12 23:03:09 +02001169.5. fcgi-app
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200117
Christopher Fauletb30b3102019-09-12 23:03:09 +020011810. FastCGI applications
11910.1. Setup
12010.1.1. Fcgi-app section
12110.1.2. Proxy section
12210.1.3. Example
12310.2. Default parameters
12410.3. Limitations
125
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200126
1271. Quick reminder about HTTP
128----------------------------
129
Davor Ocelice9ed2812017-12-25 17:49:28 +0100130When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131fully analyzed and indexed, thus it becomes possible to build matching criteria
132on almost anything found in the contents.
133
134However, it is important to understand how HTTP requests and responses are
135formed, and how HAProxy decomposes them. It will then become easier to write
136correct rules and to debug existing configurations.
137
138
1391.1. The HTTP transaction model
140-------------------------------
141
142The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100143to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100144from the client to the server, a request is sent by the client through the
145connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200146will involve a new connection :
147
148 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
149
150In this mode, called the "HTTP close" mode, there are as many connection
151establishments as there are HTTP transactions. Since the connection is closed
152by the server after the response, the client does not need to know the content
153length.
154
155Due to the transactional nature of the protocol, it was possible to improve it
156to avoid closing a connection between two subsequent transactions. In this mode
157however, it is mandatory that the server indicates the content length for each
158response so that the client does not wait indefinitely. For this, a special
159header is used: "Content-length". This mode is called the "keep-alive" mode :
160
161 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
162
163Its advantages are a reduced latency between transactions, and less processing
164power required on the server side. It is generally better than the close mode,
165but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200166a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200167
Willy Tarreau95c4e142017-11-26 12:18:55 +0100168Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169keep-alive, but the client does not wait for the first response to send the
170second request. This is useful for fetching large number of images composing a
171page :
172
173 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
174
175This can obviously have a tremendous benefit on performance because the network
176latency is eliminated between subsequent requests. Many HTTP agents do not
177correctly support pipelining since there is no way to associate a response with
178the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100179server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200180
Willy Tarreau95c4e142017-11-26 12:18:55 +0100181The next improvement is the multiplexed mode, as implemented in HTTP/2. This
182time, each transaction is assigned a single stream identifier, and all streams
183are multiplexed over an existing connection. Many requests can be sent in
184parallel by the client, and responses can arrive in any order since they also
185carry the stream identifier.
186
Willy Tarreau70dffda2014-01-30 03:07:23 +0100187By default HAProxy operates in keep-alive mode with regards to persistent
188connections: for each connection it processes each request and response, and
189leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100190start of a new request. When it receives HTTP/2 connections from a client, it
191processes all the requests in parallel and leaves the connection idling,
192waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200193
Christopher Faulet315b39c2018-09-21 16:26:19 +0200194HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100195 - keep alive : all requests and responses are processed (default)
196 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100197 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100198 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200199 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100200
Davor Ocelice9ed2812017-12-25 17:49:28 +0100201For HTTP/2, the connection mode resembles more the "server close" mode : given
202the independence of all streams, there is currently no place to hook the idle
Willy Tarreau95c4e142017-11-26 12:18:55 +0100203server connection after a response, so it is closed after the response. HTTP/2
204is only supported for incoming connections, not on connections going to
205servers.
206
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207
2081.2. HTTP request
209-----------------
210
211First, let's consider this HTTP request :
212
213 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100214 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200215 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
216 2 Host: www.mydomain.com
217 3 User-agent: my small browser
218 4 Accept: image/jpeg, image/gif
219 5 Accept: image/png
220
221
2221.2.1. The Request line
223-----------------------
224
225Line 1 is the "request line". It is always composed of 3 fields :
226
227 - a METHOD : GET
228 - a URI : /serv/login.php?lang=en&profile=2
229 - a version tag : HTTP/1.1
230
231All of them are delimited by what the standard calls LWS (linear white spaces),
232which are commonly spaces, but can also be tabs or line feeds/carriage returns
233followed by spaces/tabs. The method itself cannot contain any colon (':') and
234is limited to alphabetic letters. All those various combinations make it
235desirable that HAProxy performs the splitting itself rather than leaving it to
236the user to write a complex or inaccurate regular expression.
237
238The URI itself can have several forms :
239
240 - A "relative URI" :
241
242 /serv/login.php?lang=en&profile=2
243
244 It is a complete URL without the host part. This is generally what is
245 received by servers, reverse proxies and transparent proxies.
246
247 - An "absolute URI", also called a "URL" :
248
249 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
250
251 It is composed of a "scheme" (the protocol name followed by '://'), a host
252 name or address, optionally a colon (':') followed by a port number, then
253 a relative URI beginning at the first slash ('/') after the address part.
254 This is generally what proxies receive, but a server supporting HTTP/1.1
255 must accept this form too.
256
257 - a star ('*') : this form is only accepted in association with the OPTIONS
258 method and is not relayable. It is used to inquiry a next hop's
259 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100260
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200261 - an address:port combination : 192.168.0.12:80
262 This is used with the CONNECT method, which is used to establish TCP
263 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
264 other protocols too.
265
266In a relative URI, two sub-parts are identified. The part before the question
267mark is called the "path". It is typically the relative path to static objects
268on the server. The part after the question mark is called the "query string".
269It is mostly used with GET requests sent to dynamic scripts and is very
270specific to the language, framework or application in use.
271
Willy Tarreau95c4e142017-11-26 12:18:55 +0100272HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100273assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100274However, haproxy natively processes HTTP/1.x requests and headers, so requests
275received over an HTTP/2 connection are transcoded to HTTP/1.1 before being
276processed. This explains why they still appear as "HTTP/1.1" in haproxy's logs
277as well as in server logs.
278
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200279
2801.2.2. The request headers
281--------------------------
282
283The headers start at the second line. They are composed of a name at the
284beginning of the line, immediately followed by a colon (':'). Traditionally,
285an LWS is added after the colon but that's not required. Then come the values.
286Multiple identical headers may be folded into one single line, delimiting the
287values with commas, provided that their order is respected. This is commonly
288encountered in the "Cookie:" field. A header may span over multiple lines if
289the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
290define a total of 3 values for the "Accept:" header.
291
Davor Ocelice9ed2812017-12-25 17:49:28 +0100292Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200293their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100294"Connection:" header). In HTTP/2, header names are always sent in lower case,
295as can be seen when running in debug mode.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200296
297The end of the headers is indicated by the first empty line. People often say
298that it's a double line feed, which is not exact, even if a double line feed
299is one valid form of empty line.
300
301Fortunately, HAProxy takes care of all these complex combinations when indexing
302headers, checking values and counting them, so there is no reason to worry
303about the way they could be written, but it is important not to accuse an
304application of being buggy if it does unusual, valid things.
305
306Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000307 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200308 in the middle of headers by LWS in order to join multi-line headers. This
309 is necessary for proper analysis and helps less capable HTTP parsers to work
310 correctly and not to be fooled by such complex constructs.
311
312
3131.3. HTTP response
314------------------
315
316An HTTP response looks very much like an HTTP request. Both are called HTTP
317messages. Let's consider this HTTP response :
318
319 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100320 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200321 1 HTTP/1.1 200 OK
322 2 Content-length: 350
323 3 Content-Type: text/html
324
Willy Tarreau816b9792009-09-15 21:25:21 +0200325As a special case, HTTP supports so called "Informational responses" as status
326codes 1xx. These messages are special in that they don't convey any part of the
327response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100328continue to post its request for instance. In the case of a status 100 response
329the requested information will be carried by the next non-100 response message
330following the informational one. This implies that multiple responses may be
331sent to a single request, and that this only works when keep-alive is enabled
332(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
333correctly forward and skip them, and only process the next non-100 response. As
334such, these messages are neither logged nor transformed, unless explicitly
335state otherwise. Status 101 messages indicate that the protocol is changing
336over the same connection and that haproxy must switch to tunnel mode, just as
337if a CONNECT had occurred. Then the Upgrade header would contain additional
338information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200339
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200340
Davor Ocelice9ed2812017-12-25 17:49:28 +01003411.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200342------------------------
343
344Line 1 is the "response line". It is always composed of 3 fields :
345
346 - a version tag : HTTP/1.1
347 - a status code : 200
348 - a reason : OK
349
350The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100351 - 1xx = informational message to be skipped (e.g. 100, 101)
352 - 2xx = OK, content is following (e.g. 200, 206)
353 - 3xx = OK, no content following (e.g. 302, 304)
354 - 4xx = error caused by the client (e.g. 401, 403, 404)
355 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200356
Lukas Tribus23953682017-04-28 13:24:30 +0000357Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100358"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200359found there, but it's a common practice to respect the well-established
360messages. It can be composed of one or multiple words, such as "OK", "Found",
361or "Authentication Required".
362
Davor Ocelice9ed2812017-12-25 17:49:28 +0100363HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200364
365 Code When / reason
366 200 access to stats page, and when replying to monitoring requests
367 301 when performing a redirection, depending on the configured code
368 302 when performing a redirection, depending on the configured code
369 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100370 307 when performing a redirection, depending on the configured code
371 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200372 400 for an invalid or too large request
373 401 when an authentication is required to perform the action (when
374 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200375 403 when a request is forbidden by a "http-request deny" rule
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200376 408 when the request timeout strikes before the request is complete
377 500 when haproxy encounters an unrecoverable internal error, such as a
378 memory allocation failure, which should never happen
379 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200380 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381 503 when no server was available to handle the request, or in response to
382 monitoring requests which match the "monitor fail" condition
383 504 when the response timeout strikes before the server responds
384
385The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3864.2).
387
388
3891.3.2. The response headers
390---------------------------
391
392Response headers work exactly like request headers, and as such, HAProxy uses
393the same parsing function for both. Please refer to paragraph 1.2.2 for more
394details.
395
396
3972. Configuring HAProxy
398----------------------
399
4002.1. Configuration file format
401------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200402
403HAProxy's configuration process involves 3 major sources of parameters :
404
405 - the arguments from the command-line, which always take precedence
406 - the "global" section, which sets process-wide parameters
407 - the proxies sections which can take form of "defaults", "listen",
408 "frontend" and "backend".
409
Willy Tarreau0ba27502007-12-24 16:55:16 +0100410The configuration file syntax consists in lines beginning with a keyword
411referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200412delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100413
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200414
William Lallemandf9873ba2015-05-05 17:37:14 +02004152.2. Quoting and escaping
416-------------------------
417
418HAProxy's configuration introduces a quoting and escaping system similar to
419many programming languages. The configuration file supports 3 types: escaping
420with a backslash, weak quoting with double quotes, and strong quoting with
421single quotes.
422
423If spaces have to be entered in strings, then they must be escaped by preceding
424them by a backslash ('\') or by quoting them. Backslashes also have to be
425escaped by doubling or strong quoting them.
426
427Escaping is achieved by preceding a special character by a backslash ('\'):
428
429 \ to mark a space and differentiate it from a delimiter
430 \# to mark a hash and differentiate it from a comment
431 \\ to use a backslash
432 \' to use a single quote and differentiate it from strong quoting
433 \" to use a double quote and differentiate it from weak quoting
434
435Weak quoting is achieved by using double quotes (""). Weak quoting prevents
436the interpretation of:
437
438 space as a parameter separator
439 ' single quote as a strong quoting delimiter
440 # hash as a comment start
441
William Lallemandb2f07452015-05-12 14:27:13 +0200442Weak quoting permits the interpretation of variables, if you want to use a non
443-interpreted dollar within a double quoted string, you should escape it with a
444backslash ("\$"), it does not work outside weak quoting.
445
446Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200447quoting.
448
449Strong quoting is achieved by using single quotes (''). Inside single quotes,
450nothing is interpreted, it's the efficient way to quote regexes.
451
452Quoted and escaped strings are replaced in memory by their interpreted
453equivalent, it allows you to perform concatenation.
454
455 Example:
456 # those are equivalents:
457 log-format %{+Q}o\ %t\ %s\ %{-Q}r
458 log-format "%{+Q}o %t %s %{-Q}r"
459 log-format '%{+Q}o %t %s %{-Q}r'
460 log-format "%{+Q}o %t"' %s %{-Q}r'
461 log-format "%{+Q}o %t"' %s'\ %{-Q}r
462
463 # those are equivalents:
464 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
465 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
466 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
467 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
468
469
William Lallemandb2f07452015-05-12 14:27:13 +02004702.3. Environment variables
471--------------------------
472
473HAProxy's configuration supports environment variables. Those variables are
474interpreted only within double quotes. Variables are expanded during the
475configuration parsing. Variable names must be preceded by a dollar ("$") and
476optionally enclosed with braces ("{}") similarly to what is done in Bourne
477shell. Variable names can contain alphanumerical characters or the character
478underscore ("_") but should not start with a digit.
479
480 Example:
481
482 bind "fd@${FD_APP1}"
483
484 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
485
486 user "$HAPROXY_USER"
487
William Lallemand4d03e432019-06-14 15:35:37 +0200488Some variables are defined by HAProxy, they can be used in the configuration
489file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200490
William Lallemand4d03e432019-06-14 15:35:37 +0200491* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
492 name of the local peer. (See "-L" in the management guide.)
493
494* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
495 separated by semicolons. Can be useful in the case you specified a
496 directory.
497
498* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
499
John Roeslerfb2fce12019-07-10 15:45:51 -0500500* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200501 processes, separated by semicolons.
502
John Roeslerfb2fce12019-07-10 15:45:51 -0500503* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200504 CLI, separated by semicolons.
505
506See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200507
5082.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200509----------------
510
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100511Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100512values are generally expressed in milliseconds (unless explicitly stated
513otherwise) but may be expressed in any other unit by suffixing the unit to the
514numeric value. It is important to consider this because it will not be repeated
515for every keyword. Supported units are :
516
517 - us : microseconds. 1 microsecond = 1/1000000 second
518 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
519 - s : seconds. 1s = 1000ms
520 - m : minutes. 1m = 60s = 60000ms
521 - h : hours. 1h = 60m = 3600s = 3600000ms
522 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
523
524
Lukas Tribusaa83a312017-03-21 09:25:09 +00005252.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200526-------------
527
528 # Simple configuration for an HTTP proxy listening on port 80 on all
529 # interfaces and forwarding requests to a single backend "servers" with a
530 # single server "server1" listening on 127.0.0.1:8000
531 global
532 daemon
533 maxconn 256
534
535 defaults
536 mode http
537 timeout connect 5000ms
538 timeout client 50000ms
539 timeout server 50000ms
540
541 frontend http-in
542 bind *:80
543 default_backend servers
544
545 backend servers
546 server server1 127.0.0.1:8000 maxconn 32
547
548
549 # The same configuration defined with a single listen block. Shorter but
550 # less expressive, especially in HTTP mode.
551 global
552 daemon
553 maxconn 256
554
555 defaults
556 mode http
557 timeout connect 5000ms
558 timeout client 50000ms
559 timeout server 50000ms
560
561 listen http-in
562 bind *:80
563 server server1 127.0.0.1:8000 maxconn 32
564
565
566Assuming haproxy is in $PATH, test these configurations in a shell with:
567
Willy Tarreauccb289d2010-12-11 20:19:38 +0100568 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200569
570
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005713. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200572--------------------
573
574Parameters in the "global" section are process-wide and often OS-specific. They
575are generally set once for all and do not need being changed once correct. Some
576of them have command-line equivalents.
577
578The following keywords are supported in the "global" section :
579
580 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200581 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200582 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200583 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200584 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200585 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200586 - description
587 - deviceatlas-json-file
588 - deviceatlas-log-level
589 - deviceatlas-separator
590 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900591 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200592 - gid
593 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100594 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200595 - h1-case-adjust
596 - h1-case-adjust-file
Willy Tarreau6a06a402007-07-15 20:15:28 +0200597 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200598 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100599 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200600 - lua-load
William Lallemand27edc4b2019-05-07 17:49:33 +0200601 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200602 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200603 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200604 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200605 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100606 - presetenv
607 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200608 - uid
609 - ulimit-n
610 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200611 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100612 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200613 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200614 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200615 - ssl-default-bind-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200616 - ssl-default-bind-options
617 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200618 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200619 - ssl-default-server-options
620 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100621 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100622 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100623 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100624 - 51degrees-data-file
625 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200626 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200627 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200628 - wurfl-data-file
629 - wurfl-information-list
630 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200631 - wurfl-cache-size
Willy Tarreaud72758d2010-01-12 10:42:19 +0100632
Willy Tarreau6a06a402007-07-15 20:15:28 +0200633 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200634 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200635 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200636 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100637 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100638 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100639 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200640 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200641 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200642 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200643 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200644 - noepoll
645 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000646 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200647 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100648 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300649 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000650 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100651 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200652 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200653 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200654 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000655 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000656 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200657 - tune.buffers.limit
658 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200659 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200660 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100661 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200662 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200663 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200664 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100665 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200666 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200667 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100668 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100669 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100670 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100671 - tune.lua.session-timeout
672 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200673 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100674 - tune.maxaccept
675 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200676 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200677 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200678 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100679 - tune.rcvbuf.client
680 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100681 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200682 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100683 - tune.sndbuf.client
684 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100685 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100686 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200687 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100688 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200689 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200690 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100691 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200692 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100693 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200694 - tune.vars.reqres-max-size
695 - tune.vars.sess-max-size
696 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100697 - tune.zlib.memlevel
698 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100699
Willy Tarreau6a06a402007-07-15 20:15:28 +0200700 * Debugging
701 - debug
702 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200703
704
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007053.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200706------------------------------------
707
Emeric Brunc8e8d122012-10-02 18:42:10 +0200708ca-base <dir>
709 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200710 relative path is used with "ca-file" or "crl-file" directives. Absolute
711 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200712
Willy Tarreau6a06a402007-07-15 20:15:28 +0200713chroot <jail dir>
714 Changes current directory to <jail dir> and performs a chroot() there before
715 dropping privileges. This increases the security level in case an unknown
716 vulnerability would be exploited, since it would make it very hard for the
717 attacker to exploit the system. This only works when the process is started
718 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100719 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100720
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100721cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
722 On Linux 2.6 and above, it is possible to bind a process or a thread to a
723 specific CPU set. This means that the process or the thread will never run on
724 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
725 sets. The first argument is a process set, eventually followed by a thread
726 set. These sets have the format
727
728 all | odd | even | number[-[number]]
729
730 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100731 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100732 ignored. It is possible to specify a range with two such number delimited by
733 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100734 "all", only odd numbers using "odd" or even numbers using "even", just like
735 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100736 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 +0100737 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100738 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100739 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100740 specified. Each "cpu-map" directive will replace the previous ones when they
741 overlap. A thread will be bound on the intersection of its mapping and the
742 one of the process on which it is attached. If the intersection is null, no
743 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100744
Christopher Fauletff4121f2017-11-22 16:38:49 +0100745 Ranges can be partially defined. The higher bound can be omitted. In such
746 case, it is replaced by the corresponding maximum value, 32 or 64 depending
747 on the machine's word size.
748
Christopher Faulet26028f62017-11-22 15:01:51 +0100749 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100750 automatically bind a process or a thread to a CPU by incrementing
751 process/thread and CPU sets. To be valid, both sets must have the same
752 size. No matter the declaration order of the CPU sets, it will be bound from
753 the lowest to the highest bound. Having a process and a thread range with the
754 "auto:" prefix is not supported. Only one range is supported, the other one
755 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100756
757 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100758 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
759
760 cpu-map 1/all 0-3 # bind all threads of the first process on the
761 # first 4 CPUs
762
763 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
764 # or "cpu-map 1-32 0-31" depending on the machine's
765 # word size.
766
Christopher Faulet26028f62017-11-22 15:01:51 +0100767 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100768 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100769 cpu-map auto:1-4 0-3
770 cpu-map auto:1-4 0-1 2-3
771 cpu-map auto:1-4 3 2 1 0
772
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100773 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
774 # and so on.
775 cpu-map auto:1/1-4 0-3
776 cpu-map auto:1/1-4 0-1 2-3
777 cpu-map auto:1/1-4 3 2 1 0
778
Davor Ocelice9ed2812017-12-25 17:49:28 +0100779 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100780 cpu-map auto:all 0-63
781 cpu-map auto:even 0-31
782 cpu-map auto:odd 32-63
783
784 # invalid cpu-map because process and CPU sets have different sizes.
785 cpu-map auto:1-4 0 # invalid
786 cpu-map auto:1 0-3 # invalid
787
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100788 # invalid cpu-map because automatic binding is used with a process range
789 # and a thread range.
790 cpu-map auto:all/all 0 # invalid
791 cpu-map auto:all/1-4 0 # invalid
792 cpu-map auto:1-4/all 0 # invalid
793
Emeric Brunc8e8d122012-10-02 18:42:10 +0200794crt-base <dir>
795 Assigns a default directory to fetch SSL certificates from when a relative
796 path is used with "crtfile" directives. Absolute locations specified after
797 "crtfile" prevail and ignore "crt-base".
798
Willy Tarreau6a06a402007-07-15 20:15:28 +0200799daemon
800 Makes the process fork into background. This is the recommended mode of
801 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100802 disabled by the command line "-db" argument. This option is ignored in
803 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200804
David Carlier8167f302015-06-01 13:50:06 +0200805deviceatlas-json-file <path>
806 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100807 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200808
809deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100810 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200811 optional and set to 0 by default if not set.
812
813deviceatlas-separator <char>
814 Sets the character separator for the API properties results. This directive
815 is optional and set to | by default if not set.
816
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100817deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200818 Sets the client cookie's name used for the detection if the DeviceAtlas
819 Client-side component was used during the request. This directive is optional
820 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100821
Simon Horman98637e52014-06-20 12:30:16 +0900822external-check
823 Allows the use of an external agent to perform health checks.
824 This is disabled by default as a security precaution.
825 See "option external-check".
826
Willy Tarreau6a06a402007-07-15 20:15:28 +0200827gid <number>
828 Changes the process' group ID to <number>. It is recommended that the group
829 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
830 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100831 Note that if haproxy is started from a user having supplementary groups, it
832 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200833 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100834
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100835hard-stop-after <time>
836 Defines the maximum time allowed to perform a clean soft-stop.
837
838 Arguments :
839 <time> is the maximum time (by default in milliseconds) for which the
840 instance will remain alive when a soft-stop is received via the
841 SIGUSR1 signal.
842
843 This may be used to ensure that the instance will quit even if connections
844 remain opened during a soft-stop (for example with long timeouts for a proxy
845 in tcp mode). It applies both in TCP and HTTP mode.
846
847 Example:
848 global
849 hard-stop-after 30s
850
Christopher Faulet98fbe952019-07-22 16:18:24 +0200851h1-case-adjust <from> <to>
852 Defines the case adjustment to apply, when enabled, to the header name
853 <from>, to change it to <to> before sending it to HTTP/1 clients or
854 servers. <from> must be in lower case, and <from> and <to> must not differ
855 except for their case. It may be repeated if several header names need to be
856 ajusted. Duplicate entries are not allowed. If a lot of header names have to
857 be adjusted, it might be more convenient to use "h1-case-adjust-file".
858 Please note that no transformation will be applied unless "option
859 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
860 specified in a proxy.
861
862 There is no standard case for header names because, as stated in RFC7230,
863 they are case-insensitive. So applications must handle them in a case-
864 insensitive manner. But some bogus applications violate the standards and
865 erroneously rely on the cases most commonly used by browsers. This problem
866 becomes critical with HTTP/2 because all header names must be exchanged in
867 lower case, and HAProxy follows the same convention. All header names are
868 sent in lower case to clients and servers, regardless of the HTTP version.
869
870 Applications which fail to properly process requests or responses may require
871 to temporarily use such workarounds to adjust header names sent to them for
872 the time it takes the application to be fixed. Please note that an
873 application which requires such workarounds might be vulnerable to content
874 smuggling attacks and must absolutely be fixed.
875
876 Example:
877 global
878 h1-case-adjust content-length Content-Length
879
880 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
881 "option h1-case-adjust-bogus-server".
882
883h1-case-adjust-file <hdrs-file>
884 Defines a file containing a list of key/value pairs used to adjust the case
885 of some header names before sending them to HTTP/1 clients or servers. The
886 file <hdrs-file> must contain 2 header names per line. The first one must be
887 in lower case and both must not differ except for their case. Lines which
888 start with '#' are ignored, just like empty lines. Leading and trailing tabs
889 and spaces are stripped. Duplicate entries are not allowed. Please note that
890 no transformation will be applied unless "option h1-case-adjust-bogus-client"
891 or "option h1-case-adjust-bogus-server" is specified in a proxy.
892
893 If this directive is repeated, only the last one will be processed. It is an
894 alternative to the directive "h1-case-adjust" if a lot of header names need
895 to be adjusted. Please read the risks associated with using this.
896
897 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
898 "option h1-case-adjust-bogus-server".
899
Willy Tarreau6a06a402007-07-15 20:15:28 +0200900group <group name>
901 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
902 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100903
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200904log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
905 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100906 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100907 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100908 configured with "log global".
909
910 <address> can be one of:
911
Willy Tarreau2769aa02007-12-27 18:26:09 +0100912 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100913 no port is specified, 514 is used by default (the standard syslog
914 port).
915
David du Colombier24bb5f52011-03-17 10:40:23 +0100916 - An IPv6 address followed by a colon and optionally a UDP port. If
917 no port is specified, 514 is used by default (the standard syslog
918 port).
919
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100920 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100921 considerations for chroot (be sure the path is accessible inside
922 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100923 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100924
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100925 - A file descriptor number in the form "fd@<number>", which may point
926 to a pipe, terminal, or socket. In this case unbuffered logs are used
927 and one writev() call per log is performed. This is a bit expensive
928 but acceptable for most workloads. Messages sent this way will not be
929 truncated but may be dropped, in which case the DroppedLogs counter
930 will be incremented. The writev() call is atomic even on pipes for
931 messages up to PIPE_BUF size, which POSIX recommends to be at least
932 512 and which is 4096 bytes on most modern operating systems. Any
933 larger message may be interleaved with messages from other processes.
934 Exceptionally for debugging purposes the file descriptor may also be
935 directed to a file, but doing so will significantly slow haproxy down
936 as non-blocking calls will be ignored. Also there will be no way to
937 purge nor rotate this file without restarting the process. Note that
938 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100939 for use with a TCP syslog server. See also the "short" and "raw"
940 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100941
942 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
943 "fd@2", see above.
944
Willy Tarreauc046d162019-08-30 15:24:59 +0200945 - A ring buffer in the form "ring@<name>", which will correspond to an
946 in-memory ring buffer accessible over the CLI using the "show events"
947 command, which will also list existing rings and their sizes. Such
948 buffers are lost on reload or restart but when used as a complement
949 this can help troubleshooting by having the logs instantly available.
950
William Lallemandb2f07452015-05-12 14:27:13 +0200951 You may want to reference some environment variables in the address
952 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100953
Willy Tarreau18324f52014-06-27 18:10:07 +0200954 <length> is an optional maximum line length. Log lines larger than this value
955 will be truncated before being sent. The reason is that syslog
956 servers act differently on log line length. All servers support the
957 default value of 1024, but some servers simply drop larger lines
958 while others do log them. If a server supports long lines, it may
959 make sense to set this value here in order to avoid truncating long
960 lines. Similarly, if a server drops long lines, it is preferable to
961 truncate them before sending them. Accepted values are 80 to 65535
962 inclusive. The default value of 1024 is generally fine for all
963 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100964 JSON-formatted logs may require larger values. You may also need to
965 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200966
Dragan Dosen7ad31542015-09-28 17:16:47 +0200967 <format> is the log format used when generating syslog messages. It may be
968 one of the following :
969
970 rfc3164 The RFC3164 syslog message format. This is the default.
971 (https://tools.ietf.org/html/rfc3164)
972
973 rfc5424 The RFC5424 syslog message format.
974 (https://tools.ietf.org/html/rfc5424)
975
Willy Tarreaue8746a02018-11-12 08:45:00 +0100976 short A message containing only a level between angle brackets such as
977 '<3>', followed by the text. The PID, date, time, process name
978 and system name are omitted. This is designed to be used with a
979 local log server. This format is compatible with what the systemd
980 logger consumes.
981
Willy Tarreauc1b06452018-11-12 11:57:56 +0100982 raw A message containing only the text. The level, PID, date, time,
983 process name and system name are omitted. This is designed to be
984 used in containers or during development, where the severity only
985 depends on the file descriptor used (stdout/stderr).
986
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200987 <ranges> A list of comma-separated ranges to identify the logs to sample.
988 This is used to balance the load of the logs to send to the log
989 server. The limits of the ranges cannot be null. They are numbered
990 from 1. The size or period (in number of logs) of the sample must be
991 set with <sample_size> parameter.
992
993 <sample_size>
994 The size of the sample in number of logs to consider when balancing
995 their logging loads. It is used to balance the load of the logs to
996 send to the syslog server. This size must be greater or equal to the
997 maximum of the high limits of the ranges.
998 (see also <ranges> parameter).
999
Robert Tsai81ae1952007-12-05 10:47:29 +01001000 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001001
Willy Tarreaue8746a02018-11-12 08:45:00 +01001002 kern user mail daemon auth syslog lpr news
1003 uucp cron auth2 ftp ntp audit alert cron2
1004 local0 local1 local2 local3 local4 local5 local6 local7
1005
Willy Tarreauc1b06452018-11-12 11:57:56 +01001006 Note that the facility is ignored for the "short" and "raw"
1007 formats, but still required as a positional field. It is
1008 recommended to use "daemon" in this case to make it clear that
1009 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001010
1011 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001012 all messages are sent. If a maximum level is specified, only messages with a
1013 severity at least as important as this level will be sent. An optional minimum
1014 level can be specified. If it is set, logs emitted with a more severe level
1015 than this one will be capped to this level. This is used to avoid sending
1016 "emerg" messages on all terminals on some default syslog configurations.
1017 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001018
Cyril Bontédc4d9032012-04-08 21:57:39 +02001019 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001020
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001021log-send-hostname [<string>]
1022 Sets the hostname field in the syslog header. If optional "string" parameter
1023 is set the header is set to the string contents, otherwise uses the hostname
1024 of the system. Generally used if one is not relaying logs through an
1025 intermediate syslog server or for simply customizing the hostname printed in
1026 the logs.
1027
Kevinm48936af2010-12-22 16:08:21 +00001028log-tag <string>
1029 Sets the tag field in the syslog header to this string. It defaults to the
1030 program name as launched from the command line, which usually is "haproxy".
1031 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001032 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001033
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001034lua-load <file>
1035 This global directive loads and executes a Lua file. This directive can be
1036 used multiple times.
1037
William Lallemand4cfede82017-11-24 22:02:34 +01001038master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001039 Master-worker mode. It is equivalent to the command line "-W" argument.
1040 This mode will launch a "master" which will monitor the "workers". Using
1041 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001042 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001043 or daemon mode. It is recommended to use this mode with multiprocess and
1044 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001045 By default, if a worker exits with a bad return code, in the case of a
1046 segfault for example, all workers will be killed, and the master will leave.
1047 It is convenient to combine this behavior with Restart=on-failure in a
1048 systemd unit file in order to relaunch the whole process. If you don't want
1049 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001050
William Lallemand4cfede82017-11-24 22:02:34 +01001051 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001052
William Lallemand27edc4b2019-05-07 17:49:33 +02001053mworker-max-reloads <number>
1054 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001055 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001056 number of reloads is greater than this number, the worker will receive a
1057 SIGTERM. This option helps to keep under control the number of workers.
1058 See also "show proc" in the Management Guide.
1059
Willy Tarreau6a06a402007-07-15 20:15:28 +02001060nbproc <number>
1061 Creates <number> processes when going daemon. This requires the "daemon"
1062 mode. By default, only one process is created, which is the recommended mode
1063 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001064 process, it may be needed to fork multiple daemons. When set to a value
1065 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +01001066 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
1067 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001068
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001069nbthread <number>
1070 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001071 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1072 "nbproc" historically used to be the only way to use multiple processors, it
1073 also involved a number of shortcomings related to the lack of synchronization
1074 between processes (health-checks, peers, stick-tables, stats, ...) which do
1075 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001076 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1077 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1078 when nbproc is not used, the default "nbthread" value is automatically set to
1079 the number of CPUs the process is bound to upon startup. This means that the
1080 thread count can easily be adjusted from the calling process using commands
1081 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1082 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001083
Willy Tarreau6a06a402007-07-15 20:15:28 +02001084pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001085 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +02001086 the "-p" command line argument. The file must be accessible to the user
1087 starting the process. See also "daemon".
1088
Willy Tarreau1d549722016-02-16 12:41:57 +01001089presetenv <name> <value>
1090 Sets environment variable <name> to value <value>. If the variable exists, it
1091 is NOT overwritten. The changes immediately take effect so that the next line
1092 in the configuration file sees the new value. See also "setenv", "resetenv",
1093 and "unsetenv".
1094
1095resetenv [<name> ...]
1096 Removes all environment variables except the ones specified in argument. It
1097 allows to use a clean controlled environment before setting new values with
1098 setenv or unsetenv. Please note that some internal functions may make use of
1099 some environment variables, such as time manipulation functions, but also
1100 OpenSSL or even external checks. This must be used with extreme care and only
1101 after complete validation. The changes immediately take effect so that the
1102 next line in the configuration file sees the new environment. See also
1103 "setenv", "presetenv", and "unsetenv".
1104
Christopher Fauletff4121f2017-11-22 16:38:49 +01001105stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001106 Limits the stats socket to a certain set of processes numbers. By default the
1107 stats socket is bound to all processes, causing a warning to be emitted when
1108 nbproc is greater than 1 because there is no way to select the target process
1109 when connecting. However, by using this setting, it becomes possible to pin
1110 the stats socket to a specific set of processes, typically the first one. The
1111 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001112 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001113 word size (32 or 64). Ranges can be partially defined. The higher bound can
1114 be omitted. In such case, it is replaced by the corresponding maximum
1115 value. A better option consists in using the "process" setting of the "stats
1116 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001117
Baptiste Assmann5626f482015-08-23 10:00:10 +02001118server-state-base <directory>
1119 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001120 file names which do not start with a '/'. See also "server-state-file",
1121 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001122
1123server-state-file <file>
1124 Specifies the path to the file containing state of servers. If the path starts
1125 with a slash ('/'), it is considered absolute, otherwise it is considered
1126 relative to the directory specified using "server-state-base" (if set) or to
1127 the current directory. Before reloading HAProxy, it is possible to save the
1128 servers' current state using the stats command "show servers state". The
1129 output of this command must be written in the file pointed by <file>. When
1130 starting up, before handling traffic, HAProxy will read, load and apply state
1131 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001132 configuration. See also "server-state-base" and "show servers state",
1133 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001134
Willy Tarreau1d549722016-02-16 12:41:57 +01001135setenv <name> <value>
1136 Sets environment variable <name> to value <value>. If the variable exists, it
1137 is overwritten. The changes immediately take effect so that the next line in
1138 the configuration file sees the new value. See also "presetenv", "resetenv",
1139 and "unsetenv".
1140
Willy Tarreau636848a2019-04-15 19:38:50 +02001141set-dumpable
1142 This option is better left disabled by default and enabled only upon a
1143 developer's request. It has no impact on performance nor stability but will
1144 try hard to re-enable core dumps that were possibly disabled by file size
1145 limitations (ulimit -f), core size limitations (ulimit -c), or "dumpability"
1146 of a process after changing its UID/GID (such as /proc/sys/fs/suid_dumpable
1147 on Linux). Core dumps might still be limited by the current directory's
1148 permissions (check what directory the file is started from), the chroot
1149 directory's permission (it may be needed to temporarily disable the chroot
1150 directive or to move it to a dedicated writable location), or any other
1151 system-specific constraint. For example, some Linux flavours are notorious
1152 for replacing the default core file with a path to an executable not even
1153 installed on the system (check /proc/sys/kernel/core_pattern). Often, simply
1154 writing "core", "core.%p" or "/var/log/core/core.%p" addresses the issue.
1155 When trying to enable this option waiting for a rare issue to re-appear, it's
1156 often a good idea to first try to obtain such a dump by issuing, for example,
1157 "kill -11" to the haproxy process and verify that it leaves a core where
1158 expected when dying.
1159
Willy Tarreau610f04b2014-02-13 11:36:41 +01001160ssl-default-bind-ciphers <ciphers>
1161 This setting is only available when support for OpenSSL was built in. It sets
1162 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001163 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001164 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001165 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1166 information and recommendations see e.g.
1167 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1168 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1169 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1170 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001171
1172ssl-default-bind-ciphersuites <ciphersuites>
1173 This setting is only available when support for OpenSSL was built in and
1174 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1175 describing the list of cipher algorithms ("cipher suite") that are negotiated
1176 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1177 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001178 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1179 cipher configuration for TLSv1.2 and earlier, please check the
1180 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001181 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001182
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001183ssl-default-bind-options [<option>]...
1184 This setting is only available when support for OpenSSL was built in. It sets
1185 default ssl-options to force on all "bind" lines. Please check the "bind"
1186 keyword to see available options.
1187
1188 Example:
1189 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001190 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001191
Willy Tarreau610f04b2014-02-13 11:36:41 +01001192ssl-default-server-ciphers <ciphers>
1193 This setting is only available when support for OpenSSL was built in. It
1194 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001195 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001196 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001197 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1198 information and recommendations see e.g.
1199 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1200 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1201 For TLSv1.3 cipher configuration, please check the
1202 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1203 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001204
1205ssl-default-server-ciphersuites <ciphersuites>
1206 This setting is only available when support for OpenSSL was built in and
1207 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1208 string describing the list of cipher algorithms that are negotiated during
1209 the TLSv1.3 handshake with the server, for all "server" lines which do not
1210 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001211 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1212 cipher configuration for TLSv1.2 and earlier, please check the
1213 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1214 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001215
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001216ssl-default-server-options [<option>]...
1217 This setting is only available when support for OpenSSL was built in. It sets
1218 default ssl-options to force on all "server" lines. Please check the "server"
1219 keyword to see available options.
1220
Remi Gacogne47783ef2015-05-29 15:53:22 +02001221ssl-dh-param-file <file>
1222 This setting is only available when support for OpenSSL was built in. It sets
1223 the default DH parameters that are used during the SSL/TLS handshake when
1224 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001225 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001226 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001227 are not specified either by using ssl-dh-param-file or by setting them
1228 directly in the certificate file, pre-generated DH parameters of the size
1229 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1230 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001231 Custom DH parameters may be generated by using the OpenSSL command
1232 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1233 parameters should not be considered secure anymore.
1234
Emeric Brun850efd52014-01-29 12:24:34 +01001235ssl-server-verify [none|required]
1236 The default behavior for SSL verify on servers side. If specified to 'none',
1237 servers certificates are not verified. The default is 'required' except if
1238 forced using cmdline option '-dV'.
1239
Willy Tarreauabb175f2012-09-24 12:43:26 +02001240stats socket [<address:port>|<path>] [param*]
1241 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1242 Connections to this socket will return various statistics outputs and even
1243 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001244 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001245 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001246
Willy Tarreauabb175f2012-09-24 12:43:26 +02001247 All parameters supported by "bind" lines are supported, for instance to
1248 restrict access to some users or their access rights. Please consult
1249 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001250
1251stats timeout <timeout, in milliseconds>
1252 The default timeout on the stats socket is set to 10 seconds. It is possible
1253 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001254 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001255
1256stats maxconn <connections>
1257 By default, the stats socket is limited to 10 concurrent connections. It is
1258 possible to change this value with "stats maxconn".
1259
Willy Tarreau6a06a402007-07-15 20:15:28 +02001260uid <number>
1261 Changes the process' user ID to <number>. It is recommended that the user ID
1262 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1263 be started with superuser privileges in order to be able to switch to another
1264 one. See also "gid" and "user".
1265
1266ulimit-n <number>
1267 Sets the maximum number of per-process file-descriptors to <number>. By
1268 default, it is automatically computed, so it is recommended not to use this
1269 option.
1270
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001271unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1272 [ group <group> ] [ gid <gid> ]
1273
1274 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1275 This is mainly used to simplify declaration of those UNIX sockets and reduce
1276 the risk of errors, since those settings are most commonly required but are
1277 also process-specific. The <prefix> setting can be used to force all socket
1278 path to be relative to that directory. This might be needed to access another
1279 component's chroot. Note that those paths are resolved before haproxy chroots
1280 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1281 all have the same meaning as their homonyms used by the "bind" statement. If
1282 both are specified, the "bind" statement has priority, meaning that the
1283 "unix-bind" settings may be seen as process-wide default settings.
1284
Willy Tarreau1d549722016-02-16 12:41:57 +01001285unsetenv [<name> ...]
1286 Removes environment variables specified in arguments. This can be useful to
1287 hide some sensitive information that are occasionally inherited from the
1288 user's environment during some operations. Variables which did not exist are
1289 silently ignored so that after the operation, it is certain that none of
1290 these variables remain. The changes immediately take effect so that the next
1291 line in the configuration file will not see these variables. See also
1292 "setenv", "presetenv", and "resetenv".
1293
Willy Tarreau6a06a402007-07-15 20:15:28 +02001294user <user name>
1295 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1296 See also "uid" and "group".
1297
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001298node <name>
1299 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1300
1301 This statement is useful in HA configurations where two or more processes or
1302 servers share the same IP address. By setting a different node-name on all
1303 nodes, it becomes easy to immediately spot what server is handling the
1304 traffic.
1305
1306description <text>
1307 Add a text that describes the instance.
1308
1309 Please note that it is required to escape certain characters (# for example)
1310 and this text is inserted into a html page so you should avoid using
1311 "<" and ">" characters.
1312
Thomas Holmesdb04f192015-05-18 13:21:39 +0100131351degrees-data-file <file path>
1314 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001315 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001316
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001317 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001318 compiled with USE_51DEGREES.
1319
Ben Shillitof25e8e52016-12-02 14:25:37 +0000132051degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001321 A list of 51Degrees property names to be load from the dataset. A full list
1322 of names is available on the 51Degrees website:
1323 https://51degrees.com/resources/property-dictionary
1324
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001325 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001326 compiled with USE_51DEGREES.
1327
Dragan Dosen93b38d92015-06-29 16:43:25 +0200132851degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001329 A char that will be appended to every property value in a response header
1330 containing 51Degrees results. If not set that will be set as ','.
1331
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001332 Please note that this option is only available when haproxy has been
1333 compiled with USE_51DEGREES.
1334
133551degrees-cache-size <number>
1336 Sets the size of the 51Degrees converter cache to <number> entries. This
1337 is an LRU cache which reminds previous device detections and their results.
1338 By default, this cache is disabled.
1339
1340 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001341 compiled with USE_51DEGREES.
1342
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001343wurfl-data-file <file path>
1344 The path of the WURFL data file to provide device detection services. The
1345 file should be accessible by HAProxy with relevant permissions.
1346
1347 Please note that this option is only available when haproxy has been compiled
1348 with USE_WURFL=1.
1349
1350wurfl-information-list [<capability>]*
1351 A space-delimited list of WURFL capabilities, virtual capabilities, property
1352 names we plan to use in injected headers. A full list of capability and
1353 virtual capability names is available on the Scientiamobile website :
1354
1355 https://www.scientiamobile.com/wurflCapability
1356
1357 Valid WURFL properties are:
1358 - wurfl_id Contains the device ID of the matched device.
1359
1360 - wurfl_root_id Contains the device root ID of the matched
1361 device.
1362
1363 - wurfl_isdevroot Tells if the matched device is a root device.
1364 Possible values are "TRUE" or "FALSE".
1365
1366 - wurfl_useragent The original useragent coming with this
1367 particular web request.
1368
1369 - wurfl_api_version Contains a string representing the currently
1370 used Libwurfl API version.
1371
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001372 - wurfl_info A string containing information on the parsed
1373 wurfl.xml and its full path.
1374
1375 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1376 WURFL has been loaded successfully.
1377
1378 - wurfl_normalized_useragent The normalized useragent.
1379
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001380 Please note that this option is only available when haproxy has been compiled
1381 with USE_WURFL=1.
1382
1383wurfl-information-list-separator <char>
1384 A char that will be used to separate values in a response header containing
1385 WURFL results. If not set that a comma (',') will be used by default.
1386
1387 Please note that this option is only available when haproxy has been compiled
1388 with USE_WURFL=1.
1389
1390wurfl-patch-file [<file path>]
1391 A list of WURFL patch file paths. Note that patches are loaded during startup
1392 thus before the chroot.
1393
1394 Please note that this option is only available when haproxy has been compiled
1395 with USE_WURFL=1.
1396
paulborilebad132c2019-04-18 11:57:04 +02001397wurfl-cache-size <size>
1398 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1399 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001400 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001401 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001402
1403 Please note that this option is only available when haproxy has been compiled
1404 with USE_WURFL=1.
1405
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014063.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001407-----------------------
1408
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001409busy-polling
1410 In some situations, especially when dealing with low latency on processors
1411 supporting a variable frequency or when running inside virtual machines, each
1412 time the process waits for an I/O using the poller, the processor goes back
1413 to sleep or is offered to another VM for a long time, and it causes
1414 excessively high latencies. This option provides a solution preventing the
1415 processor from sleeping by always using a null timeout on the pollers. This
1416 results in a significant latency reduction (30 to 100 microseconds observed)
1417 at the expense of a risk to overheat the processor. It may even be used with
1418 threads, in which case improperly bound threads may heavily conflict,
1419 resulting in a worse performance and high values for the CPU stolen fields
1420 in "show info" output, indicating which threads are misconfigured. It is
1421 important not to let the process run on the same processor as the network
1422 interrupts when this option is used. It is also better to avoid using it on
1423 multiple CPU threads sharing the same core. This option is disabled by
1424 default. If it has been enabled, it may still be forcibly disabled by
1425 prefixing it with the "no" keyword. It is ignored by the "select" and
1426 "poll" pollers.
1427
Willy Tarreau1746eec2014-04-25 10:46:47 +02001428max-spread-checks <delay in milliseconds>
1429 By default, haproxy tries to spread the start of health checks across the
1430 smallest health check interval of all the servers in a farm. The principle is
1431 to avoid hammering services running on the same server. But when using large
1432 check intervals (10 seconds or more), the last servers in the farm take some
1433 time before starting to be tested, which can be a problem. This parameter is
1434 used to enforce an upper bound on delay between the first and the last check,
1435 even if the servers' check intervals are larger. When servers run with
1436 shorter intervals, their intervals will be respected though.
1437
Willy Tarreau6a06a402007-07-15 20:15:28 +02001438maxconn <number>
1439 Sets the maximum per-process number of concurrent connections to <number>. It
1440 is equivalent to the command-line argument "-n". Proxies will stop accepting
1441 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001442 automatically adjusted according to this value. See also "ulimit-n". Note:
1443 the "select" poller cannot reliably use more than 1024 file descriptors on
1444 some platforms. If your platform only supports select and reports "select
1445 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001446 below 500 in general). If this value is not set, it will automatically be
1447 calculated based on the current file descriptors limit reported by the
1448 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1449 is enforced, based on the buffer size, memory allocated to compression, SSL
1450 cache size, and use or not of SSL and the associated maxsslconn (which can
1451 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001452
Willy Tarreau81c25d02011-09-07 15:17:21 +02001453maxconnrate <number>
1454 Sets the maximum per-process number of connections per second to <number>.
1455 Proxies will stop accepting connections when this limit is reached. It can be
1456 used to limit the global capacity regardless of each frontend capacity. It is
1457 important to note that this can only be used as a service protection measure,
1458 as there will not necessarily be a fair share between frontends when the
1459 limit is reached, so it's a good idea to also limit each frontend to some
1460 value close to its expected share. Also, lowering tune.maxaccept can improve
1461 fairness.
1462
William Lallemandd85f9172012-11-09 17:05:39 +01001463maxcomprate <number>
1464 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001465 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001466 level will be decreased during the session. If the maximum is reached at the
1467 beginning of a session, the session will not compress at all. If the maximum
1468 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001469 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001470 default value.
1471
William Lallemand072a2bf2012-11-20 17:01:01 +01001472maxcompcpuusage <number>
1473 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1474 for new requests or decreasing the compression level of current requests.
1475 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1476 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1477 case of multiple processes (nbproc > 1), each process manages its individual
1478 usage. A value of 100 disable the limit. The default value is 100. Setting
1479 a lower value will prevent the compression work from slowing the whole
1480 process down and from introducing high latencies.
1481
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001482maxpipes <number>
1483 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1484 are only used by kernel-based tcp splicing. Since a pipe contains two file
1485 descriptors, the "ulimit-n" value will be increased accordingly. The default
1486 value is maxconn/4, which seems to be more than enough for most heavy usages.
1487 The splice code dynamically allocates and releases pipes, and can fall back
1488 to standard copy, so setting this value too low may only impact performance.
1489
Willy Tarreau93e7c002013-10-07 18:51:07 +02001490maxsessrate <number>
1491 Sets the maximum per-process number of sessions per second to <number>.
1492 Proxies will stop accepting connections when this limit is reached. It can be
1493 used to limit the global capacity regardless of each frontend capacity. It is
1494 important to note that this can only be used as a service protection measure,
1495 as there will not necessarily be a fair share between frontends when the
1496 limit is reached, so it's a good idea to also limit each frontend to some
1497 value close to its expected share. Also, lowering tune.maxaccept can improve
1498 fairness.
1499
Willy Tarreau403edff2012-09-06 11:58:37 +02001500maxsslconn <number>
1501 Sets the maximum per-process number of concurrent SSL connections to
1502 <number>. By default there is no SSL-specific limit, which means that the
1503 global maxconn setting will apply to all connections. Setting this limit
1504 avoids having openssl use too much memory and crash when malloc returns NULL
1505 (since it unfortunately does not reliably check for such conditions). Note
1506 that the limit applies both to incoming and outgoing connections, so one
1507 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001508 If this value is not set, but a memory limit is enforced, this value will be
1509 automatically computed based on the memory limit, maxconn, the buffer size,
1510 memory allocated to compression, SSL cache size, and use of SSL in either
1511 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1512 when there is a memory limit, haproxy will automatically adjust these values
1513 so that 100% of the connections can be made over SSL with no risk, and will
1514 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001515
Willy Tarreaue43d5322013-10-07 20:01:52 +02001516maxsslrate <number>
1517 Sets the maximum per-process number of SSL sessions per second to <number>.
1518 SSL listeners will stop accepting connections when this limit is reached. It
1519 can be used to limit the global SSL CPU usage regardless of each frontend
1520 capacity. It is important to note that this can only be used as a service
1521 protection measure, as there will not necessarily be a fair share between
1522 frontends when the limit is reached, so it's a good idea to also limit each
1523 frontend to some value close to its expected share. It is also important to
1524 note that the sessions are accounted before they enter the SSL stack and not
1525 after, which also protects the stack against bad handshakes. Also, lowering
1526 tune.maxaccept can improve fairness.
1527
William Lallemand9d5f5482012-11-07 16:12:57 +01001528maxzlibmem <number>
1529 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1530 When the maximum amount is reached, future sessions will not compress as long
1531 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001532 The default value is 0. The value is available in bytes on the UNIX socket
1533 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1534 "ZlibMemUsage" in bytes.
1535
Willy Tarreau6a06a402007-07-15 20:15:28 +02001536noepoll
1537 Disables the use of the "epoll" event polling system on Linux. It is
1538 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001539 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001540
1541nokqueue
1542 Disables the use of the "kqueue" event polling system on BSD. It is
1543 equivalent to the command-line argument "-dk". The next polling system
1544 used will generally be "poll". See also "nopoll".
1545
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001546noevports
1547 Disables the use of the event ports event polling system on SunOS systems
1548 derived from Solaris 10 and later. It is equivalent to the command-line
1549 argument "-dv". The next polling system used will generally be "poll". See
1550 also "nopoll".
1551
Willy Tarreau6a06a402007-07-15 20:15:28 +02001552nopoll
1553 Disables the use of the "poll" event polling system. It is equivalent to the
1554 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001555 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001556 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1557 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001558
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001559nosplice
1560 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001561 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001562 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001563 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001564 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1565 be used. This option makes it easier to globally disable kernel splicing in
1566 case of doubt. See also "option splice-auto", "option splice-request" and
1567 "option splice-response".
1568
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001569nogetaddrinfo
1570 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1571 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1572
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001573noreuseport
1574 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1575 command line argument "-dR".
1576
Willy Tarreaud2d33482019-04-25 17:09:07 +02001577profiling.tasks { auto | on | off }
1578 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1579 the profiling automatically turns on a thread when it starts to suffer from
1580 an average latency of 1000 microseconds or higher as reported in the
1581 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05001582 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001583 loops so it does not vary quickly and tends to significantly smooth short
1584 spikes). It may also spontaneously trigger from time to time on overloaded
1585 systems, containers, or virtual machines, or when the system swaps (which
1586 must absolutely never happen on a load balancer).
1587
1588 CPU profiling per task can be very convenient to report where the time is
1589 spent and which requests have what effect on which other request. Enabling
1590 it will typically affect the overall's performance by less than 1%, thus it
1591 is recommended to leave it to the default 'auto' value so that it only
1592 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001593 supporting the clock_gettime(2) syscall with clock identifiers
1594 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1595 be zero. This option may be changed at run time using "set profiling" on the
1596 CLI.
1597
Willy Tarreaufe255b72007-10-14 23:09:26 +02001598spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001599 Sometimes it is desirable to avoid sending agent and health checks to
1600 servers at exact intervals, for instance when many logical servers are
1601 located on the same physical server. With the help of this parameter, it
1602 becomes possible to add some randomness in the check interval between 0
1603 and +/- 50%. A value between 2 and 5 seems to show good results. The
1604 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001605
Davor Ocelice9ed2812017-12-25 17:49:28 +01001606ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001607 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001608 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001609 multiple times, it will simply enable multiple crypto engines. Referencing an
1610 unsupported engine will prevent haproxy from starting. Note that many engines
1611 will lead to lower HTTPS performance than pure software with recent
1612 processors. The optional command "algo" sets the default algorithms an ENGINE
1613 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001614 of "ALL" uses the engine for all cryptographic operations. If no list of
1615 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001616 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1617 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1618 openssl configuration file uses:
1619 https://www.openssl.org/docs/man1.0.2/apps/config.html
1620
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001621ssl-mode-async
1622 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001623 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001624 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1625 doesn't support moving read/write buffers and is not compliant with
1626 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05001627 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001628 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001629
Willy Tarreau33cb0652014-12-23 22:52:37 +01001630tune.buffers.limit <number>
1631 Sets a hard limit on the number of buffers which may be allocated per process.
1632 The default value is zero which means unlimited. The minimum non-zero value
1633 will always be greater than "tune.buffers.reserve" and should ideally always
1634 be about twice as large. Forcing this value can be particularly useful to
1635 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001636 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001637 another one to be released by another session. Since buffers are dynamically
1638 allocated and released, the waiting time is very short and not perceptible
1639 provided that limits remain reasonable. In fact sometimes reducing the limit
1640 may even increase performance by increasing the CPU cache's efficiency. Tests
1641 have shown good results on average HTTP traffic with a limit to 1/10 of the
1642 expected global maxconn setting, which also significantly reduces memory
1643 usage. The memory savings come from the fact that a number of connections
1644 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1645 advised to do so by an haproxy core developer.
1646
Willy Tarreau1058ae72014-12-23 22:40:40 +01001647tune.buffers.reserve <number>
1648 Sets the number of buffers which are pre-allocated and reserved for use only
1649 during memory shortage conditions resulting in failed memory allocations. The
1650 minimum value is 2 and is also the default. There is no reason a user would
1651 want to change this value, it's mostly aimed at haproxy core developers.
1652
Willy Tarreau27a674e2009-08-17 07:23:33 +02001653tune.bufsize <number>
1654 Sets the buffer size to this size (in bytes). Lower values allow more
1655 sessions to coexist in the same amount of RAM, and higher values allow some
1656 applications with very large cookies to work. The default value is 16384 and
1657 can be changed at build time. It is strongly recommended not to change this
1658 from the default value, as very low values will break some services such as
1659 statistics, and values larger than default size will increase memory usage,
1660 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001661 parameter should be decreased by the same factor as this one is increased. In
1662 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1663 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001664 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001665 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1666 value set using this parameter will automatically be rounded up to the next
1667 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001668
Willy Tarreau43961d52010-10-04 20:39:20 +02001669tune.chksize <number>
1670 Sets the check buffer size to this size (in bytes). Higher values may help
1671 find string or regex patterns in very large pages, though doing so may imply
1672 more memory and CPU usage. The default value is 16384 and can be changed at
1673 build time. It is not recommended to change this value, but to use better
1674 checks whenever possible.
1675
William Lallemandf3747832012-11-09 12:33:10 +01001676tune.comp.maxlevel <number>
1677 Sets the maximum compression level. The compression level affects CPU
1678 usage during compression. This value affects CPU usage during compression.
1679 Each session using compression initializes the compression algorithm with
1680 this value. The default value is 1.
1681
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001682tune.fail-alloc
1683 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1684 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1685 success). This is useful to debug and make sure memory failures are handled
1686 gracefully.
1687
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001688tune.h2.header-table-size <number>
1689 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1690 cannot be larger than 65536 bytes. A larger value may help certain clients
1691 send more compact requests, depending on their capabilities. This amount of
1692 memory is consumed for each HTTP/2 connection. It is recommended not to
1693 change it.
1694
Willy Tarreaue6baec02017-07-27 11:45:11 +02001695tune.h2.initial-window-size <number>
1696 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001697 can upload before waiting for an acknowledgment from haproxy. This setting
1698 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001699 The default value is 65535, which roughly allows up to 5 Mbps of upload
1700 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1701 over a 1-ms local network. It can make sense to increase this value to allow
1702 faster uploads, or to reduce it to increase fairness when dealing with many
1703 clients. It doesn't affect resource usage.
1704
Willy Tarreau5242ef82017-07-27 11:47:28 +02001705tune.h2.max-concurrent-streams <number>
1706 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1707 number of outstanding requests on a single connection). The default value is
1708 100. A larger one may slightly improve page load time for complex sites when
1709 visited over high latency networks, but increases the amount of resources a
1710 single client may allocate. A value of zero disables the limit so a single
1711 client may create as many streams as allocatable by haproxy. It is highly
1712 recommended not to change this value.
1713
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001714tune.h2.max-frame-size <number>
1715 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1716 receive to its peers. The default value is the largest between 16384 and the
1717 buffer size (tune.bufsize). In any case, haproxy will not announce support
1718 for frame sizes larger than buffers. The main purpose of this setting is to
1719 allow to limit the maximum frame size setting when using large buffers. Too
1720 large frame sizes might have performance impact or cause some peers to
1721 misbehave. It is highly recommended not to change this value.
1722
Willy Tarreau193b8c62012-11-22 00:17:38 +01001723tune.http.cookielen <number>
1724 Sets the maximum length of captured cookies. This is the maximum value that
1725 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1726 will automatically be truncated to this one. It is important not to set too
1727 high a value because all cookie captures still allocate this size whatever
1728 their configured value (they share a same pool). This value is per request
1729 per response, so the memory allocated is twice this value per connection.
1730 When not specified, the limit is set to 63 characters. It is recommended not
1731 to change this value.
1732
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001733tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001734 Sets the maximum length of request URI in logs. This prevents truncating long
1735 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001736 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001737 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001738 configuration directives too.
1739 The default value is 1024.
1740
Willy Tarreauac1932d2011-10-24 19:14:41 +02001741tune.http.maxhdr <number>
1742 Sets the maximum number of headers in a request. When a request comes with a
1743 number of headers greater than this value (including the first line), it is
1744 rejected with a "400 Bad Request" status code. Similarly, too large responses
1745 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1746 for all usages, considering that the widely deployed Apache server uses the
1747 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001748 a buggy application to work by the time it gets fixed. The accepted range is
1749 1..32767. Keep in mind that each new header consumes 32bits of memory for
1750 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001751
Willy Tarreau7e312732014-02-12 16:35:14 +01001752tune.idletimer <timeout>
1753 Sets the duration after which haproxy will consider that an empty buffer is
1754 probably associated with an idle stream. This is used to optimally adjust
1755 some packet sizes while forwarding large and small data alternatively. The
1756 decision to use splice() or to send large buffers in SSL is modulated by this
1757 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1758 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001759 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05001760 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01001761 tune.ssl.maxrecord below.
1762
Willy Tarreau7ac908b2019-02-27 12:02:18 +01001763tune.listener.multi-queue { on | off }
1764 Enables ('on') or disables ('off') the listener's multi-queue accept which
1765 spreads the incoming traffic to all threads a "bind" line is allowed to run
1766 on instead of taking them for itself. This provides a smoother traffic
1767 distribution and scales much better, especially in environments where threads
1768 may be unevenly loaded due to external activity (network interrupts colliding
1769 with one thread for example). This option is enabled by default, but it may
1770 be forcefully disabled for troubleshooting or for situations where it is
1771 estimated that the operating system already provides a good enough
1772 distribution and connections are extremely short-lived.
1773
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001774tune.lua.forced-yield <number>
1775 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001776 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001777 HAProxy scheduler to process other tasks like accepting connections or
1778 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001779 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001780 lowered. If the Lua code is quite long and its result is absolutely required
1781 to process the data, the <number> can be increased.
1782
Willy Tarreau32f61e22015-03-18 17:54:59 +01001783tune.lua.maxmem
1784 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1785 default it is zero which means unlimited. It is important to set a limit to
1786 ensure that a bug in a script will not result in the system running out of
1787 memory.
1788
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001789tune.lua.session-timeout <timeout>
1790 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001791 preventing infinite loops or spending too much time in Lua. This timeout
1792 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001793 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001794
1795tune.lua.task-timeout <timeout>
1796 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1797 dedicated to the tasks. By default, this timeout isn't set because a task may
1798 remain alive during of the lifetime of HAProxy. For example, a task used to
1799 check servers.
1800
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001801tune.lua.service-timeout <timeout>
1802 This is the execution timeout for the Lua services. This is useful for
1803 preventing infinite loops or spending too much time in Lua. This timeout
1804 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001805 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001806
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001807tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001808 Sets the maximum number of consecutive connections a process may accept in a
1809 row before switching to other work. In single process mode, higher numbers
1810 give better performance at high connection rates. However in multi-process
1811 modes, keeping a bit of fairness between processes generally is better to
1812 increase performance. This value applies individually to each listener, so
1813 that the number of processes a listener is bound to is taken into account.
1814 This value defaults to 64. In multi-process mode, it is divided by twice
1815 the number of processes the listener is bound to. Setting this value to -1
1816 completely disables the limitation. It should normally not be needed to tweak
1817 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001818
1819tune.maxpollevents <number>
1820 Sets the maximum amount of events that can be processed at once in a call to
1821 the polling system. The default value is adapted to the operating system. It
1822 has been noticed that reducing it below 200 tends to slightly decrease
1823 latency at the expense of network bandwidth, and increasing it above 200
1824 tends to trade latency for slightly increased bandwidth.
1825
Willy Tarreau27a674e2009-08-17 07:23:33 +02001826tune.maxrewrite <number>
1827 Sets the reserved buffer space to this size in bytes. The reserved space is
1828 used for header rewriting or appending. The first reads on sockets will never
1829 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1830 bufsize, though that does not make much sense since there are rarely large
1831 numbers of headers to add. Setting it too high prevents processing of large
1832 requests or responses. Setting it too low prevents addition of new headers
1833 to already large requests or to POST requests. It is generally wise to set it
1834 to about 1024. It is automatically readjusted to half of bufsize if it is
1835 larger than that. This means you don't have to worry about it when changing
1836 bufsize.
1837
Willy Tarreauf3045d22015-04-29 16:24:50 +02001838tune.pattern.cache-size <number>
1839 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1840 cache which reminds previous lookups and their results. It is used by ACLs
1841 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1842 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1843 strings. It applies to pattern expressions which means that it will be able
1844 to memorize the result of a lookup among all the patterns specified on a
1845 configuration line (including all those loaded from files). It automatically
1846 invalidates entries which are updated using HTTP actions or on the CLI. The
1847 default cache size is set to 10000 entries, which limits its footprint to
1848 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1849 risk of collision in this cache, which is in the order of the size of the
1850 cache divided by 2^64. Typically, at 10000 requests per second with the
1851 default cache size of 10000 entries, there's 1% chance that a brute force
1852 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1853 This is considered much lower than the risk of a memory corruption caused by
1854 aging components. If this is not acceptable, the cache can be disabled by
1855 setting this parameter to 0.
1856
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001857tune.pipesize <number>
1858 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1859 are the default size for the system. But sometimes when using TCP splicing,
1860 it can improve performance to increase pipe sizes, especially if it is
1861 suspected that pipes are not filled and that many calls to splice() are
1862 performed. This has an impact on the kernel's memory footprint, so this must
1863 not be changed if impacts are not understood.
1864
Olivier Houchard88698d92019-04-16 19:07:22 +02001865tune.pool-low-fd-ratio <number>
1866 This setting sets the max number of file descriptors (in percentage) used by
1867 haproxy globally against the maximum number of file descriptors haproxy can
1868 use before we stop putting connection into the idle pool for reuse. The
1869 default is 20.
1870
1871tune.pool-high-fd-ratio <number>
1872 This setting sets the max number of file descriptors (in percentage) used by
1873 haproxy globally against the maximum number of file descriptors haproxy can
1874 use before we start killing idle connections when we can't reuse a connection
1875 and we have to create a new one. The default is 25 (one quarter of the file
1876 descriptor will mean that roughly half of the maximum front connections can
1877 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05001878 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02001879
Willy Tarreaue803de22010-01-21 17:43:04 +01001880tune.rcvbuf.client <number>
1881tune.rcvbuf.server <number>
1882 Forces the kernel socket receive buffer size on the client or the server side
1883 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1884 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05001885 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001886 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001887 order to save kernel memory by preventing it from buffering too large amounts
1888 of received data. Lower values will significantly increase CPU usage though.
1889
Willy Tarreaub22fc302015-12-14 12:04:35 +01001890tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001891 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001892 socket buffers. One of them is that a read returns more than <recv_enough>
1893 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1894 may be changed by this setting to better deal with workloads involving lots
1895 of short messages such as telnet or SSH sessions.
1896
Olivier Houchard1599b802018-05-24 18:59:04 +02001897tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05001898 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02001899 tasks. The default value is 200. Increasing it may incur latency when
1900 dealing with I/Os, making it too small can incur extra overhead.
1901
Willy Tarreaue803de22010-01-21 17:43:04 +01001902tune.sndbuf.client <number>
1903tune.sndbuf.server <number>
1904 Forces the kernel socket send buffer size on the client or the server side to
1905 the specified value in bytes. This value applies to all TCP/HTTP frontends
1906 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05001907 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001908 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001909 order to save kernel memory by preventing it from buffering too large amounts
1910 of received data. Lower values will significantly increase CPU usage though.
1911 Another use case is to prevent write timeouts with extremely slow clients due
1912 to the kernel waiting for a large part of the buffer to be read before
1913 notifying haproxy again.
1914
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001915tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001916 Sets the size of the global SSL session cache, in a number of blocks. A block
1917 is large enough to contain an encoded session without peer certificate.
1918 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001919 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001920 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001921 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001922 and reassigned. Higher values reduce the occurrence of such a purge, hence
1923 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1924 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001925 and are shared between all processes if "nbproc" is greater than 1. Setting
1926 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001927
Emeric Brun8dc60392014-05-09 13:52:00 +02001928tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001929 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001930 should normally not be used since it will force many renegotiations due to
1931 clients hitting a random process. But it may be required on some operating
1932 systems where none of the SSL cache synchronization method may be used. In
1933 this case, adding a first layer of hash-based load balancing before the SSL
1934 layer might limit the impact of the lack of session sharing.
1935
Emeric Brun4f65bff2012-11-16 15:11:00 +01001936tune.ssl.lifetime <timeout>
1937 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001938 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001939 does not guarantee that sessions will last that long, because if the cache is
1940 full, the longest idle sessions will be purged despite their configured
1941 lifetime. The real usefulness of this setting is to prevent sessions from
1942 being used for too long.
1943
Willy Tarreaubfd59462013-02-21 07:46:09 +01001944tune.ssl.maxrecord <number>
1945 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1946 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1947 data only once it has received a full record. With large records, it means
1948 that clients might have to download up to 16kB of data before starting to
1949 process them. Limiting the value can improve page load times on browsers
1950 located over high latency or low bandwidth networks. It is suggested to find
1951 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1952 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1953 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1954 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001955 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001956 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001957
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001958tune.ssl.default-dh-param <number>
1959 Sets the maximum size of the Diffie-Hellman parameters used for generating
1960 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1961 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1962 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1963 this maximum value. Default value if 1024. Only 1024 or higher values are
1964 allowed. Higher values will increase the CPU load, and values greater than
1965 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001966 used if static Diffie-Hellman parameters are supplied either directly
1967 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001968
Christopher Faulet31af49d2015-06-09 17:29:50 +02001969tune.ssl.ssl-ctx-cache-size <number>
1970 Sets the size of the cache used to store generated certificates to <number>
1971 entries. This is a LRU cache. Because generating a SSL certificate
1972 dynamically is expensive, they are cached. The default cache size is set to
1973 1000 entries.
1974
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001975tune.ssl.capture-cipherlist-size <number>
1976 Sets the maximum size of the buffer used for capturing client-hello cipher
1977 list. If the value is 0 (default value) the capture is disabled, otherwise
1978 a buffer is allocated for each SSL/TLS connection.
1979
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001980tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001981tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001982tune.vars.reqres-max-size <size>
1983tune.vars.sess-max-size <size>
1984tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001985 These five tunes help to manage the maximum amount of memory used by the
1986 variables system. "global" limits the overall amount of memory available for
1987 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1988 memory for the session scope, "txn" for the transaction scope, and "reqres"
1989 limits the memory for each request or response processing.
1990 Memory accounting is hierarchical, meaning more coarse grained limits include
1991 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1992 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001993
Daniel Schneller0b547052016-03-21 20:46:57 +01001994 For example, when "tune.vars.sess-max-size" is limited to 100,
1995 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1996 100 either. If we create a variable "txn.var" that contains 100 bytes,
1997 all available space is consumed.
1998 Notice that exceeding the limits at runtime will not result in an error
1999 message, but values might be cut off or corrupted. So make sure to accurately
2000 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002001
William Lallemanda509e4c2012-11-07 16:54:34 +01002002tune.zlib.memlevel <number>
2003 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002004 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002005 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002006 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002007 between 1 and 9. The default value is 8.
2008
2009tune.zlib.windowsize <number>
2010 Sets the window size (the size of the history buffer) as a parameter of the
2011 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002012 in better compression at the expense of memory usage. Can be a value between
2013 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002014
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020153.3. Debugging
2016--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002017
2018debug
2019 Enables debug mode which dumps to stdout all exchanges, and disables forking
2020 into background. It is the equivalent of the command-line argument "-d". It
2021 should never be used in a production configuration since it may prevent full
2022 system startup.
2023
2024quiet
2025 Do not display any message during startup. It is equivalent to the command-
2026 line argument "-q".
2027
Emeric Brunf099e792010-09-27 12:05:28 +02002028
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010020293.4. Userlists
2030--------------
2031It is possible to control access to frontend/backend/listen sections or to
2032http stats by allowing only authenticated and authorized users. To do this,
2033it is required to create at least one userlist and to define users.
2034
2035userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002036 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002037 used to store authentication & authorization data for independent customers.
2038
2039group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002040 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002041 attach users to this group by using a comma separated list of names
2042 proceeded by "users" keyword.
2043
Cyril Bontéf0c60612010-02-06 14:44:47 +01002044user <username> [password|insecure-password <password>]
2045 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002046 Adds user <username> to the current userlist. Both secure (encrypted) and
2047 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002048 evaluated using the crypt(3) function, so depending on the system's
2049 capabilities, different algorithms are supported. For example, modern Glibc
2050 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2051 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002052
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002053 Attention: Be aware that using encrypted passwords might cause significantly
2054 increased CPU usage, depending on the number of requests, and the algorithm
2055 used. For any of the hashed variants, the password for each request must
2056 be processed through the chosen algorithm, before it can be compared to the
2057 value specified in the config file. Most current algorithms are deliberately
2058 designed to be expensive to compute to achieve resistance against brute
2059 force attacks. They do not simply salt/hash the clear text password once,
2060 but thousands of times. This can quickly become a major factor in haproxy's
2061 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002062
2063 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002064 userlist L1
2065 group G1 users tiger,scott
2066 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002067
Cyril Bontéf0c60612010-02-06 14:44:47 +01002068 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2069 user scott insecure-password elgato
2070 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002071
Cyril Bontéf0c60612010-02-06 14:44:47 +01002072 userlist L2
2073 group G1
2074 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002075
Cyril Bontéf0c60612010-02-06 14:44:47 +01002076 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2077 user scott insecure-password elgato groups G1,G2
2078 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002079
2080 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002081
Emeric Brunf099e792010-09-27 12:05:28 +02002082
20833.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002084----------
Emeric Brun94900952015-06-11 18:25:54 +02002085It is possible to propagate entries of any data-types in stick-tables between
2086several haproxy instances over TCP connections in a multi-master fashion. Each
2087instance pushes its local updates and insertions to remote peers. The pushed
2088values overwrite remote ones without aggregation. Interrupted exchanges are
2089automatically detected and recovered from the last known point.
2090In addition, during a soft restart, the old process connects to the new one
2091using such a TCP connection to push all its entries before the new process
2092tries to connect to other peers. That ensures very fast replication during a
2093reload, it typically takes a fraction of a second even for large tables.
2094Note that Server IDs are used to identify servers remotely, so it is important
2095that configurations look similar or at least that the same IDs are forced on
2096each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002097
2098peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002099 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002100 which is referenced by one or more stick-tables.
2101
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002102bind [<address>]:<port_range> [, ...] [param*]
2103 Defines the binding parameters of the local peer of this "peers" section.
2104 Such lines are not supported with "peer" line in the same "peers" section.
2105
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002106disabled
2107 Disables a peers section. It disables both listening and any synchronization
2108 related to this section. This is provided to disable synchronization of stick
2109 tables without having to comment out all "peers" references.
2110
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002111default-bind [param*]
2112 Defines the binding parameters for the local peer, excepted its address.
2113
2114default-server [param*]
2115 Change default options for a server in a "peers" section.
2116
2117 Arguments:
2118 <param*> is a list of parameters for this server. The "default-server"
2119 keyword accepts an important number of options and has a complete
2120 section dedicated to it. Please refer to section 5 for more
2121 details.
2122
2123
2124 See also: "server" and section 5 about server options
2125
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002126enable
2127 This re-enables a disabled peers section which was previously disabled.
2128
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002129peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002130 Defines a peer inside a peers section.
2131 If <peername> is set to the local peer name (by default hostname, or forced
2132 using "-L" command line option), haproxy will listen for incoming remote peer
2133 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
2134 to join the remote peer, and <peername> is used at the protocol level to
2135 identify and validate the remote peer on the server side.
2136
2137 During a soft restart, local peer <ip>:<port> is used by the old instance to
2138 connect the new one and initiate a complete replication (teaching process).
2139
2140 It is strongly recommended to have the exact same peers declaration on all
2141 peers and to only rely on the "-L" command line argument to change the local
2142 peer name. This makes it easier to maintain coherent configuration files
2143 across all peers.
2144
William Lallemandb2f07452015-05-12 14:27:13 +02002145 You may want to reference some environment variables in the address
2146 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002147
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002148 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2149 "server" keyword explanation below).
2150
2151server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002152 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002153 with a support for all "server" parameters found in 5.2 paragraph.
2154 If the underlying peer is local, <ip>:<port> parameters must not be present.
2155 These parameters must be provided on a "bind" line (see "bind" keyword
2156 of this "peers" section).
2157 Some of these parameters are irrelevant for "peers" sections.
2158
2159
Cyril Bontédc4d9032012-04-08 21:57:39 +02002160 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002161 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002162 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002163 peer haproxy1 192.168.0.1:1024
2164 peer haproxy2 192.168.0.2:1024
2165 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002166
2167 backend mybackend
2168 mode tcp
2169 balance roundrobin
2170 stick-table type ip size 20k peers mypeers
2171 stick on src
2172
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002173 server srv1 192.168.0.30:80
2174 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002175
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002176 Example:
2177 peers mypeers
2178 bind 127.0.0.11:10001 ssl crt mycerts/pem
2179 default-server ssl verify none
2180 server hostA 127.0.0.10:10000
2181 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002182
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002183
2184table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2185 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2186
2187 Configure a stickiness table for the current section. This line is parsed
2188 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002189 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002190 mandatory first parameter to designate the stick-table. Contrary to others
2191 sections, there may be several "table" lines in "peers" sections (see also
2192 "stick-table" keyword).
2193
2194 Also be aware of the fact that "peers" sections have their own stick-table
2195 namespaces to avoid collisions between stick-table names identical in
2196 different "peers" section. This is internally handled prepending the "peers"
2197 sections names to the name of the stick-tables followed by a '/' character.
2198 If somewhere else in the configuration file you have to refer to such
2199 stick-tables declared in "peers" sections you must use the prefixed version
2200 of the stick-table name as follows:
2201
2202 peers mypeers
2203 peer A ...
2204 peer B ...
2205 table t1 ...
2206
2207 frontend fe1
2208 tcp-request content track-sc0 src table mypeers/t1
2209
2210 This is also this prefixed version of the stick-table names which must be
2211 used to refer to stick-tables through the CLI.
2212
2213 About "peers" protocol, as only "peers" belonging to the same section may
2214 communicate with each others, there is no need to do such a distinction.
2215 Several "peers" sections may declare stick-tables with the same name.
2216 This is shorter version of the stick-table name which is sent over the network.
2217 There is only a '/' character as prefix to avoid stick-table name collisions between
2218 stick-tables declared as backends and stick-table declared in "peers" sections
2219 as follows in this weird but supported configuration:
2220
2221 peers mypeers
2222 peer A ...
2223 peer B ...
2224 table t1 type string size 10m store gpc0
2225
2226 backend t1
2227 stick-table type string size 10m store gpc0 peers mypeers
2228
2229 Here "t1" table declared in "mypeeers" section has "mypeers/t1" as global name.
2230 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2231 level the former table is named "/t1", the latter is again named "t1".
2232
Simon Horman51a1cf62015-02-03 13:00:44 +090022333.6. Mailers
2234------------
2235It is possible to send email alerts when the state of servers changes.
2236If configured email alerts are sent to each mailer that is configured
2237in a mailers section. Email is sent to mailers using SMTP.
2238
Pieter Baauw386a1272015-08-16 15:26:24 +02002239mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002240 Creates a new mailer list with the name <mailersect>. It is an
2241 independent section which is referenced by one or more proxies.
2242
2243mailer <mailername> <ip>:<port>
2244 Defines a mailer inside a mailers section.
2245
2246 Example:
2247 mailers mymailers
2248 mailer smtp1 192.168.0.1:587
2249 mailer smtp2 192.168.0.2:587
2250
2251 backend mybackend
2252 mode tcp
2253 balance roundrobin
2254
2255 email-alert mailers mymailers
2256 email-alert from test1@horms.org
2257 email-alert to test2@horms.org
2258
2259 server srv1 192.168.0.30:80
2260 server srv2 192.168.0.31:80
2261
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002262timeout mail <time>
2263 Defines the time available for a mail/connection to be made and send to
2264 the mail-server. If not defined the default value is 10 seconds. To allow
2265 for at least two SYN-ACK packets to be send during initial TCP handshake it
2266 is advised to keep this value above 4 seconds.
2267
2268 Example:
2269 mailers mymailers
2270 timeout mail 20s
2271 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002272
William Lallemandc9515522019-06-12 16:32:11 +020022733.7. Programs
2274-------------
2275In master-worker mode, it is possible to launch external binaries with the
2276master, these processes are called programs. These programs are launched and
2277managed the same way as the workers.
2278
2279During a reload of HAProxy, those processes are dealing with the same
2280sequence as a worker:
2281
2282 - the master is re-executed
2283 - the master sends a SIGUSR1 signal to the program
2284 - if "option start-on-reload" is not disabled, the master launches a new
2285 instance of the program
2286
2287During a stop, or restart, a SIGTERM is sent to the programs.
2288
2289program <name>
2290 This is a new program section, this section will create an instance <name>
2291 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2292 the management guide).
2293
2294command <command> [arguments*]
2295 Define the command to start with optional arguments. The command is looked
2296 up in the current PATH if it does not include an absolute path. This is a
2297 mandatory option of the program section. Arguments containing spaces must
2298 be enclosed in quotes or double quotes or be prefixed by a backslash.
2299
Andrew Heberle97236962019-07-12 11:50:26 +08002300user <user name>
2301 Changes the executed command user ID to the <user name> from /etc/passwd.
2302 See also "group".
2303
2304group <group name>
2305 Changes the executed command group ID to the <group name> from /etc/group.
2306 See also "user".
2307
William Lallemandc9515522019-06-12 16:32:11 +02002308option start-on-reload
2309no option start-on-reload
2310 Start (or not) a new instance of the program upon a reload of the master.
2311 The default is to start a new instance. This option may only be used in a
2312 program section.
2313
2314
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023154. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002316----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002317
Willy Tarreau6a06a402007-07-15 20:15:28 +02002318Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002319 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002320 - frontend <name>
2321 - backend <name>
2322 - listen <name>
2323
2324A "defaults" section sets default parameters for all other sections following
2325its declaration. Those default parameters are reset by the next "defaults"
2326section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002327section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002328
2329A "frontend" section describes a set of listening sockets accepting client
2330connections.
2331
2332A "backend" section describes a set of servers to which the proxy will connect
2333to forward incoming connections.
2334
2335A "listen" section defines a complete proxy with its frontend and backend
2336parts combined in one section. It is generally useful for TCP-only traffic.
2337
Willy Tarreau0ba27502007-12-24 16:55:16 +01002338All proxy names must be formed from upper and lower case letters, digits,
2339'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2340case-sensitive, which means that "www" and "WWW" are two different proxies.
2341
2342Historically, all proxy names could overlap, it just caused troubles in the
2343logs. Since the introduction of content switching, it is mandatory that two
2344proxies with overlapping capabilities (frontend/backend) have different names.
2345However, it is still permitted that a frontend and a backend share the same
2346name, as this configuration seems to be commonly encountered.
2347
2348Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2349and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002350bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002351protocol, and can interact with it by allowing, blocking, switching, adding,
2352modifying, or removing arbitrary contents in requests or responses, based on
2353arbitrary criteria.
2354
Willy Tarreau70dffda2014-01-30 03:07:23 +01002355In HTTP mode, the processing applied to requests and responses flowing over
2356a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02002357the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002358
2359 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2360 requests and responses are processed, and connections remain open but idle
2361 between responses and new requests.
2362
Willy Tarreau70dffda2014-01-30 03:07:23 +01002363 - SCL: server close ("option http-server-close") : the server-facing
2364 connection is closed after the end of the response is received, but the
2365 client-facing connection remains open.
2366
Christopher Faulet315b39c2018-09-21 16:26:19 +02002367 - CLO: close ("option httpclose"): the connection is closed after the end of
2368 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002369
2370The effective mode that will be applied to a connection passing through a
2371frontend and a backend can be determined by both proxy modes according to the
2372following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002373weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002374
Christopher Faulet315b39c2018-09-21 16:26:19 +02002375 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002376
Christopher Faulet315b39c2018-09-21 16:26:19 +02002377 | KAL | SCL | CLO
2378 ----+-----+-----+----
2379 KAL | KAL | SCL | CLO
2380 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02002381 mode SCL | SCL | SCL | CLO
2382 ----+-----+-----+----
2383 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002384
Willy Tarreau0ba27502007-12-24 16:55:16 +01002385
Willy Tarreau70dffda2014-01-30 03:07:23 +01002386
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023874.1. Proxy keywords matrix
2388--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002389
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002390The following list of keywords is supported. Most of them may only be used in a
2391limited set of section types. Some of them are marked as "deprecated" because
2392they are inherited from an old syntax which may be confusing or functionally
2393limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002394marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002395option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002396and must be disabled for a specific instance. Such options may also be prefixed
2397with "default" in order to restore default settings regardless of what has been
2398specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002399
Willy Tarreau6a06a402007-07-15 20:15:28 +02002400
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002401 keyword defaults frontend listen backend
2402------------------------------------+----------+----------+---------+---------
2403acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002404backlog X X X -
2405balance X - X X
2406bind - X X -
2407bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002408capture cookie - X X -
2409capture request header - X X -
2410capture response header - X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002411compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002412cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002413declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002414default-server X - X X
2415default_backend X X X -
2416description - X X X
2417disabled X X X X
2418dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002419email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002420email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002421email-alert mailers X X X X
2422email-alert myhostname X X X X
2423email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002424enabled X X X X
2425errorfile X X X X
2426errorloc X X X X
2427errorloc302 X X X X
2428-- keyword -------------------------- defaults - frontend - listen -- backend -
2429errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002430force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002431filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002432fullconn X - X X
2433grace X X X X
2434hash-type X - X X
2435http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002436http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002437http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002438http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002439http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002440http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002441http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002442id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002443ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002444load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002445log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002446log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002447log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002448log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002449max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002450maxconn X X X -
2451mode X X X X
2452monitor fail - X X -
2453monitor-net X X X -
2454monitor-uri X X X -
2455option abortonclose (*) X - X X
2456option accept-invalid-http-request (*) X X X -
2457option accept-invalid-http-response (*) X - X X
2458option allbackups (*) X - X X
2459option checkcache (*) X - X X
2460option clitcpka (*) X X X -
2461option contstats (*) X X X -
2462option dontlog-normal (*) X X X -
2463option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002464-- keyword -------------------------- defaults - frontend - listen -- backend -
2465option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02002466option h1-case-adjust-bogus-client (*) X X X -
2467option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002468option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002469option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002470option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002471option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002472option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002473option http-server-close (*) X X X X
2474option http-use-proxy-header (*) X X X -
2475option httpchk X - X X
2476option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01002477option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002478option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002479option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002480option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002481option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002482option log-health-checks (*) X - X X
2483option log-separate-errors (*) X X X -
2484option logasap (*) X X X -
2485option mysql-check X - X X
2486option nolinger (*) X X X X
2487option originalto X X X X
2488option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002489option pgsql-check X - X X
2490option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002491option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002492option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002493option smtpchk X - X X
2494option socket-stats (*) X X X -
2495option splice-auto (*) X X X X
2496option splice-request (*) X X X X
2497option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002498option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002499option srvtcpka (*) X - X X
2500option ssl-hello-chk X - X X
2501-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002502option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002503option tcp-smart-accept (*) X X X -
2504option tcp-smart-connect (*) X - X X
2505option tcpka X X X X
2506option tcplog X X X X
2507option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002508external-check command X - X X
2509external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002510persist rdp-cookie X - X X
2511rate-limit sessions X X X -
2512redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002513-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002514retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02002515retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002516server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002517server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002518server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002519source X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002520stats admin - X X X
2521stats auth X X X X
2522stats enable X X X X
2523stats hide-version X X X X
2524stats http-request - X X X
2525stats realm X X X X
2526stats refresh X X X X
2527stats scope X X X X
2528stats show-desc X X X X
2529stats show-legends X X X X
2530stats show-node X X X X
2531stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002532-- keyword -------------------------- defaults - frontend - listen -- backend -
2533stick match - - X X
2534stick on - - X X
2535stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002536stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002537stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002538tcp-check connect - - X X
2539tcp-check expect - - X X
2540tcp-check send - - X X
2541tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002542tcp-request connection - X X -
2543tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002544tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002545tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002546tcp-response content - - X X
2547tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002548timeout check X - X X
2549timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002550timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002551timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002552timeout http-keep-alive X X X X
2553timeout http-request X X X X
2554timeout queue X - X X
2555timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002556timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002557timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002558timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002559transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002560unique-id-format X X X -
2561unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002562use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02002563use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002564use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002565------------------------------------+----------+----------+---------+---------
2566 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002567
Willy Tarreau0ba27502007-12-24 16:55:16 +01002568
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025694.2. Alphabetically sorted keywords reference
2570---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002571
2572This section provides a description of each keyword and its usage.
2573
2574
2575acl <aclname> <criterion> [flags] [operator] <value> ...
2576 Declare or complete an access list.
2577 May be used in sections : defaults | frontend | listen | backend
2578 no | yes | yes | yes
2579 Example:
2580 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2581 acl invalid_src src_port 0:1023
2582 acl local_dst hdr(host) -i localhost
2583
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002584 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002585
2586
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002587backlog <conns>
2588 Give hints to the system about the approximate listen backlog desired size
2589 May be used in sections : defaults | frontend | listen | backend
2590 yes | yes | yes | no
2591 Arguments :
2592 <conns> is the number of pending connections. Depending on the operating
2593 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002594 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002595
2596 In order to protect against SYN flood attacks, one solution is to increase
2597 the system's SYN backlog size. Depending on the system, sometimes it is just
2598 tunable via a system parameter, sometimes it is not adjustable at all, and
2599 sometimes the system relies on hints given by the application at the time of
2600 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2601 to the listen() syscall. On systems which can make use of this value, it can
2602 sometimes be useful to be able to specify a different value, hence this
2603 backlog parameter.
2604
2605 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2606 used as a hint and the system accepts up to the smallest greater power of
2607 two, and never more than some limits (usually 32768).
2608
2609 See also : "maxconn" and the target operating system's tuning guide.
2610
2611
Willy Tarreau0ba27502007-12-24 16:55:16 +01002612balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002613balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002614 Define the load balancing algorithm to be used in a backend.
2615 May be used in sections : defaults | frontend | listen | backend
2616 yes | no | yes | yes
2617 Arguments :
2618 <algorithm> is the algorithm used to select a server when doing load
2619 balancing. This only applies when no persistence information
2620 is available, or when a connection is redispatched to another
2621 server. <algorithm> may be one of the following :
2622
2623 roundrobin Each server is used in turns, according to their weights.
2624 This is the smoothest and fairest algorithm when the server's
2625 processing time remains equally distributed. This algorithm
2626 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002627 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002628 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002629 large farms, when a server becomes up after having been down
2630 for a very short time, it may sometimes take a few hundreds
2631 requests for it to be re-integrated into the farm and start
2632 receiving traffic. This is normal, though very rare. It is
2633 indicated here in case you would have the chance to observe
2634 it, so that you don't worry.
2635
2636 static-rr Each server is used in turns, according to their weights.
2637 This algorithm is as similar to roundrobin except that it is
2638 static, which means that changing a server's weight on the
2639 fly will have no effect. On the other hand, it has no design
2640 limitation on the number of servers, and when a server goes
2641 up, it is always immediately reintroduced into the farm, once
2642 the full map is recomputed. It also uses slightly less CPU to
2643 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002644
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002645 leastconn The server with the lowest number of connections receives the
2646 connection. Round-robin is performed within groups of servers
2647 of the same load to ensure that all servers will be used. Use
2648 of this algorithm is recommended where very long sessions are
2649 expected, such as LDAP, SQL, TSE, etc... but is not very well
2650 suited for protocols using short sessions such as HTTP. This
2651 algorithm is dynamic, which means that server weights may be
2652 adjusted on the fly for slow starts for instance.
2653
Willy Tarreauf09c6602012-02-13 17:12:08 +01002654 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002655 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002656 identifier to the highest (see server parameter "id"), which
2657 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002658 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002659 not make sense to use this algorithm without setting maxconn.
2660 The purpose of this algorithm is to always use the smallest
2661 number of servers so that extra servers can be powered off
2662 during non-intensive hours. This algorithm ignores the server
2663 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002664 or IMAP than HTTP, though it can be useful there too. In
2665 order to use this algorithm efficiently, it is recommended
2666 that a cloud controller regularly checks server usage to turn
2667 them off when unused, and regularly checks backend queue to
2668 turn new servers on when the queue inflates. Alternatively,
2669 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002670
Willy Tarreau0ba27502007-12-24 16:55:16 +01002671 source The source IP address is hashed and divided by the total
2672 weight of the running servers to designate which server will
2673 receive the request. This ensures that the same client IP
2674 address will always reach the same server as long as no
2675 server goes down or up. If the hash result changes due to the
2676 number of running servers changing, many clients will be
2677 directed to a different server. This algorithm is generally
2678 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002679 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002680 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002681 static by default, which means that changing a server's
2682 weight on the fly will have no effect, but this can be
2683 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002684
Oskar Stolc8dc41842012-05-19 10:19:54 +01002685 uri This algorithm hashes either the left part of the URI (before
2686 the question mark) or the whole URI (if the "whole" parameter
2687 is present) and divides the hash value by the total weight of
2688 the running servers. The result designates which server will
2689 receive the request. This ensures that the same URI will
2690 always be directed to the same server as long as no server
2691 goes up or down. This is used with proxy caches and
2692 anti-virus proxies in order to maximize the cache hit rate.
2693 Note that this algorithm may only be used in an HTTP backend.
2694 This algorithm is static by default, which means that
2695 changing a server's weight on the fly will have no effect,
2696 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002697
Oskar Stolc8dc41842012-05-19 10:19:54 +01002698 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002699 "depth", both followed by a positive integer number. These
2700 options may be helpful when it is needed to balance servers
2701 based on the beginning of the URI only. The "len" parameter
2702 indicates that the algorithm should only consider that many
2703 characters at the beginning of the URI to compute the hash.
2704 Note that having "len" set to 1 rarely makes sense since most
2705 URIs start with a leading "/".
2706
2707 The "depth" parameter indicates the maximum directory depth
2708 to be used to compute the hash. One level is counted for each
2709 slash in the request. If both parameters are specified, the
2710 evaluation stops when either is reached.
2711
Willy Tarreau0ba27502007-12-24 16:55:16 +01002712 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002713 the query string of each HTTP GET request.
2714
2715 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002716 request entity will be searched for the parameter argument,
2717 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002718 ('?') in the URL. The message body will only start to be
2719 analyzed once either the advertised amount of data has been
2720 received or the request buffer is full. In the unlikely event
2721 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002722 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002723 be randomly balanced if at all. This keyword used to support
2724 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002725
2726 If the parameter is found followed by an equal sign ('=') and
2727 a value, then the value is hashed and divided by the total
2728 weight of the running servers. The result designates which
2729 server will receive the request.
2730
2731 This is used to track user identifiers in requests and ensure
2732 that a same user ID will always be sent to the same server as
2733 long as no server goes up or down. If no value is found or if
2734 the parameter is not found, then a round robin algorithm is
2735 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002736 backend. This algorithm is static by default, which means
2737 that changing a server's weight on the fly will have no
2738 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002739
Cyril Bontédc4d9032012-04-08 21:57:39 +02002740 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2741 request. Just as with the equivalent ACL 'hdr()' function,
2742 the header name in parenthesis is not case sensitive. If the
2743 header is absent or if it does not contain any value, the
2744 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002745
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002746 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002747 reducing the hash algorithm to the main domain part with some
2748 specific headers such as 'Host'. For instance, in the Host
2749 value "haproxy.1wt.eu", only "1wt" will be considered.
2750
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002751 This algorithm is static by default, which means that
2752 changing a server's weight on the fly will have no effect,
2753 but this can be changed using "hash-type".
2754
Willy Tarreau21c741a2019-01-14 18:14:27 +01002755 random
2756 random(<draws>)
2757 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02002758 hashing function. This means that the servers' weights are
2759 respected, dynamic weight changes immediately take effect, as
2760 well as new server additions. Random load balancing can be
2761 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01002762 or removed as it may avoid the hammering effect that could
2763 result from roundrobin or leastconn in this situation. The
2764 hash-balance-factor directive can be used to further improve
2765 fairness of the load balancing, especially in situations
2766 where servers show highly variable response times. When an
2767 argument <draws> is present, it must be an integer value one
2768 or greater, indicating the number of draws before selecting
2769 the least loaded of these servers. It was indeed demonstrated
2770 that picking the least loaded of two servers is enough to
2771 significantly improve the fairness of the algorithm, by
2772 always avoiding to pick the most loaded server within a farm
2773 and getting rid of any bias that could be induced by the
2774 unfair distribution of the consistent list. Higher values N
2775 will take away N-1 of the highest loaded servers at the
2776 expense of performance. With very high values, the algorithm
2777 will converge towards the leastconn's result but much slower.
2778 The default value is 2, which generally shows very good
2779 distribution and performance. This algorithm is also known as
2780 the Power of Two Random Choices and is described here :
2781 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02002782
Emeric Brun736aa232009-06-30 17:56:00 +02002783 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002784 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002785 The RDP cookie <name> (or "mstshash" if omitted) will be
2786 looked up and hashed for each incoming TCP request. Just as
2787 with the equivalent ACL 'req_rdp_cookie()' function, the name
2788 is not case-sensitive. This mechanism is useful as a degraded
2789 persistence mode, as it makes it possible to always send the
2790 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002791 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002792 used instead.
2793
2794 Note that for this to work, the frontend must ensure that an
2795 RDP cookie is already present in the request buffer. For this
2796 you must use 'tcp-request content accept' rule combined with
2797 a 'req_rdp_cookie_cnt' ACL.
2798
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002799 This algorithm is static by default, which means that
2800 changing a server's weight on the fly will have no effect,
2801 but this can be changed using "hash-type".
2802
Cyril Bontédc4d9032012-04-08 21:57:39 +02002803 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002804
Willy Tarreau0ba27502007-12-24 16:55:16 +01002805 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002806 algorithms. Right now, only "url_param" and "uri" support an
2807 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002808
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002809 The load balancing algorithm of a backend is set to roundrobin when no other
2810 algorithm, mode nor option have been set. The algorithm may only be set once
2811 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002812
Lukas Tribus80512b12018-10-27 20:07:40 +02002813 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05002814 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02002815 to be routed to different backend servers, breaking the invalid assumptions
2816 NTLM relies on.
2817
Willy Tarreau0ba27502007-12-24 16:55:16 +01002818 Examples :
2819 balance roundrobin
2820 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002821 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002822 balance hdr(User-Agent)
2823 balance hdr(host)
2824 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002825
2826 Note: the following caveats and limitations on using the "check_post"
2827 extension with "url_param" must be considered :
2828
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002829 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002830 to determine if the parameters will be found in the body or entity which
2831 may contain binary data. Therefore another method may be required to
2832 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02002833 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002834
2835 - using a <max_wait> value larger than the request buffer size does not
2836 make sense and is useless. The buffer size is set at build time, and
2837 defaults to 16 kB.
2838
2839 - Content-Encoding is not supported, the parameter search will probably
2840 fail; and load balancing will fall back to Round Robin.
2841
2842 - Expect: 100-continue is not supported, load balancing will fall back to
2843 Round Robin.
2844
Lukas Tribus23953682017-04-28 13:24:30 +00002845 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002846 If the entire parameter value is not present in the first chunk, the
2847 selection of server is undefined (actually, defined by how little
2848 actually appeared in the first chunk).
2849
2850 - This feature does not support generation of a 100, 411 or 501 response.
2851
2852 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002853 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002854 white space or control characters are found, indicating the end of what
2855 might be a URL parameter list. This is probably not a concern with SGML
2856 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002857
Willy Tarreau294d0f02015-08-10 19:40:12 +02002858 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002859
2860
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002861bind [<address>]:<port_range> [, ...] [param*]
2862bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002863 Define one or several listening addresses and/or ports in a frontend.
2864 May be used in sections : defaults | frontend | listen | backend
2865 no | yes | yes | no
2866 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002867 <address> is optional and can be a host name, an IPv4 address, an IPv6
2868 address, or '*'. It designates the address the frontend will
2869 listen on. If unset, all IPv4 addresses of the system will be
2870 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002871 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002872 Optionally, an address family prefix may be used before the
2873 address to force the family regardless of the address format,
2874 which can be useful to specify a path to a unix socket with
2875 no slash ('/'). Currently supported prefixes are :
2876 - 'ipv4@' -> address is always IPv4
2877 - 'ipv6@' -> address is always IPv6
2878 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002879 - 'abns@' -> address is in abstract namespace (Linux only).
2880 Note: since abstract sockets are not "rebindable", they
2881 do not cope well with multi-process mode during
2882 soft-restart, so it is better to avoid them if
2883 nbproc is greater than 1. The effect is that if the
2884 new process fails to start, only one of the old ones
2885 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002886 - 'fd@<n>' -> use file descriptor <n> inherited from the
2887 parent. The fd must be bound and may or may not already
2888 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002889 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2890 connected unix socket or of a socketpair. The bind waits
2891 to receive a FD over the unix socket and uses it as if it
2892 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002893 You may want to reference some environment variables in the
2894 address parameter, see section 2.3 about environment
2895 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002896
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002897 <port_range> is either a unique TCP port, or a port range for which the
2898 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002899 above. The port is mandatory for TCP listeners. Note that in
2900 the case of an IPv6 address, the port is always the number
2901 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002902 - a numerical port (ex: '80')
2903 - a dash-delimited ports range explicitly stating the lower
2904 and upper bounds (ex: '2000-2100') which are included in
2905 the range.
2906
2907 Particular care must be taken against port ranges, because
2908 every <address:port> couple consumes one socket (= a file
2909 descriptor), so it's easy to consume lots of descriptors
2910 with a simple range, and to run out of sockets. Also, each
2911 <address:port> couple must be used only once among all
2912 instances running on a same system. Please note that binding
2913 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002914 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002915 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002916
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002917 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002918 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002919 receive UNIX connections on the socket located at this place.
2920 The path must begin with a slash and by default is absolute.
2921 It can be relative to the prefix defined by "unix-bind" in
2922 the global section. Note that the total length of the prefix
2923 followed by the socket path cannot exceed some system limits
2924 for UNIX sockets, which commonly are set to 107 characters.
2925
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002926 <param*> is a list of parameters common to all sockets declared on the
2927 same line. These numerous parameters depend on OS and build
2928 options and have a complete section dedicated to them. Please
2929 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002930
Willy Tarreau0ba27502007-12-24 16:55:16 +01002931 It is possible to specify a list of address:port combinations delimited by
2932 commas. The frontend will then listen on all of these addresses. There is no
2933 fixed limit to the number of addresses and ports which can be listened on in
2934 a frontend, as well as there is no limit to the number of "bind" statements
2935 in a frontend.
2936
2937 Example :
2938 listen http_proxy
2939 bind :80,:443
2940 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002941 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002942
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002943 listen http_https_proxy
2944 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002945 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002946
Willy Tarreau24709282013-03-10 21:32:12 +01002947 listen http_https_proxy_explicit
2948 bind ipv6@:80
2949 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2950 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2951
Willy Tarreaudad36a32013-03-11 01:20:04 +01002952 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002953 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002954
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002955 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2956 sun_path length is used for the address length. Some other programs
2957 such as socat use the string length only by default. Pass the option
2958 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2959 make it compatible with HAProxy's.
2960
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002961 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002962 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002963
2964
Christopher Fauletff4121f2017-11-22 16:38:49 +01002965bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002966 Limit visibility of an instance to a certain set of processes numbers.
2967 May be used in sections : defaults | frontend | listen | backend
2968 yes | yes | yes | yes
2969 Arguments :
2970 all All process will see this instance. This is the default. It
2971 may be used to override a default value.
2972
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002973 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002974 option may be combined with other numbers.
2975
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002976 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002977 option may be combined with other numbers. Do not use it
2978 with less than 2 processes otherwise some instances might be
2979 missing from all processes.
2980
Christopher Fauletff4121f2017-11-22 16:38:49 +01002981 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002982 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002983 the machine's word size. Ranges can be partially defined. The
2984 higher bound can be omitted. In such case, it is replaced by
2985 the corresponding maximum value. If a proxy is bound to
2986 process numbers greater than the configured global.nbproc, it
2987 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02002988 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002989
2990 This keyword limits binding of certain instances to certain processes. This
2991 is useful in order not to have too many processes listening to the same
2992 ports. For instance, on a dual-core machine, it might make sense to set
2993 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2994 and 'even' instances.
2995
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002996 At the moment, it is not possible to reference more than 32 or 64 processes
2997 using this keyword, but this should be more than enough for most setups.
2998 Please note that 'all' really means all processes regardless of the machine's
2999 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003000
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003001 Each "bind" line may further be limited to a subset of the proxy's processes,
3002 please consult the "process" bind keyword in section 5.1.
3003
Willy Tarreaub369a042014-09-16 13:21:03 +02003004 When a frontend has no explicit "bind-process" line, it tries to bind to all
3005 the processes referenced by its "bind" lines. That means that frontends can
3006 easily adapt to their listeners' processes.
3007
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003008 If some backends are referenced by frontends bound to other processes, the
3009 backend automatically inherits the frontend's processes.
3010
3011 Example :
3012 listen app_ip1
3013 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003014 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003015
3016 listen app_ip2
3017 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003018 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003019
3020 listen management
3021 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003022 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003023
Willy Tarreau110ecc12012-11-15 17:50:01 +01003024 listen management
3025 bind 10.0.0.4:80
3026 bind-process 1-4
3027
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003028 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003029
3030
Willy Tarreau0ba27502007-12-24 16:55:16 +01003031capture cookie <name> len <length>
3032 Capture and log a cookie in the request and in the response.
3033 May be used in sections : defaults | frontend | listen | backend
3034 no | yes | yes | no
3035 Arguments :
3036 <name> is the beginning of the name of the cookie to capture. In order
3037 to match the exact name, simply suffix the name with an equal
3038 sign ('='). The full name will appear in the logs, which is
3039 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003040 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003041
3042 <length> is the maximum number of characters to report in the logs, which
3043 include the cookie name, the equal sign and the value, all in the
3044 standard "name=value" form. The string will be truncated on the
3045 right if it exceeds <length>.
3046
3047 Only the first cookie is captured. Both the "cookie" request headers and the
3048 "set-cookie" response headers are monitored. This is particularly useful to
3049 check for application bugs causing session crossing or stealing between
3050 users, because generally the user's cookies can only change on a login page.
3051
3052 When the cookie was not presented by the client, the associated log column
3053 will report "-". When a request does not cause a cookie to be assigned by the
3054 server, a "-" is reported in the response column.
3055
3056 The capture is performed in the frontend only because it is necessary that
3057 the log format does not change for a given frontend depending on the
3058 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003059 "capture cookie" statement in a frontend. The maximum capture length is set
3060 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3061 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003062
3063 Example:
3064 capture cookie ASPSESSION len 32
3065
3066 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003067 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003068
3069
3070capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003071 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003072 May be used in sections : defaults | frontend | listen | backend
3073 no | yes | yes | no
3074 Arguments :
3075 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003076 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003077 appear in the requests, with the first letter of each word in
3078 upper case. The header name will not appear in the logs, only the
3079 value is reported, but the position in the logs is respected.
3080
3081 <length> is the maximum number of characters to extract from the value and
3082 report in the logs. The string will be truncated on the right if
3083 it exceeds <length>.
3084
Willy Tarreau4460d032012-11-21 23:37:37 +01003085 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003086 value will be added to the logs between braces ('{}'). If multiple headers
3087 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003088 in the same order they were declared in the configuration. Non-existent
3089 headers will be logged just as an empty string. Common uses for request
3090 header captures include the "Host" field in virtual hosting environments, the
3091 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003092 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003093 environments to find where the request came from.
3094
3095 Note that when capturing headers such as "User-agent", some spaces may be
3096 logged, making the log analysis more difficult. Thus be careful about what
3097 you log if you know your log parser is not smart enough to rely on the
3098 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003099
Willy Tarreau0900abb2012-11-22 00:21:46 +01003100 There is no limit to the number of captured request headers nor to their
3101 length, though it is wise to keep them low to limit memory usage per session.
3102 In order to keep log format consistent for a same frontend, header captures
3103 can only be declared in a frontend. It is not possible to specify a capture
3104 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003105
3106 Example:
3107 capture request header Host len 15
3108 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003109 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003110
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003111 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003112 about logging.
3113
3114
3115capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003116 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003117 May be used in sections : defaults | frontend | listen | backend
3118 no | yes | yes | no
3119 Arguments :
3120 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003121 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003122 appear in the response, with the first letter of each word in
3123 upper case. The header name will not appear in the logs, only the
3124 value is reported, but the position in the logs is respected.
3125
3126 <length> is the maximum number of characters to extract from the value and
3127 report in the logs. The string will be truncated on the right if
3128 it exceeds <length>.
3129
Willy Tarreau4460d032012-11-21 23:37:37 +01003130 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003131 result will be added to the logs between braces ('{}') after the captured
3132 request headers. If multiple headers are captured, they will be delimited by
3133 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003134 the configuration. Non-existent headers will be logged just as an empty
3135 string. Common uses for response header captures include the "Content-length"
3136 header which indicates how many bytes are expected to be returned, the
3137 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003138
Willy Tarreau0900abb2012-11-22 00:21:46 +01003139 There is no limit to the number of captured response headers nor to their
3140 length, though it is wise to keep them low to limit memory usage per session.
3141 In order to keep log format consistent for a same frontend, header captures
3142 can only be declared in a frontend. It is not possible to specify a capture
3143 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003144
3145 Example:
3146 capture response header Content-length len 9
3147 capture response header Location len 15
3148
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003149 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003150 about logging.
3151
3152
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003153compression algo <algorithm> ...
3154compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003155compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003156 Enable HTTP compression.
3157 May be used in sections : defaults | frontend | listen | backend
3158 yes | yes | yes | yes
3159 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003160 algo is followed by the list of supported compression algorithms.
3161 type is followed by the list of MIME types that will be compressed.
3162 offload makes haproxy work as a compression offloader only (see notes).
3163
3164 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003165 identity this is mostly for debugging, and it was useful for developing
3166 the compression feature. Identity does not apply any change on
3167 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003168
Willy Tarreauc91840a2015-03-28 17:00:39 +01003169 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003170 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003171
3172 deflate same as "gzip", but with deflate algorithm and zlib format.
3173 Note that this algorithm has ambiguous support on many
3174 browsers and no support at all from recent ones. It is
3175 strongly recommended not to use it for anything else than
3176 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003177 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003178
Willy Tarreauc91840a2015-03-28 17:00:39 +01003179 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3180 alternative when the browser wants "deflate". All major
3181 browsers understand it and despite violating the standards,
3182 it is known to work better than "deflate", at least on MSIE
3183 and some versions of Safari. Do not use it in conjunction
3184 with "deflate", use either one or the other since both react
3185 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003186 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003187
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003188 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003189 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003190 If backend servers support HTTP compression, these directives
3191 will be no-op: haproxy will see the compressed response and will not
3192 compress again. If backend servers do not support HTTP compression and
3193 there is Accept-Encoding header in request, haproxy will compress the
3194 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003195
3196 The "offload" setting makes haproxy remove the Accept-Encoding header to
3197 prevent backend servers from compressing responses. It is strongly
3198 recommended not to do this because this means that all the compression work
3199 will be done on the single point where haproxy is located. However in some
3200 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003201 with broken HTTP compression implementation which can't be turned off.
3202 In that case haproxy can be used to prevent that gateway from emitting
3203 invalid payloads. In this case, simply removing the header in the
3204 configuration does not work because it applies before the header is parsed,
3205 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003206 then be used for such scenarios. Note: for now, the "offload" setting is
3207 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003208
William Lallemand05097442012-11-20 12:14:28 +01003209 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003210 * the request does not advertise a supported compression algorithm in the
3211 "Accept-Encoding" header
3212 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003213 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003214 * response contain neither a "Content-Length" header nor a
3215 "Transfer-Encoding" whose last value is "chunked"
3216 * response contains a "Content-Type" header whose first value starts with
3217 "multipart"
3218 * the response contains the "no-transform" value in the "Cache-control"
3219 header
3220 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3221 and later
3222 * The response contains a "Content-Encoding" header, indicating that the
3223 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003224 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003225
Tim Duesterhusb229f012019-01-29 16:38:56 +01003226 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003227
William Lallemand82fe75c2012-10-23 10:25:10 +02003228 Examples :
3229 compression algo gzip
3230 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003231
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003232
Willy Tarreau55165fe2009-05-10 12:02:55 +02003233cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003234 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3235 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003236 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003237 Enable cookie-based persistence in a backend.
3238 May be used in sections : defaults | frontend | listen | backend
3239 yes | no | yes | yes
3240 Arguments :
3241 <name> is the name of the cookie which will be monitored, modified or
3242 inserted in order to bring persistence. This cookie is sent to
3243 the client via a "Set-Cookie" header in the response, and is
3244 brought back by the client in a "Cookie" header in all requests.
3245 Special care should be taken to choose a name which does not
3246 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003247 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003248 HTTP/HTTPS), care should be taken to use different cookie names
3249 between all backends if persistence between them is not desired.
3250
3251 rewrite This keyword indicates that the cookie will be provided by the
3252 server and that haproxy will have to modify its value to set the
3253 server's identifier in it. This mode is handy when the management
3254 of complex combinations of "Set-cookie" and "Cache-control"
3255 headers is left to the application. The application can then
3256 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003257 cookie. Since all responses should be monitored, this mode
3258 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003259 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003260 start with this mode for new deployments. This keyword is
3261 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003262
3263 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003264 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003265
Willy Tarreaua79094d2010-08-31 22:54:15 +02003266 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003267 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003268 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003269 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003270 existing configurations running in the "rewrite" mode. The cookie
3271 will only be a session cookie and will not be stored on the
3272 client's disk. By default, unless the "indirect" option is added,
3273 the server will see the cookies emitted by the client. Due to
3274 caching effects, it is generally wise to add the "nocache" or
3275 "postonly" keywords (see below). The "insert" keyword is not
3276 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003277
3278 prefix This keyword indicates that instead of relying on a dedicated
3279 cookie for the persistence, an existing one will be completed.
3280 This may be needed in some specific environments where the client
3281 does not support more than one single cookie and the application
3282 already needs it. In this case, whenever the server sets a cookie
3283 named <name>, it will be prefixed with the server's identifier
3284 and a delimiter. The prefix will be removed from all client
3285 requests so that the server still finds the cookie it emitted.
3286 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003287 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003288 not compatible with "rewrite" and "insert". Note: it is highly
3289 recommended not to use "indirect" with "prefix", otherwise server
3290 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003291
Willy Tarreaua79094d2010-08-31 22:54:15 +02003292 indirect When this option is specified, no cookie will be emitted to a
3293 client which already has a valid one for the server which has
3294 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003295 it will be removed, unless the "preserve" option is also set. In
3296 "insert" mode, this will additionally remove cookies from the
3297 requests transmitted to the server, making the persistence
3298 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003299 Note: it is highly recommended not to use "indirect" with
3300 "prefix", otherwise server cookie updates would not be sent to
3301 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003302
3303 nocache This option is recommended in conjunction with the insert mode
3304 when there is a cache between the client and HAProxy, as it
3305 ensures that a cacheable response will be tagged non-cacheable if
3306 a cookie needs to be inserted. This is important because if all
3307 persistence cookies are added on a cacheable home page for
3308 instance, then all customers will then fetch the page from an
3309 outer cache and will all share the same persistence cookie,
3310 leading to one server receiving much more traffic than others.
3311 See also the "insert" and "postonly" options.
3312
3313 postonly This option ensures that cookie insertion will only be performed
3314 on responses to POST requests. It is an alternative to the
3315 "nocache" option, because POST responses are not cacheable, so
3316 this ensures that the persistence cookie will never get cached.
3317 Since most sites do not need any sort of persistence before the
3318 first POST which generally is a login request, this is a very
3319 efficient method to optimize caching without risking to find a
3320 persistence cookie in the cache.
3321 See also the "insert" and "nocache" options.
3322
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003323 preserve This option may only be used with "insert" and/or "indirect". It
3324 allows the server to emit the persistence cookie itself. In this
3325 case, if a cookie is found in the response, haproxy will leave it
3326 untouched. This is useful in order to end persistence after a
3327 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003328 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003329 the past. By combining this mechanism with the "disable-on-404"
3330 check option, it is possible to perform a completely graceful
3331 shutdown because users will definitely leave the server after
3332 they logout.
3333
Willy Tarreau4992dd22012-05-31 21:02:17 +02003334 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3335 when a cookie is inserted. This attribute is used so that a
3336 user agent doesn't share the cookie with non-HTTP components.
3337 Please check RFC6265 for more information on this attribute.
3338
3339 secure This option tells haproxy to add a "Secure" cookie attribute when
3340 a cookie is inserted. This attribute is used so that a user agent
3341 never emits this cookie over non-secure channels, which means
3342 that a cookie learned with this flag will be presented only over
3343 SSL/TLS connections. Please check RFC6265 for more information on
3344 this attribute.
3345
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003346 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003347 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003348 name. If the domain begins with a dot, the browser is allowed to
3349 use it for any host ending with that name. It is also possible to
3350 specify several domain names by invoking this option multiple
3351 times. Some browsers might have small limits on the number of
3352 domains, so be careful when doing that. For the record, sending
3353 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003354
Willy Tarreau996a92c2010-10-13 19:30:47 +02003355 maxidle This option allows inserted cookies to be ignored after some idle
3356 time. It only works with insert-mode cookies. When a cookie is
3357 sent to the client, the date this cookie was emitted is sent too.
3358 Upon further presentations of this cookie, if the date is older
3359 than the delay indicated by the parameter (in seconds), it will
3360 be ignored. Otherwise, it will be refreshed if needed when the
3361 response is sent to the client. This is particularly useful to
3362 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003363 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003364 this option is set and a cookie has no date, it is always
3365 accepted, but gets refreshed in the response. This maintains the
3366 ability for admins to access their sites. Cookies that have a
3367 date in the future further than 24 hours are ignored. Doing so
3368 lets admins fix timezone issues without risking kicking users off
3369 the site.
3370
3371 maxlife This option allows inserted cookies to be ignored after some life
3372 time, whether they're in use or not. It only works with insert
3373 mode cookies. When a cookie is first sent to the client, the date
3374 this cookie was emitted is sent too. Upon further presentations
3375 of this cookie, if the date is older than the delay indicated by
3376 the parameter (in seconds), it will be ignored. If the cookie in
3377 the request has no date, it is accepted and a date will be set.
3378 Cookies that have a date in the future further than 24 hours are
3379 ignored. Doing so lets admins fix timezone issues without risking
3380 kicking users off the site. Contrary to maxidle, this value is
3381 not refreshed, only the first visit date counts. Both maxidle and
3382 maxlife may be used at the time. This is particularly useful to
3383 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003384 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003385 is stronger than the maxidle method in that it forces a
3386 redispatch after some absolute delay.
3387
Olivier Houchard4e694042017-03-14 20:01:29 +01003388 dynamic Activate dynamic cookies. When used, a session cookie is
3389 dynamically created for each server, based on the IP and port
3390 of the server, and a secret key, specified in the
3391 "dynamic-cookie-key" backend directive.
3392 The cookie will be regenerated each time the IP address change,
3393 and is only generated for IPv4/IPv6.
3394
Willy Tarreau0ba27502007-12-24 16:55:16 +01003395 There can be only one persistence cookie per HTTP backend, and it can be
3396 declared in a defaults section. The value of the cookie will be the value
3397 indicated after the "cookie" keyword in a "server" statement. If no cookie
3398 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003399
Willy Tarreau0ba27502007-12-24 16:55:16 +01003400 Examples :
3401 cookie JSESSIONID prefix
3402 cookie SRV insert indirect nocache
3403 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003404 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003405
Willy Tarreau294d0f02015-08-10 19:40:12 +02003406 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003407
Willy Tarreau983e01e2010-01-11 18:42:06 +01003408
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003409declare capture [ request | response ] len <length>
3410 Declares a capture slot.
3411 May be used in sections : defaults | frontend | listen | backend
3412 no | yes | yes | no
3413 Arguments:
3414 <length> is the length allowed for the capture.
3415
3416 This declaration is only available in the frontend or listen section, but the
3417 reserved slot can be used in the backends. The "request" keyword allocates a
3418 capture slot for use in the request, and "response" allocates a capture slot
3419 for use in the response.
3420
3421 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003422 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003423 "http-request capture" and "http-response capture".
3424
3425
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003426default-server [param*]
3427 Change default options for a server in a backend
3428 May be used in sections : defaults | frontend | listen | backend
3429 yes | no | yes | yes
3430 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003431 <param*> is a list of parameters for this server. The "default-server"
3432 keyword accepts an important number of options and has a complete
3433 section dedicated to it. Please refer to section 5 for more
3434 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003435
Willy Tarreau983e01e2010-01-11 18:42:06 +01003436 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003437 default-server inter 1000 weight 13
3438
3439 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003440
Willy Tarreau983e01e2010-01-11 18:42:06 +01003441
Willy Tarreau0ba27502007-12-24 16:55:16 +01003442default_backend <backend>
3443 Specify the backend to use when no "use_backend" rule has been matched.
3444 May be used in sections : defaults | frontend | listen | backend
3445 yes | yes | yes | no
3446 Arguments :
3447 <backend> is the name of the backend to use.
3448
3449 When doing content-switching between frontend and backends using the
3450 "use_backend" keyword, it is often useful to indicate which backend will be
3451 used when no rule has matched. It generally is the dynamic backend which
3452 will catch all undetermined requests.
3453
Willy Tarreau0ba27502007-12-24 16:55:16 +01003454 Example :
3455
3456 use_backend dynamic if url_dyn
3457 use_backend static if url_css url_img extension_img
3458 default_backend dynamic
3459
Willy Tarreau98d04852015-05-26 12:18:29 +02003460 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003461
Willy Tarreau0ba27502007-12-24 16:55:16 +01003462
Baptiste Assmann27f51342013-10-09 06:51:49 +02003463description <string>
3464 Describe a listen, frontend or backend.
3465 May be used in sections : defaults | frontend | listen | backend
3466 no | yes | yes | yes
3467 Arguments : string
3468
3469 Allows to add a sentence to describe the related object in the HAProxy HTML
3470 stats page. The description will be printed on the right of the object name
3471 it describes.
3472 No need to backslash spaces in the <string> arguments.
3473
3474
Willy Tarreau0ba27502007-12-24 16:55:16 +01003475disabled
3476 Disable a proxy, frontend or backend.
3477 May be used in sections : defaults | frontend | listen | backend
3478 yes | yes | yes | yes
3479 Arguments : none
3480
3481 The "disabled" keyword is used to disable an instance, mainly in order to
3482 liberate a listening port or to temporarily disable a service. The instance
3483 will still be created and its configuration will be checked, but it will be
3484 created in the "stopped" state and will appear as such in the statistics. It
3485 will not receive any traffic nor will it send any health-checks or logs. It
3486 is possible to disable many instances at once by adding the "disabled"
3487 keyword in a "defaults" section.
3488
3489 See also : "enabled"
3490
3491
Willy Tarreau5ce94572010-06-07 14:35:41 +02003492dispatch <address>:<port>
3493 Set a default server address
3494 May be used in sections : defaults | frontend | listen | backend
3495 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003496 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003497
3498 <address> is the IPv4 address of the default server. Alternatively, a
3499 resolvable hostname is supported, but this name will be resolved
3500 during start-up.
3501
3502 <ports> is a mandatory port specification. All connections will be sent
3503 to this port, and it is not permitted to use port offsets as is
3504 possible with normal servers.
3505
Willy Tarreau787aed52011-04-15 06:45:37 +02003506 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003507 server can take the connection. In the past it was used to forward non
3508 persistent connections to an auxiliary load balancer. Due to its simple
3509 syntax, it has also been used for simple TCP relays. It is recommended not to
3510 use it for more clarity, and to use the "server" directive instead.
3511
3512 See also : "server"
3513
Olivier Houchard4e694042017-03-14 20:01:29 +01003514
3515dynamic-cookie-key <string>
3516 Set the dynamic cookie secret key for a backend.
3517 May be used in sections : defaults | frontend | listen | backend
3518 yes | no | yes | yes
3519 Arguments : The secret key to be used.
3520
3521 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003522 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003523 specified on the "server" line), using a hash of the IP address of the
3524 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003525 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003526 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003527
Willy Tarreau0ba27502007-12-24 16:55:16 +01003528enabled
3529 Enable a proxy, frontend or backend.
3530 May be used in sections : defaults | frontend | listen | backend
3531 yes | yes | yes | yes
3532 Arguments : none
3533
3534 The "enabled" keyword is used to explicitly enable an instance, when the
3535 defaults has been set to "disabled". This is very rarely used.
3536
3537 See also : "disabled"
3538
3539
3540errorfile <code> <file>
3541 Return a file contents instead of errors generated by HAProxy
3542 May be used in sections : defaults | frontend | listen | backend
3543 yes | yes | yes | yes
3544 Arguments :
3545 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003546 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3547 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003548
3549 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003550 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003551 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003552 error pages, and to use absolute paths, since files are read
3553 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003554
3555 It is important to understand that this keyword is not meant to rewrite
3556 errors returned by the server, but errors detected and returned by HAProxy.
3557 This is why the list of supported errors is limited to a small set.
3558
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003559 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3560
Willy Tarreau0ba27502007-12-24 16:55:16 +01003561 The files are returned verbatim on the TCP socket. This allows any trick such
3562 as redirections to another URL or site, as well as tricks to clean cookies,
3563 force enable or disable caching, etc... The package provides default error
3564 files returning the same contents as default errors.
3565
Willy Tarreau59140a22009-02-22 12:02:30 +01003566 The files should not exceed the configured buffer size (BUFSIZE), which
3567 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003568 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003569 loops between the client and HAProxy when all servers are down, causing an
3570 error to be returned instead of an image. For better HTTP compliance, it is
3571 recommended that all header lines end with CR-LF and not LF alone.
3572
Willy Tarreau0ba27502007-12-24 16:55:16 +01003573 The files are read at the same time as the configuration and kept in memory.
3574 For this reason, the errors continue to be returned even when the process is
3575 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003576 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003577 403 status code and interrogating a blocked URL.
3578
3579 See also : "errorloc", "errorloc302", "errorloc303"
3580
Willy Tarreau59140a22009-02-22 12:02:30 +01003581 Example :
3582 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003583 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003584 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3585 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3586
Willy Tarreau2769aa02007-12-27 18:26:09 +01003587
3588errorloc <code> <url>
3589errorloc302 <code> <url>
3590 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3591 May be used in sections : defaults | frontend | listen | backend
3592 yes | yes | yes | yes
3593 Arguments :
3594 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003595 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3596 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003597
3598 <url> it is the exact contents of the "Location" header. It may contain
3599 either a relative URI to an error page hosted on the same site,
3600 or an absolute URI designating an error page on another site.
3601 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003602 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003603
3604 It is important to understand that this keyword is not meant to rewrite
3605 errors returned by the server, but errors detected and returned by HAProxy.
3606 This is why the list of supported errors is limited to a small set.
3607
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003608 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3609
Willy Tarreau2769aa02007-12-27 18:26:09 +01003610 Note that both keyword return the HTTP 302 status code, which tells the
3611 client to fetch the designated URL using the same HTTP method. This can be
3612 quite problematic in case of non-GET methods such as POST, because the URL
3613 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003614 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003615 status code, indicating to the client that the URL must be fetched with a GET
3616 request.
3617
3618 See also : "errorfile", "errorloc303"
3619
3620
3621errorloc303 <code> <url>
3622 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3623 May be used in sections : defaults | frontend | listen | backend
3624 yes | yes | yes | yes
3625 Arguments :
3626 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003627 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3628 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003629
3630 <url> it is the exact contents of the "Location" header. It may contain
3631 either a relative URI to an error page hosted on the same site,
3632 or an absolute URI designating an error page on another site.
3633 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003634 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003635
3636 It is important to understand that this keyword is not meant to rewrite
3637 errors returned by the server, but errors detected and returned by HAProxy.
3638 This is why the list of supported errors is limited to a small set.
3639
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003640 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3641
Willy Tarreau2769aa02007-12-27 18:26:09 +01003642 Note that both keyword return the HTTP 303 status code, which tells the
3643 client to fetch the designated URL using the same HTTP GET method. This
3644 solves the usual problems associated with "errorloc" and the 302 code. It is
3645 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003646 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003647
3648 See also : "errorfile", "errorloc", "errorloc302"
3649
3650
Simon Horman51a1cf62015-02-03 13:00:44 +09003651email-alert from <emailaddr>
3652 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003653 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003654 May be used in sections: defaults | frontend | listen | backend
3655 yes | yes | yes | yes
3656
3657 Arguments :
3658
3659 <emailaddr> is the from email address to use when sending email alerts
3660
3661 Also requires "email-alert mailers" and "email-alert to" to be set
3662 and if so sending email alerts is enabled for the proxy.
3663
Simon Horman64e34162015-02-06 11:11:57 +09003664 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003665 "email-alert myhostname", "email-alert to", section 3.6 about
3666 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003667
3668
3669email-alert level <level>
3670 Declare the maximum log level of messages for which email alerts will be
3671 sent. This acts as a filter on the sending of email alerts.
3672 May be used in sections: defaults | frontend | listen | backend
3673 yes | yes | yes | yes
3674
3675 Arguments :
3676
3677 <level> One of the 8 syslog levels:
3678 emerg alert crit err warning notice info debug
3679 The above syslog levels are ordered from lowest to highest.
3680
3681 By default level is alert
3682
3683 Also requires "email-alert from", "email-alert mailers" and
3684 "email-alert to" to be set and if so sending email alerts is enabled
3685 for the proxy.
3686
Simon Horman1421e212015-04-30 13:10:35 +09003687 Alerts are sent when :
3688
3689 * An un-paused server is marked as down and <level> is alert or lower
3690 * A paused server is marked as down and <level> is notice or lower
3691 * A server is marked as up or enters the drain state and <level>
3692 is notice or lower
3693 * "option log-health-checks" is enabled, <level> is info or lower,
3694 and a health check status update occurs
3695
Simon Horman64e34162015-02-06 11:11:57 +09003696 See also : "email-alert from", "email-alert mailers",
3697 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003698 section 3.6 about mailers.
3699
3700
3701email-alert mailers <mailersect>
3702 Declare the mailers to be used when sending email alerts
3703 May be used in sections: defaults | frontend | listen | backend
3704 yes | yes | yes | yes
3705
3706 Arguments :
3707
3708 <mailersect> is the name of the mailers section to send email alerts.
3709
3710 Also requires "email-alert from" and "email-alert to" to be set
3711 and if so sending email alerts is enabled for the proxy.
3712
Simon Horman64e34162015-02-06 11:11:57 +09003713 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3714 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003715
3716
3717email-alert myhostname <hostname>
3718 Declare the to hostname address to be used when communicating with
3719 mailers.
3720 May be used in sections: defaults | frontend | listen | backend
3721 yes | yes | yes | yes
3722
3723 Arguments :
3724
Baptiste Assmann738bad92015-12-21 15:27:53 +01003725 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003726
3727 By default the systems hostname is used.
3728
3729 Also requires "email-alert from", "email-alert mailers" and
3730 "email-alert to" to be set and if so sending email alerts is enabled
3731 for the proxy.
3732
Simon Horman64e34162015-02-06 11:11:57 +09003733 See also : "email-alert from", "email-alert level", "email-alert mailers",
3734 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003735
3736
3737email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003738 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003739 header of email alerts. This is the address that email alerts are sent to.
3740 May be used in sections: defaults | frontend | listen | backend
3741 yes | yes | yes | yes
3742
3743 Arguments :
3744
3745 <emailaddr> is the to email address to use when sending email alerts
3746
3747 Also requires "email-alert mailers" and "email-alert to" to be set
3748 and if so sending email alerts is enabled for the proxy.
3749
Simon Horman64e34162015-02-06 11:11:57 +09003750 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003751 "email-alert myhostname", section 3.6 about mailers.
3752
3753
Willy Tarreau4de91492010-01-22 19:10:05 +01003754force-persist { if | unless } <condition>
3755 Declare a condition to force persistence on down servers
3756 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003757 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003758
3759 By default, requests are not dispatched to down servers. It is possible to
3760 force this using "option persist", but it is unconditional and redispatches
3761 to a valid server if "option redispatch" is set. That leaves with very little
3762 possibilities to force some requests to reach a server which is artificially
3763 marked down for maintenance operations.
3764
3765 The "force-persist" statement allows one to declare various ACL-based
3766 conditions which, when met, will cause a request to ignore the down status of
3767 a server and still try to connect to it. That makes it possible to start a
3768 server, still replying an error to the health checks, and run a specially
3769 configured browser to test the service. Among the handy methods, one could
3770 use a specific source IP address, or a specific cookie. The cookie also has
3771 the advantage that it can easily be added/removed on the browser from a test
3772 page. Once the service is validated, it is then possible to open the service
3773 to the world by returning a valid response to health checks.
3774
3775 The forced persistence is enabled when an "if" condition is met, or unless an
3776 "unless" condition is met. The final redispatch is always disabled when this
3777 is used.
3778
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003779 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003780 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003781
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003782
3783filter <name> [param*]
3784 Add the filter <name> in the filter list attached to the proxy.
3785 May be used in sections : defaults | frontend | listen | backend
3786 no | yes | yes | yes
3787 Arguments :
3788 <name> is the name of the filter. Officially supported filters are
3789 referenced in section 9.
3790
Tim Düsterhus4896c442016-11-29 02:15:19 +01003791 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003792 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003793 filter. Please refer to the documentation of the corresponding
3794 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003795
3796 Multiple occurrences of the filter line can be used for the same proxy. The
3797 same filter can be referenced many times if needed.
3798
3799 Example:
3800 listen
3801 bind *:80
3802
3803 filter trace name BEFORE-HTTP-COMP
3804 filter compression
3805 filter trace name AFTER-HTTP-COMP
3806
3807 compression algo gzip
3808 compression offload
3809
3810 server srv1 192.168.0.1:80
3811
3812 See also : section 9.
3813
Willy Tarreau4de91492010-01-22 19:10:05 +01003814
Willy Tarreau2769aa02007-12-27 18:26:09 +01003815fullconn <conns>
3816 Specify at what backend load the servers will reach their maxconn
3817 May be used in sections : defaults | frontend | listen | backend
3818 yes | no | yes | yes
3819 Arguments :
3820 <conns> is the number of connections on the backend which will make the
3821 servers use the maximal number of connections.
3822
Willy Tarreau198a7442008-01-17 12:05:32 +01003823 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003824 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003825 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003826 load. The server will then always accept at least <minconn> connections,
3827 never more than <maxconn>, and the limit will be on the ramp between both
3828 values when the backend has less than <conns> concurrent connections. This
3829 makes it possible to limit the load on the servers during normal loads, but
3830 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003831 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003832
Willy Tarreaufbb78422011-06-05 15:38:35 +02003833 Since it's hard to get this value right, haproxy automatically sets it to
3834 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003835 backend (based on "use_backend" and "default_backend" rules). That way it's
3836 safe to leave it unset. However, "use_backend" involving dynamic names are
3837 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003838
Willy Tarreau2769aa02007-12-27 18:26:09 +01003839 Example :
3840 # The servers will accept between 100 and 1000 concurrent connections each
3841 # and the maximum of 1000 will be reached when the backend reaches 10000
3842 # connections.
3843 backend dynamic
3844 fullconn 10000
3845 server srv1 dyn1:80 minconn 100 maxconn 1000
3846 server srv2 dyn2:80 minconn 100 maxconn 1000
3847
3848 See also : "maxconn", "server"
3849
3850
3851grace <time>
3852 Maintain a proxy operational for some time after a soft stop
3853 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003854 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003855 Arguments :
3856 <time> is the time (by default in milliseconds) for which the instance
3857 will remain operational with the frontend sockets still listening
3858 when a soft-stop is received via the SIGUSR1 signal.
3859
3860 This may be used to ensure that the services disappear in a certain order.
3861 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003862 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003863 needed by the equipment to detect the failure.
3864
3865 Note that currently, there is very little benefit in using this parameter,
3866 and it may in fact complicate the soft-reconfiguration process more than
3867 simplify it.
3868
Willy Tarreau0ba27502007-12-24 16:55:16 +01003869
Andrew Rodland17be45e2016-10-25 17:04:12 -04003870hash-balance-factor <factor>
3871 Specify the balancing factor for bounded-load consistent hashing
3872 May be used in sections : defaults | frontend | listen | backend
3873 yes | no | no | yes
3874 Arguments :
3875 <factor> is the control for the maximum number of concurrent requests to
3876 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01003877 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04003878
3879 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3880 enables an algorithm that prevents any one server from getting too many
3881 requests at once, even if some hash buckets receive many more requests than
3882 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3883 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3884 then no server will be allowed to have a load more than 1.5 times the average.
3885 If server weights are used, they will be respected.
3886
3887 If the first-choice server is disqualified, the algorithm will choose another
3888 server based on the request hash, until a server with additional capacity is
3889 found. A higher <factor> allows more imbalance between the servers, while a
3890 lower <factor> means that more servers will be checked on average, affecting
3891 performance. Reasonable values are from 125 to 200.
3892
Willy Tarreau760e81d2018-05-03 07:20:40 +02003893 This setting is also used by "balance random" which internally relies on the
3894 consistent hashing mechanism.
3895
Andrew Rodland17be45e2016-10-25 17:04:12 -04003896 See also : "balance" and "hash-type".
3897
3898
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003899hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003900 Specify a method to use for mapping hashes to servers
3901 May be used in sections : defaults | frontend | listen | backend
3902 yes | no | yes | yes
3903 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003904 <method> is the method used to select a server from the hash computed by
3905 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003906
Bhaskar98634f02013-10-29 23:30:51 -04003907 map-based the hash table is a static array containing all alive servers.
3908 The hashes will be very smooth, will consider weights, but
3909 will be static in that weight changes while a server is up
3910 will be ignored. This means that there will be no slow start.
3911 Also, since a server is selected by its position in the array,
3912 most mappings are changed when the server count changes. This
3913 means that when a server goes up or down, or when a server is
3914 added to a farm, most connections will be redistributed to
3915 different servers. This can be inconvenient with caches for
3916 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003917
Bhaskar98634f02013-10-29 23:30:51 -04003918 consistent the hash table is a tree filled with many occurrences of each
3919 server. The hash key is looked up in the tree and the closest
3920 server is chosen. This hash is dynamic, it supports changing
3921 weights while the servers are up, so it is compatible with the
3922 slow start feature. It has the advantage that when a server
3923 goes up or down, only its associations are moved. When a
3924 server is added to the farm, only a few part of the mappings
3925 are redistributed, making it an ideal method for caches.
3926 However, due to its principle, the distribution will never be
3927 very smooth and it may sometimes be necessary to adjust a
3928 server's weight or its ID to get a more balanced distribution.
3929 In order to get the same distribution on multiple load
3930 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003931 same IDs. Note: consistent hash uses sdbm and avalanche if no
3932 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003933
3934 <function> is the hash function to be used :
3935
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003936 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003937 reimplementation of ndbm) database library. It was found to do
3938 well in scrambling bits, causing better distribution of the keys
3939 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003940 function with good distribution, unless the total server weight
3941 is a multiple of 64, in which case applying the avalanche
3942 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003943
3944 djb2 this function was first proposed by Dan Bernstein many years ago
3945 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003946 function provides a better distribution than sdbm. It generally
3947 works well with text-based inputs though it can perform extremely
3948 poorly with numeric-only input or when the total server weight is
3949 a multiple of 33, unless the avalanche modifier is also used.
3950
Willy Tarreaua0f42712013-11-14 14:30:35 +01003951 wt6 this function was designed for haproxy while testing other
3952 functions in the past. It is not as smooth as the other ones, but
3953 is much less sensible to the input data set or to the number of
3954 servers. It can make sense as an alternative to sdbm+avalanche or
3955 djb2+avalanche for consistent hashing or when hashing on numeric
3956 data such as a source IP address or a visitor identifier in a URL
3957 parameter.
3958
Willy Tarreau324f07f2015-01-20 19:44:50 +01003959 crc32 this is the most common CRC32 implementation as used in Ethernet,
3960 gzip, PNG, etc. It is slower than the other ones but may provide
3961 a better distribution or less predictable results especially when
3962 used on strings.
3963
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003964 <modifier> indicates an optional method applied after hashing the key :
3965
3966 avalanche This directive indicates that the result from the hash
3967 function above should not be used in its raw form but that
3968 a 4-byte full avalanche hash must be applied first. The
3969 purpose of this step is to mix the resulting bits from the
3970 previous hash in order to avoid any undesired effect when
3971 the input contains some limited values or when the number of
3972 servers is a multiple of one of the hash's components (64
3973 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3974 result less predictable, but it's also not as smooth as when
3975 using the original function. Some testing might be needed
3976 with some workloads. This hash is one of the many proposed
3977 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003978
Bhaskar98634f02013-10-29 23:30:51 -04003979 The default hash type is "map-based" and is recommended for most usages. The
3980 default function is "sdbm", the selection of a function should be based on
3981 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003982
Andrew Rodland17be45e2016-10-25 17:04:12 -04003983 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003984
3985
Willy Tarreau0ba27502007-12-24 16:55:16 +01003986http-check disable-on-404
3987 Enable a maintenance mode upon HTTP/404 response to health-checks
3988 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003989 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003990 Arguments : none
3991
3992 When this option is set, a server which returns an HTTP code 404 will be
3993 excluded from further load-balancing, but will still receive persistent
3994 connections. This provides a very convenient method for Web administrators
3995 to perform a graceful shutdown of their servers. It is also important to note
3996 that a server which is detected as failed while it was in this mode will not
3997 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3998 will immediately be reinserted into the farm. The status on the stats page
3999 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01004000 option only works in conjunction with the "httpchk" option. If this option
4001 is used with "http-check expect", then it has precedence over it so that 404
4002 responses will still be considered as soft-stop.
4003
4004 See also : "option httpchk", "http-check expect"
4005
4006
4007http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004008 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01004009 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02004010 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01004011 Arguments :
4012 <match> is a keyword indicating how to look for a specific pattern in the
4013 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004014 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01004015 exclamation mark ("!") to negate the match. Spaces are allowed
4016 between the exclamation mark and the keyword. See below for more
4017 details on the supported keywords.
4018
4019 <pattern> is the pattern to look for. It may be a string or a regular
4020 expression. If the pattern contains spaces, they must be escaped
4021 with the usual backslash ('\').
4022
4023 By default, "option httpchk" considers that response statuses 2xx and 3xx
4024 are valid, and that others are invalid. When "http-check expect" is used,
4025 it defines what is considered valid or invalid. Only one "http-check"
4026 statement is supported in a backend. If a server fails to respond or times
4027 out, the check obviously fails. The available matches are :
4028
4029 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004030 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004031 response's status code is exactly this string. If the
4032 "status" keyword is prefixed with "!", then the response
4033 will be considered invalid if the status code matches.
4034
4035 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004036 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004037 response's status code matches the expression. If the
4038 "rstatus" keyword is prefixed with "!", then the response
4039 will be considered invalid if the status code matches.
4040 This is mostly used to check for multiple codes.
4041
4042 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004043 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004044 response's body contains this exact string. If the
4045 "string" keyword is prefixed with "!", then the response
4046 will be considered invalid if the body contains this
4047 string. This can be used to look for a mandatory word at
4048 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004049 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01004050 trace).
4051
4052 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004053 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004054 response's body matches this expression. If the "rstring"
4055 keyword is prefixed with "!", then the response will be
4056 considered invalid if the body matches the expression.
4057 This can be used to look for a mandatory word at the end
4058 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01004059 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01004060
4061 It is important to note that the responses will be limited to a certain size
4062 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
4063 Thus, too large responses may not contain the mandatory pattern when using
4064 "string" or "rstring". If a large response is absolutely required, it is
4065 possible to change the default max size by setting the global variable.
4066 However, it is worth keeping in mind that parsing very large responses can
4067 waste some CPU cycles, especially when regular expressions are used, and that
4068 it is always better to focus the checks on smaller resources.
4069
Cyril Bonté32602d22015-01-30 00:07:07 +01004070 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
4071 will automatically append a "Connection: close" header, meaning that this
4072 header should not be present in the request provided by "option httpchk".
4073
Willy Tarreaubd741542010-03-16 18:46:54 +01004074 Last, if "http-check expect" is combined with "http-check disable-on-404",
4075 then this last one has precedence when the server responds with 404.
4076
4077 Examples :
4078 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004079 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01004080
4081 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004082 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01004083
4084 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004085 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01004086
4087 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004088 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01004089
Willy Tarreaubd741542010-03-16 18:46:54 +01004090 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004091
4092
Willy Tarreauef781042010-01-27 11:53:01 +01004093http-check send-state
4094 Enable emission of a state header with HTTP health checks
4095 May be used in sections : defaults | frontend | listen | backend
4096 yes | no | yes | yes
4097 Arguments : none
4098
4099 When this option is set, haproxy will systematically send a special header
4100 "X-Haproxy-Server-State" with a list of parameters indicating to each server
4101 how they are seen by haproxy. This can be used for instance when a server is
4102 manipulated without access to haproxy and the operator needs to know whether
4103 haproxy still sees it up or not, or if the server is the last one in a farm.
4104
4105 The header is composed of fields delimited by semi-colons, the first of which
4106 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
4107 checks on the total number before transition, just as appears in the stats
4108 interface. Next headers are in the form "<variable>=<value>", indicating in
4109 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08004110 - a variable "address", containing the address of the backend server.
4111 This corresponds to the <address> field in the server declaration. For
4112 unix domain sockets, it will read "unix".
4113
4114 - a variable "port", containing the port of the backend server. This
4115 corresponds to the <port> field in the server declaration. For unix
4116 domain sockets, it will read "unix".
4117
Willy Tarreauef781042010-01-27 11:53:01 +01004118 - a variable "name", containing the name of the backend followed by a slash
4119 ("/") then the name of the server. This can be used when a server is
4120 checked in multiple backends.
4121
4122 - a variable "node" containing the name of the haproxy node, as set in the
4123 global "node" variable, otherwise the system's hostname if unspecified.
4124
4125 - a variable "weight" indicating the weight of the server, a slash ("/")
4126 and the total weight of the farm (just counting usable servers). This
4127 helps to know if other servers are available to handle the load when this
4128 one fails.
4129
4130 - a variable "scur" indicating the current number of concurrent connections
4131 on the server, followed by a slash ("/") then the total number of
4132 connections on all servers of the same backend.
4133
4134 - a variable "qcur" indicating the current number of requests in the
4135 server's queue.
4136
4137 Example of a header received by the application server :
4138 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4139 scur=13/22; qcur=0
4140
4141 See also : "option httpchk", "http-check disable-on-404"
4142
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004143
4144http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004145 Access control for Layer 7 requests
4146
4147 May be used in sections: defaults | frontend | listen | backend
4148 no | yes | yes | yes
4149
Willy Tarreau20b0de52012-12-24 15:45:22 +01004150 The http-request statement defines a set of rules which apply to layer 7
4151 processing. The rules are evaluated in their declaration order when they are
4152 met in a frontend, listen or backend section. Any rule may optionally be
4153 followed by an ACL-based condition, in which case it will only be evaluated
4154 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004155
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004156 The first keyword is the rule's action. The supported actions are described
4157 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004158
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004159 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004160
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004161 Example:
4162 acl nagios src 192.168.129.3
4163 acl local_net src 192.168.0.0/16
4164 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004165
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004166 http-request allow if nagios
4167 http-request allow if local_net auth_ok
4168 http-request auth realm Gimme if local_net auth_ok
4169 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004170
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004171 Example:
4172 acl key req.hdr(X-Add-Acl-Key) -m found
4173 acl add path /addacl
4174 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004175
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004176 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004177
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004178 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4179 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004180
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004181 Example:
4182 acl value req.hdr(X-Value) -m found
4183 acl setmap path /setmap
4184 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004185
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004186 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004187
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004188 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4189 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004190
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004191 See also : "stats http-request", section 3.4 about userlists and section 7
4192 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004193
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004194http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004195
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004196 This is used to add a new entry into an ACL. The ACL must be loaded from a
4197 file (even a dummy empty file). The file name of the ACL to be updated is
4198 passed between parentheses. It takes one argument: <key fmt>, which follows
4199 log-format rules, to collect content of the new entry. It performs a lookup
4200 in the ACL before insertion, to avoid duplicated (or more) values. This
4201 lookup is done by a linear search and can be expensive with large lists!
4202 It is the equivalent of the "add acl" command from the stats socket, but can
4203 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004204
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004205http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004206
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004207 This appends an HTTP header field whose name is specified in <name> and
4208 whose value is defined by <fmt> which follows the log-format rules (see
4209 Custom Log Format in section 8.2.4). This is particularly useful to pass
4210 connection-specific information to the server (e.g. the client's SSL
4211 certificate), or to combine several headers into one. This rule is not
4212 final, so it is possible to add other similar rules. Note that header
4213 addition is performed immediately, so one rule might reuse the resulting
4214 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004215
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004216http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004217
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004218 This stops the evaluation of the rules and lets the request pass the check.
4219 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004220
Sasha Pachev218f0642014-06-16 12:05:59 -06004221
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004222http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004223
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004224 This stops the evaluation of the rules and immediately responds with an
4225 HTTP 401 or 407 error code to invite the user to present a valid user name
4226 and password. No further "http-request" rules are evaluated. An optional
4227 "realm" parameter is supported, it sets the authentication realm that is
4228 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004229
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004230 Example:
4231 acl auth_ok http_auth_group(L1) G1
4232 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004233
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004234http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004235
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004236 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004237
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004238http-request capture <sample> [ len <length> | id <id> ]
4239 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004240
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004241 This captures sample expression <sample> from the request buffer, and
4242 converts it to a string of at most <len> characters. The resulting string is
4243 stored into the next request "capture" slot, so it will possibly appear next
4244 to some captured HTTP headers. It will then automatically appear in the logs,
4245 and it will be possible to extract it using sample fetch rules to feed it
4246 into headers or anything. The length should be limited given that this size
4247 will be allocated for each capture during the whole session life.
4248 Please check section 7.3 (Fetching samples) and "capture request header" for
4249 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004250
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004251 If the keyword "id" is used instead of "len", the action tries to store the
4252 captured string in a previously declared capture slot. This is useful to run
4253 captures in backends. The slot id can be declared by a previous directive
4254 "http-request capture" or with the "declare capture" keyword. If the slot
4255 <id> doesn't exist, then HAProxy fails parsing the configuration to prevent
4256 unexpected behavior at run time.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004257
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004258http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004259
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004260 This is used to delete an entry from an ACL. The ACL must be loaded from a
4261 file (even a dummy empty file). The file name of the ACL to be updated is
4262 passed between parentheses. It takes one argument: <key fmt>, which follows
4263 log-format rules, to collect content of the entry to delete.
4264 It is the equivalent of the "del acl" command from the stats socket, but can
4265 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004266
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004267http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004268
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004269 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004270
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004271http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004272
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004273 This is used to delete an entry from a MAP. The MAP must be loaded from a
4274 file (even a dummy empty file). The file name of the MAP to be updated is
4275 passed between parentheses. It takes one argument: <key fmt>, which follows
4276 log-format rules, to collect content of the entry to delete.
4277 It takes one argument: "file name" It is the equivalent of the "del map"
4278 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004279
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004280http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004281
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004282 This stops the evaluation of the rules and immediately rejects the request
4283 and emits an HTTP 403 error, or optionally the status code specified as an
4284 argument to "deny_status". The list of permitted status codes is limited to
4285 those that can be overridden by the "errorfile" directive.
4286 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004287
Olivier Houchard602bf7d2019-05-10 13:59:15 +02004288http-request disable-l7-retry [ { if | unless } <condition> ]
4289 This disables any attempt to retry the request if it fails for any other
4290 reason than a connection failure. This can be useful for example to make
4291 sure POST requests aren't retried on failure.
4292
Baptiste Assmann333939c2019-01-21 08:34:50 +01004293http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
4294
4295 This action performs a DNS resolution of the output of <expr> and stores
4296 the result in the variable <var>. It uses the DNS resolvers section
4297 pointed by <resolvers>.
4298 It is possible to choose a resolution preference using the optional
4299 arguments 'ipv4' or 'ipv6'.
4300 When performing the DNS resolution, the client side connection is on
4301 pause waiting till the end of the resolution.
4302 If an IP address can be found, it is stored into <var>. If any kind of
4303 error occurs, then <var> is not set.
4304 One can use this action to discover a server IP address at run time and
4305 based on information found in the request (IE a Host header).
4306 If this action is used to find the server's IP address (using the
4307 "set-dst" action), then the server IP address in the backend must be set
4308 to 0.0.0.0.
4309
4310 Example:
4311 resolvers mydns
4312 nameserver local 127.0.0.53:53
4313 nameserver google 8.8.8.8:53
4314 timeout retry 1s
4315 hold valid 10s
4316 hold nx 3s
4317 hold other 3s
4318 hold obsolete 0s
4319 accepted_payload_size 8192
4320
4321 frontend fe
4322 bind 10.42.0.1:80
4323 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
4324 http-request capture var(txn.myip) len 40
4325
4326 # return 503 when the variable is not set,
4327 # which mean DNS resolution error
4328 use_backend b_503 unless { var(txn.myip) -m found }
4329
4330 default_backend be
4331
4332 backend b_503
4333 # dummy backend used to return 503.
4334 # one can use the errorfile directive to send a nice
4335 # 503 error page to end users
4336
4337 backend be
4338 # rule to prevent HAProxy from reconnecting to services
4339 # on the local network (forged DNS name used to scan the network)
4340 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
4341 http-request set-dst var(txn.myip)
4342 server clear 0.0.0.0:0
4343
4344 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
4345 be used to scan the network or worst won't loop over itself...
4346
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004347http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4348
4349 This is used to build an HTTP 103 Early Hints response prior to any other one.
4350 This appends an HTTP header field to this response whose name is specified in
4351 <name> and whose value is defined by <fmt> which follows the log-format rules
4352 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004353 to the client some Link headers to preload resources required to render the
4354 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004355
4356 See RFC 8297 for more information.
4357
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004358http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004359
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004360 This performs an HTTP redirection based on a redirect rule. This is exactly
4361 the same as the "redirect" statement except that it inserts a redirect rule
4362 which can be processed in the middle of other "http-request" rules and that
4363 these rules use the "log-format" strings. See the "redirect" keyword for the
4364 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004365
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004366http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004367
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004368 This stops the evaluation of the rules and immediately closes the connection
4369 without sending any response. It acts similarly to the
4370 "tcp-request content reject" rules. It can be useful to force an immediate
4371 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004372
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004373http-request replace-header <name> <match-regex> <replace-fmt>
4374 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004375
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004376 This matches the regular expression in all occurrences of header field
4377 <name> according to <match-regex>, and replaces them with the
4378 <replace-fmt> argument. Format characters are allowed in replace-fmt and
4379 work like in <fmt> arguments in "http-request add-header". The match is
4380 only case-sensitive. It is important to understand that this action only
4381 considers whole header lines, regardless of the number of values they may
4382 contain. This usage is suited to headers naturally containing commas in
4383 their value, such as If-Modified-Since and so on.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004384
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004385 Example:
4386 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
William Lallemand86d0df02017-11-24 21:36:45 +01004387
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004388 # applied to:
4389 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004390
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004391 # outputs:
4392 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004393
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004394 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004395
Willy Tarreau33810222019-06-12 17:44:02 +02004396http-request replace-uri <match-regex> <replace-fmt>
4397 [ { if | unless } <condition> ]
4398
4399 This matches the regular expression in the URI part of the request
4400 according to <match-regex>, and replaces it with the <replace-fmt>
4401 argument. Standard back-references using the backslash ('\') followed by a
4402 number are supported. The <fmt> field is interpreted as a log-format string
4403 so it may contain special expressions just like the <fmt> argument passed
4404 to "http-request set-uri". The match is exclusively case-sensitive. Any
4405 optional scheme, authority or query string are considered in the matching
4406 part of the URI. It is worth noting that regular expressions may be more
4407 expensive to evaluate than certain ACLs, so rare replacements may benefit
4408 from a condition to avoid performing the evaluation at all if it does not
4409 match.
4410
4411 Example:
4412 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4413 http-request replace-uri (.*) /foo\1
4414
4415 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
4416 http-request replace-uri ([^?]*)(\?(.*))? \1/foo\2
4417
4418 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
4419 http-request replace-uri /foo/(.*) /\1
4420 # or more efficient if only some requests match :
4421 http-request replace-uri /foo/(.*) /\1 if { url_beg /foo/ }
4422
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004423http-request replace-value <name> <match-regex> <replace-fmt>
4424 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004425
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004426 This works like "replace-header" except that it matches the regex against
4427 every comma-delimited value of the header field <name> instead of the
4428 entire header. This is suited for all headers which are allowed to carry
4429 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004430
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004431 Example:
4432 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004433
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004434 # applied to:
4435 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004436
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004437 # outputs:
4438 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004439
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004440http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4441http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004442
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004443 This actions increments the GPC0 or GPC1 counter according with the sticky
4444 counter designated by <sc-id>. If an error occurs, this action silently fails
4445 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004446
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004447http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004448
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004449 This action sets the GPT0 tag according to the sticky counter designated by
4450 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4451 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004452
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004453http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004454
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004455 This is used to set the destination IP address to the value of specified
4456 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4457 but provides the correct IP in a HTTP header; or you want to mask the IP for
4458 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4459 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004460
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004461 Arguments:
4462 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4463 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004464
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004465 Example:
4466 http-request set-dst hdr(x-dst)
4467 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004468
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004469 When possible, set-dst preserves the original destination port as long as the
4470 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004471
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004472http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004473
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004474 This is used to set the destination port address to the value of specified
4475 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4476 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004477
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004478 Arguments:
4479 <expr> Is a standard HAProxy expression formed by a sample-fetch
4480 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004481
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004482 Example:
4483 http-request set-dst-port hdr(x-port)
4484 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004485
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004486 When possible, set-dst-port preserves the original destination address as
4487 long as the address family supports a port, otherwise it forces the
4488 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004489
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004490http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004491
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004492 This does the same as "http-request add-header" except that the header name
4493 is first removed if it existed. This is useful when passing security
4494 information to the server, where the header must not be manipulated by
4495 external users. Note that the new value is computed before the removal so it
4496 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004497
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004498 Example:
4499 http-request set-header X-Haproxy-Current-Date %T
4500 http-request set-header X-SSL %[ssl_fc]
4501 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4502 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4503 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4504 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4505 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4506 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4507 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004508
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004509http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004510
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004511 This is used to change the log level of the current request when a certain
4512 condition is met. Valid levels are the 8 syslog levels (see the "log"
4513 keyword) plus the special level "silent" which disables logging for this
4514 request. This rule is not final so the last matching rule wins. This rule
4515 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004516
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004517http-request set-map(<file-name>) <key fmt> <value fmt>
4518 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004519
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004520 This is used to add a new entry into a MAP. The MAP must be loaded from a
4521 file (even a dummy empty file). The file name of the MAP to be updated is
4522 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4523 log-format rules, used to collect MAP key, and <value fmt>, which follows
4524 log-format rules, used to collect content for the new entry.
4525 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4526 more) values. This lookup is done by a linear search and can be expensive
4527 with large lists! It is the equivalent of the "set map" command from the
4528 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004529
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004530http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004531
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004532 This is used to set the Netfilter MARK on all packets sent to the client to
4533 the value passed in <mark> on platforms which support it. This value is an
4534 unsigned 32 bit value which can be matched by netfilter and by the routing
4535 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4536 "0x"). This can be useful to force certain packets to take a different route
4537 (for example a cheaper network path for bulk downloads). This works on Linux
4538 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004539
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004540http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004541
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004542 This rewrites the request method with the result of the evaluation of format
4543 string <fmt>. There should be very few valid reasons for having to do so as
4544 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004545
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004546http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004547
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004548 This sets the "nice" factor of the current request being processed. It only
4549 has effect against the other requests being processed at the same time.
4550 The default value is 0, unless altered by the "nice" setting on the "bind"
4551 line. The accepted range is -1024..1024. The higher the value, the nicest
4552 the request will be. Lower values will make the request more important than
4553 other ones. This can be useful to improve the speed of some requests, or
4554 lower the priority of non-important requests. Using this setting without
4555 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004556
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004557http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004558
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004559 This rewrites the request path with the result of the evaluation of format
4560 string <fmt>. The query string, if any, is left intact. If a scheme and
4561 authority is found before the path, they are left intact as well. If the
4562 request doesn't have a path ("*"), this one is replaced with the format.
4563 This can be used to prepend a directory component in front of a path for
4564 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004565
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004566 Example :
4567 # prepend the host name before the path
4568 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004569
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004570http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004571
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004572 This is used to set the queue priority class of the current request.
4573 The value must be a sample expression which converts to an integer in the
4574 range -2047..2047. Results outside this range will be truncated.
4575 The priority class determines the order in which queued requests are
4576 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004577
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004578http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004579
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004580 This is used to set the queue priority timestamp offset of the current
4581 request. The value must be a sample expression which converts to an integer
4582 in the range -524287..524287. Results outside this range will be truncated.
4583 When a request is queued, it is ordered first by the priority class, then by
4584 the current timestamp adjusted by the given offset in milliseconds. Lower
4585 values have higher priority.
4586 Note that the resulting timestamp is is only tracked with enough precision
4587 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4588 adjusted timestamp exceeds this value, it will be misidentified as highest
4589 priority. Thus it is important to set "timeout queue" to a value, where when
4590 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004591
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004592http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004593
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004594 This rewrites the request's query string which appears after the first
4595 question mark ("?") with the result of the evaluation of format string <fmt>.
4596 The part prior to the question mark is left intact. If the request doesn't
4597 contain a question mark and the new value is not empty, then one is added at
4598 the end of the URI, followed by the new value. If a question mark was
4599 present, it will never be removed even if the value is empty. This can be
4600 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004601
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004602 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004603
4604 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004605 # replace "%3D" with "=" in the query string
4606 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004607
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004608http-request set-src <expr> [ { if | unless } <condition> ]
4609 This is used to set the source IP address to the value of specified
4610 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4611 provides the correct IP in a HTTP header; or you want to mask source IP for
4612 privacy.
4613
4614 Arguments :
4615 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4616 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004617
Cyril Bonté78caf842010-03-10 22:41:43 +01004618 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004619 http-request set-src hdr(x-forwarded-for)
4620 http-request set-src src,ipmask(24)
4621
4622 When possible, set-src preserves the original source port as long as the
4623 address family allows it, otherwise the source port is set to 0.
4624
4625http-request set-src-port <expr> [ { if | unless } <condition> ]
4626
4627 This is used to set the source port address to the value of specified
4628 expression.
4629
4630 Arguments:
4631 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4632 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004633
Willy Tarreau20b0de52012-12-24 15:45:22 +01004634 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004635 http-request set-src-port hdr(x-port)
4636 http-request set-src-port int(4000)
4637
4638 When possible, set-src-port preserves the original source address as long as
4639 the address family supports a port, otherwise it forces the source address to
4640 IPv4 "0.0.0.0" before rewriting the port.
4641
4642http-request set-tos <tos> [ { if | unless } <condition> ]
4643
4644 This is used to set the TOS or DSCP field value of packets sent to the client
4645 to the value passed in <tos> on platforms which support this. This value
4646 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4647 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4648 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4649 be used to adjust some routing behavior on border routers based on some
4650 information from the request.
4651
4652 See RFC 2474, 2597, 3260 and 4594 for more information.
4653
4654http-request set-uri <fmt> [ { if | unless } <condition> ]
4655
4656 This rewrites the request URI with the result of the evaluation of format
4657 string <fmt>. The scheme, authority, path and query string are all replaced
4658 at once. This can be used to rewrite hosts in front of proxies, or to
4659 perform complex modifications to the URI such as moving parts between the
4660 path and the query string.
4661 See also "http-request set-path" and "http-request set-query".
4662
4663http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4664
4665 This is used to set the contents of a variable. The variable is declared
4666 inline.
4667
4668 Arguments:
4669 <var-name> The name of the variable starts with an indication about its
4670 scope. The scopes allowed are:
4671 "proc" : the variable is shared with the whole process
4672 "sess" : the variable is shared with the whole session
4673 "txn" : the variable is shared with the transaction
4674 (request and response)
4675 "req" : the variable is shared only during request
4676 processing
4677 "res" : the variable is shared only during response
4678 processing
4679 This prefix is followed by a name. The separator is a '.'.
4680 The name may only contain characters 'a-z', 'A-Z', '0-9'
4681 and '_'.
4682
4683 <expr> Is a standard HAProxy expression formed by a sample-fetch
4684 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004685
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004686 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004687 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004688
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004689http-request send-spoe-group <engine-name> <group-name>
4690 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004691
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004692 This action is used to trigger sending of a group of SPOE messages. To do so,
4693 the SPOE engine used to send messages must be defined, as well as the SPOE
4694 group to send. Of course, the SPOE engine must refer to an existing SPOE
4695 filter. If not engine name is provided on the SPOE filter line, the SPOE
4696 agent name must be used.
4697
4698 Arguments:
4699 <engine-name> The SPOE engine name.
4700
4701 <group-name> The SPOE group name as specified in the engine
4702 configuration.
4703
4704http-request silent-drop [ { if | unless } <condition> ]
4705
4706 This stops the evaluation of the rules and makes the client-facing connection
4707 suddenly disappear using a system-dependent way that tries to prevent the
4708 client from being notified. The effect it then that the client still sees an
4709 established connection while there's none on HAProxy. The purpose is to
4710 achieve a comparable effect to "tarpit" except that it doesn't use any local
4711 resource at all on the machine running HAProxy. It can resist much higher
4712 loads than "tarpit", and slow down stronger attackers. It is important to
4713 understand the impact of using this mechanism. All stateful equipment placed
4714 between the client and HAProxy (firewalls, proxies, load balancers) will also
4715 keep the established connection for a long time and may suffer from this
4716 action.
4717 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4718 option is used to block the emission of a TCP reset. On other systems, the
4719 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4720 router, though it's still delivered to local networks. Do not use it unless
4721 you fully understand how it works.
4722
4723http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4724
4725 This stops the evaluation of the rules and immediately blocks the request
4726 without responding for a delay specified by "timeout tarpit" or
4727 "timeout connect" if the former is not set. After that delay, if the client
4728 is still connected, an HTTP error 500 (or optionally the status code
4729 specified as an argument to "deny_status") is returned so that the client
4730 does not suspect it has been tarpitted. Logs will report the flags "PT".
4731 The goal of the tarpit rule is to slow down robots during an attack when
4732 they're limited on the number of concurrent requests. It can be very
4733 efficient against very dumb robots, and will significantly reduce the load
4734 on firewalls compared to a "deny" rule. But when facing "correctly"
4735 developed robots, it can make things worse by forcing haproxy and the front
4736 firewall to support insane number of concurrent connections.
4737 See also the "silent-drop" action.
4738
4739http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4740http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4741http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4742
4743 This enables tracking of sticky counters from current request. These rules do
4744 not stop evaluation and do not change default action. The number of counters
4745 that may be simultaneously tracked by the same connection is set in
4746 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4747 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4748 "track-sc0" rule executed enables tracking of the counters of the specified
4749 table as the first set. The first "track-sc1" rule executed enables tracking
4750 of the counters of the specified table as the second set. The first
4751 "track-sc2" rule executed enables tracking of the counters of the specified
4752 table as the third set. It is a recommended practice to use the first set of
4753 counters for the per-frontend counters and the second set for the per-backend
4754 ones. But this is just a guideline, all may be used everywhere.
4755
4756 Arguments :
4757 <key> is mandatory, and is a sample expression rule as described in
4758 section 7.3. It describes what elements of the incoming request or
4759 connection will be analyzed, extracted, combined, and used to
4760 select which table entry to update the counters.
4761
4762 <table> is an optional table to be used instead of the default one, which
4763 is the stick-table declared in the current proxy. All the counters
4764 for the matches and updates for the key will then be performed in
4765 that table until the session ends.
4766
4767 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4768 it is not found, an entry is allocated for it. Then a pointer to that entry
4769 is kept during all the session's life, and this entry's counters are updated
4770 as often as possible, every time the session's counters are updated, and also
4771 systematically when the session ends. Counters are only updated for events
4772 that happen after the tracking has been started. As an exception, connection
4773 counters and request counters are systematically updated so that they reflect
4774 useful information.
4775
4776 If the entry tracks concurrent connection counters, one connection is counted
4777 for as long as the entry is tracked, and the entry will not expire during
4778 that time. Tracking counters also provides a performance advantage over just
4779 checking the keys, because only one table lookup is performed for all ACL
4780 checks that make use of it.
4781
4782http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4783
4784 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004785
4786 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004787 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004788
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004789http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004790
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004791 This will delay the processing of the request until the SSL handshake
4792 happened. This is mostly useful to delay processing early data until we're
4793 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004794
Willy Tarreauef781042010-01-27 11:53:01 +01004795
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004796http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004797 Access control for Layer 7 responses
4798
4799 May be used in sections: defaults | frontend | listen | backend
4800 no | yes | yes | yes
4801
4802 The http-response statement defines a set of rules which apply to layer 7
4803 processing. The rules are evaluated in their declaration order when they are
4804 met in a frontend, listen or backend section. Any rule may optionally be
4805 followed by an ACL-based condition, in which case it will only be evaluated
4806 if the condition is true. Since these rules apply on responses, the backend
4807 rules are applied first, followed by the frontend's rules.
4808
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004809 The first keyword is the rule's action. The supported actions are described
4810 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004811
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004812 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004813
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004814 Example:
4815 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004816
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004817 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06004818
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004819 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4820 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06004821
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004822 Example:
4823 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06004824
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004825 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004826
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004827 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4828 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06004829
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004830 See also : "http-request", section 3.4 about userlists and section 7 about
4831 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004832
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004833http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004834
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004835 This is used to add a new entry into an ACL. The ACL must be loaded from a
4836 file (even a dummy empty file). The file name of the ACL to be updated is
4837 passed between parentheses. It takes one argument: <key fmt>, which follows
4838 log-format rules, to collect content of the new entry. It performs a lookup
4839 in the ACL before insertion, to avoid duplicated (or more) values.
4840 This lookup is done by a linear search and can be expensive with large lists!
4841 It is the equivalent of the "add acl" command from the stats socket, but can
4842 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06004843
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004844http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004845
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004846 This appends an HTTP header field whose name is specified in <name> and whose
4847 value is defined by <fmt> which follows the log-format rules (see Custom Log
4848 Format in section 8.2.4). This may be used to send a cookie to a client for
4849 example, or to pass some internal information.
4850 This rule is not final, so it is possible to add other similar rules.
4851 Note that header addition is performed immediately, so one rule might reuse
4852 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004853
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004854http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004855
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004856 This stops the evaluation of the rules and lets the response pass the check.
4857 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06004858
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004859http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004860
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004861 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06004862
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004863http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004864
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004865 This captures sample expression <sample> from the response buffer, and
4866 converts it to a string. The resulting string is stored into the next request
4867 "capture" slot, so it will possibly appear next to some captured HTTP
4868 headers. It will then automatically appear in the logs, and it will be
4869 possible to extract it using sample fetch rules to feed it into headers or
4870 anything. Please check section 7.3 (Fetching samples) and
4871 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004872
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004873 The keyword "id" is the id of the capture slot which is used for storing the
4874 string. The capture slot must be defined in an associated frontend.
4875 This is useful to run captures in backends. The slot id can be declared by a
4876 previous directive "http-response capture" or with the "declare capture"
4877 keyword.
4878 If the slot <id> doesn't exist, then HAProxy fails parsing the configuration
4879 to prevent unexpected behavior at run time.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004880
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004881http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004882
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004883 This is used to delete an entry from an ACL. The ACL must be loaded from a
4884 file (even a dummy empty file). The file name of the ACL to be updated is
4885 passed between parentheses. It takes one argument: <key fmt>, which follows
4886 log-format rules, to collect content of the entry to delete.
4887 It is the equivalent of the "del acl" command from the stats socket, but can
4888 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004889
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004890http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004891
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004892 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004893
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004894http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02004895
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004896 This is used to delete an entry from a MAP. The MAP must be loaded from a
4897 file (even a dummy empty file). The file name of the MAP to be updated is
4898 passed between parentheses. It takes one argument: <key fmt>, which follows
4899 log-format rules, to collect content of the entry to delete.
4900 It takes one argument: "file name" It is the equivalent of the "del map"
4901 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004902
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004903http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004904
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004905 This stops the evaluation of the rules and immediately rejects the response
4906 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004907
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004908http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004909
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004910 This performs an HTTP redirection based on a redirect rule.
4911 This supports a format string similarly to "http-request redirect" rules,
4912 with the exception that only the "location" type of redirect is possible on
4913 the response. See the "redirect" keyword for the rule's syntax. When a
4914 redirect rule is applied during a response, connections to the server are
4915 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004916
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004917http-response replace-header <name> <regex-match> <replace-fmt>
4918 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004919
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004920 This matches the regular expression in all occurrences of header field <name>
4921 according to <match-regex>, and replaces them with the <replace-fmt> argument.
4922 Format characters are allowed in replace-fmt and work like in <fmt> arguments
4923 in "add-header". The match is only case-sensitive. It is important to
4924 understand that this action only considers whole header lines, regardless of
4925 the number of values they may contain. This usage is suited to headers
4926 naturally containing commas in their value, such as Set-Cookie, Expires and
4927 so on.
William Lallemand86d0df02017-11-24 21:36:45 +01004928
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004929 Example:
4930 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02004931
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004932 # applied to:
4933 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004934
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004935 # outputs:
4936 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 +02004937
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004938 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004939
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004940http-response replace-value <name> <regex-match> <replace-fmt>
4941 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004942
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004943 This works like "replace-header" except that it matches the regex against
4944 every comma-delimited value of the header field <name> instead of the entire
4945 header. This is suited for all headers which are allowed to carry more than
4946 one value. An example could be the Accept header.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004947
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004948 Example:
4949 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004950
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004951 # applied to:
4952 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01004953
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004954 # outputs:
4955 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004956
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004957http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4958http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004959
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004960 This action increments the GPC0 or GPC1 counter according with the sticky
4961 counter designated by <sc-id>. If an error occurs, this action silently fails
4962 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004963
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004964http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004965
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004966 This action sets the GPT0 tag according to the sticky counter designated by
4967 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4968 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004969
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004970http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02004971
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004972 This action is used to trigger sending of a group of SPOE messages. To do so,
4973 the SPOE engine used to send messages must be defined, as well as the SPOE
4974 group to send. Of course, the SPOE engine must refer to an existing SPOE
4975 filter. If not engine name is provided on the SPOE filter line, the SPOE
4976 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004977
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004978 Arguments:
4979 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004980
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004981 <group-name> The SPOE group name as specified in the engine
4982 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004983
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004984http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004985
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004986 This does the same as "add-header" except that the header name is first
4987 removed if it existed. This is useful when passing security information to
4988 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004989
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004990http-response set-log-level <level> [ { if | unless } <condition> ]
4991
4992 This is used to change the log level of the current request when a certain
4993 condition is met. Valid levels are the 8 syslog levels (see the "log"
4994 keyword) plus the special level "silent" which disables logging for this
4995 request. This rule is not final so the last matching rule wins. This rule can
4996 be useful to disable health checks coming from another equipment.
4997
4998http-response set-map(<file-name>) <key fmt> <value fmt>
4999
5000 This is used to add a new entry into a MAP. The MAP must be loaded from a
5001 file (even a dummy empty file). The file name of the MAP to be updated is
5002 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5003 log-format rules, used to collect MAP key, and <value fmt>, which follows
5004 log-format rules, used to collect content for the new entry. It performs a
5005 lookup in the MAP before insertion, to avoid duplicated (or more) values.
5006 This lookup is done by a linear search and can be expensive with large lists!
5007 It is the equivalent of the "set map" command from the stats socket, but can
5008 be triggered by an HTTP response.
5009
5010http-response set-mark <mark> [ { if | unless } <condition> ]
5011
5012 This is used to set the Netfilter MARK on all packets sent to the client to
5013 the value passed in <mark> on platforms which support it. This value is an
5014 unsigned 32 bit value which can be matched by netfilter and by the routing
5015 table. It can be expressed both in decimal or hexadecimal format (prefixed
5016 by "0x"). This can be useful to force certain packets to take a different
5017 route (for example a cheaper network path for bulk downloads). This works on
5018 Linux kernels 2.6.32 and above and requires admin privileges.
5019
5020http-response set-nice <nice> [ { if | unless } <condition> ]
5021
5022 This sets the "nice" factor of the current request being processed.
5023 It only has effect against the other requests being processed at the same
5024 time. The default value is 0, unless altered by the "nice" setting on the
5025 "bind" line. The accepted range is -1024..1024. The higher the value, the
5026 nicest the request will be. Lower values will make the request more important
5027 than other ones. This can be useful to improve the speed of some requests, or
5028 lower the priority of non-important requests. Using this setting without
5029 prior experimentation can cause some major slowdown.
5030
5031http-response set-status <status> [reason <str>]
5032 [ { if | unless } <condition> ]
5033
5034 This replaces the response status code with <status> which must be an integer
5035 between 100 and 999. Optionally, a custom reason text can be provided defined
5036 by <str>, or the default reason for the specified code will be used as a
5037 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005038
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005039 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005040 # return "431 Request Header Fields Too Large"
5041 http-response set-status 431
5042 # return "503 Slow Down", custom reason
5043 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005044
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005045http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005046
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005047 This is used to set the TOS or DSCP field value of packets sent to the client
5048 to the value passed in <tos> on platforms which support this.
5049 This value represents the whole 8 bits of the IP TOS field, and can be
5050 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
5051 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
5052 always 0. This can be used to adjust some routing behavior on border routers
5053 based on some information from the request.
5054
5055 See RFC 2474, 2597, 3260 and 4594 for more information.
5056
5057http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5058
5059 This is used to set the contents of a variable. The variable is declared
5060 inline.
5061
5062 Arguments:
5063 <var-name> The name of the variable starts with an indication about its
5064 scope. The scopes allowed are:
5065 "proc" : the variable is shared with the whole process
5066 "sess" : the variable is shared with the whole session
5067 "txn" : the variable is shared with the transaction
5068 (request and response)
5069 "req" : the variable is shared only during request
5070 processing
5071 "res" : the variable is shared only during response
5072 processing
5073 This prefix is followed by a name. The separator is a '.'.
5074 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5075 and '_'.
5076
5077 <expr> Is a standard HAProxy expression formed by a sample-fetch
5078 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005079
5080 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005081 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005082
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005083http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005084
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005085 This stops the evaluation of the rules and makes the client-facing connection
5086 suddenly disappear using a system-dependent way that tries to prevent the
5087 client from being notified. The effect it then that the client still sees an
5088 established connection while there's none on HAProxy. The purpose is to
5089 achieve a comparable effect to "tarpit" except that it doesn't use any local
5090 resource at all on the machine running HAProxy. It can resist much higher
5091 loads than "tarpit", and slow down stronger attackers. It is important to
5092 understand the impact of using this mechanism. All stateful equipment placed
5093 between the client and HAProxy (firewalls, proxies, load balancers) will also
5094 keep the established connection for a long time and may suffer from this
5095 action.
5096 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
5097 option is used to block the emission of a TCP reset. On other systems, the
5098 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
5099 router, though it's still delivered to local networks. Do not use it unless
5100 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005101
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005102http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
5103http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
5104http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005105
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005106 This enables tracking of sticky counters from current response. Please refer
5107 to "http-request track-sc" for a complete description. The only difference
5108 from "http-request track-sc" is the <key> sample expression can only make use
5109 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
5110 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
5111 supported, haproxy will fail and warn while parsing the config.
5112
5113http-response unset-var(<var-name>) [ { if | unless } <condition> ]
5114
5115 This is used to unset a variable. See "http-response set-var" for details
5116 about <var-name>.
5117
5118 Example:
5119 http-response unset-var(sess.last_redir)
5120
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02005121
Willy Tarreau30631952015-08-06 15:05:24 +02005122http-reuse { never | safe | aggressive | always }
5123 Declare how idle HTTP connections may be shared between requests
5124
5125 May be used in sections: defaults | frontend | listen | backend
5126 yes | no | yes | yes
5127
5128 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01005129 which is considered safe for reuse is moved back to the server's idle
5130 connections pool so that any other request can make use of it. This is the
5131 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02005132
5133 The argument indicates the desired connection reuse strategy :
5134
Olivier Houchard86006a52018-12-14 19:37:49 +01005135 - "never" : idle connections are never shared between sessions. This mode
5136 may be enforced to cancel a different strategy inherited from
5137 a defaults section or for troubleshooting. For example, if an
5138 old bogus application considers that multiple requests over
5139 the same connection come from the same client and it is not
5140 possible to fix the application, it may be desirable to
5141 disable connection sharing in a single backend. An example of
5142 such an application could be an old haproxy using cookie
5143 insertion in tunnel mode and not checking any request past the
5144 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02005145
Olivier Houchard86006a52018-12-14 19:37:49 +01005146 - "safe" : this is the default and the recommended strategy. The first
5147 request of a session is always sent over its own connection,
5148 and only subsequent requests may be dispatched over other
5149 existing connections. This ensures that in case the server
5150 closes the connection when the request is being sent, the
5151 browser can decide to silently retry it. Since it is exactly
5152 equivalent to regular keep-alive, there should be no side
5153 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02005154
5155 - "aggressive" : this mode may be useful in webservices environments where
5156 all servers are not necessarily known and where it would be
5157 appreciable to deliver most first requests over existing
5158 connections. In this case, first requests are only delivered
5159 over existing connections that have been reused at least once,
5160 proving that the server correctly supports connection reuse.
5161 It should only be used when it's sure that the client can
5162 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02005163 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02005164 downsides of rare connection failures.
5165
5166 - "always" : this mode is only recommended when the path to the server is
5167 known for never breaking existing connections quickly after
5168 releasing them. It allows the first request of a session to be
5169 sent to an existing connection. This can provide a significant
5170 performance increase over the "safe" strategy when the backend
5171 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005172 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02005173 sharing. It is recommended that the "http-keep-alive" timeout
5174 remains low in this mode so that no dead connections remain
5175 usable. In most cases, this will lead to the same performance
5176 gains as "aggressive" but with more risks. It should only be
5177 used when it improves the situation over "aggressive".
5178
5179 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005180 connection properties and compatibility. Specifically :
5181 - connections made with "usesrc" followed by a client-dependent value
5182 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005183
5184 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01005185 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005186
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02005187 - connections with certain bogus authentication schemes (relying on the
5188 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005189
5190 No connection pool is involved, once a session dies, the last idle connection
5191 it was attached to is deleted at the same time. This ensures that connections
5192 may not last after all sessions are closed.
5193
5194 Note: connection reuse improves the accuracy of the "server maxconn" setting,
5195 because almost no new connection will be established while idle connections
5196 remain available. This is particularly true with the "always" strategy.
5197
5198 See also : "option http-keep-alive", "server maxconn"
5199
5200
Mark Lamourinec2247f02012-01-04 13:02:01 -05005201http-send-name-header [<header>]
5202 Add the server name to a request. Use the header string given by <header>
5203
5204 May be used in sections: defaults | frontend | listen | backend
5205 yes | no | yes | yes
5206
5207 Arguments :
5208
5209 <header> The header string to use to send the server name
5210
5211 The "http-send-name-header" statement causes the name of the target
Davor Ocelice9ed2812017-12-25 17:49:28 +01005212 server to be added to the headers of an HTTP request. The name
Mark Lamourinec2247f02012-01-04 13:02:01 -05005213 is added with the header string proved.
5214
5215 See also : "server"
5216
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005217id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005218 Set a persistent ID to a proxy.
5219 May be used in sections : defaults | frontend | listen | backend
5220 no | yes | yes | yes
5221 Arguments : none
5222
5223 Set a persistent ID for the proxy. This ID must be unique and positive.
5224 An unused ID will automatically be assigned if unset. The first assigned
5225 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005226
5227
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005228ignore-persist { if | unless } <condition>
5229 Declare a condition to ignore persistence
5230 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005231 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005232
5233 By default, when cookie persistence is enabled, every requests containing
5234 the cookie are unconditionally persistent (assuming the target server is up
5235 and running).
5236
5237 The "ignore-persist" statement allows one to declare various ACL-based
5238 conditions which, when met, will cause a request to ignore persistence.
5239 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005240 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005241 persistence for a specific User-Agent (for example, some web crawler bots).
5242
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005243 The persistence is ignored when an "if" condition is met, or unless an
5244 "unless" condition is met.
5245
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005246 Example:
5247 acl url_static path_beg /static /images /img /css
5248 acl url_static path_end .gif .png .jpg .css .js
5249 ignore-persist if url_static
5250
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005251 See also : "force-persist", "cookie", and section 7 about ACL usage.
5252
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005253load-server-state-from-file { global | local | none }
5254 Allow seamless reload of HAProxy
5255 May be used in sections: defaults | frontend | listen | backend
5256 yes | no | yes | yes
5257
5258 This directive points HAProxy to a file where server state from previous
5259 running process has been saved. That way, when starting up, before handling
5260 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005261 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005262 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5263 loading state or load states from a file containing all backends and servers.
5264 The state file can be generated by running the command "show servers state"
5265 over the stats socket and redirect output.
5266
Davor Ocelice9ed2812017-12-25 17:49:28 +01005267 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005268 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005269 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005270
5271 Arguments:
5272 global load the content of the file pointed by the global directive
5273 named "server-state-file".
5274
5275 local load the content of the file pointed by the directive
5276 "server-state-file-name" if set. If not set, then the backend
5277 name is used as a file name.
5278
5279 none don't load any stat for this backend
5280
5281 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005282 - server's IP address is preserved across reloads by default, but the
5283 order can be changed thanks to the server's "init-addr" setting. This
5284 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005285 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005286 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005287
5288 - server's weight is applied from previous running process unless it has
5289 has changed between previous and new configuration files.
5290
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005291 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005292
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005293 global
5294 stats socket /tmp/socket
5295 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005296
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005297 defaults
5298 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005299
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005300 backend bk
5301 server s1 127.0.0.1:22 check weight 11
5302 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005303
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005304
5305 Then one can run :
5306
5307 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5308
5309 Content of the file /tmp/server_state would be like this:
5310
5311 1
5312 # <field names skipped for the doc example>
5313 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5314 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5315
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005316 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005317
5318 global
5319 stats socket /tmp/socket
5320 server-state-base /etc/haproxy/states
5321
5322 defaults
5323 load-server-state-from-file local
5324
5325 backend bk
5326 server s1 127.0.0.1:22 check weight 11
5327 server s2 127.0.0.1:22 check weight 12
5328
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005329
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005330 Then one can run :
5331
5332 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5333
5334 Content of the file /etc/haproxy/states/bk would be like this:
5335
5336 1
5337 # <field names skipped for the doc example>
5338 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5339 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5340
5341 See also: "server-state-file", "server-state-file-name", and
5342 "show servers state"
5343
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005344
Willy Tarreau2769aa02007-12-27 18:26:09 +01005345log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005346log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
5347 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005348no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005349 Enable per-instance logging of events and traffic.
5350 May be used in sections : defaults | frontend | listen | backend
5351 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005352
5353 Prefix :
5354 no should be used when the logger list must be flushed. For example,
5355 if you don't want to inherit from the default logger list. This
5356 prefix does not allow arguments.
5357
Willy Tarreau2769aa02007-12-27 18:26:09 +01005358 Arguments :
5359 global should be used when the instance's logging parameters are the
5360 same as the global ones. This is the most common usage. "global"
5361 replaces <address>, <facility> and <level> with those of the log
5362 entries found in the "global" section. Only one "log global"
5363 statement may be used per instance, and this form takes no other
5364 parameter.
5365
5366 <address> indicates where to send the logs. It takes the same format as
5367 for the "global" section's logs, and can be one of :
5368
5369 - An IPv4 address optionally followed by a colon (':') and a UDP
5370 port. If no port is specified, 514 is used by default (the
5371 standard syslog port).
5372
David du Colombier24bb5f52011-03-17 10:40:23 +01005373 - An IPv6 address followed by a colon (':') and optionally a UDP
5374 port. If no port is specified, 514 is used by default (the
5375 standard syslog port).
5376
Willy Tarreau2769aa02007-12-27 18:26:09 +01005377 - A filesystem path to a UNIX domain socket, keeping in mind
5378 considerations for chroot (be sure the path is accessible
5379 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005380 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005381
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005382 - A file descriptor number in the form "fd@<number>", which may
5383 point to a pipe, terminal, or socket. In this case unbuffered
5384 logs are used and one writev() call per log is performed. This
5385 is a bit expensive but acceptable for most workloads. Messages
5386 sent this way will not be truncated but may be dropped, in
5387 which case the DroppedLogs counter will be incremented. The
5388 writev() call is atomic even on pipes for messages up to
5389 PIPE_BUF size, which POSIX recommends to be at least 512 and
5390 which is 4096 bytes on most modern operating systems. Any
5391 larger message may be interleaved with messages from other
5392 processes. Exceptionally for debugging purposes the file
5393 descriptor may also be directed to a file, but doing so will
5394 significantly slow haproxy down as non-blocking calls will be
5395 ignored. Also there will be no way to purge nor rotate this
5396 file without restarting the process. Note that the configured
5397 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005398 with a TCP syslog server. See also the "short" and "raw"
5399 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005400
5401 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5402 and "fd@2", see above.
5403
Willy Tarreauc046d162019-08-30 15:24:59 +02005404 - A ring buffer in the form "ring@<name>", which will correspond
5405 to an in-memory ring buffer accessible over the CLI using the
5406 "show events" command, which will also list existing rings and
5407 their sizes. Such buffers are lost on reload or restart but
5408 when used as a complement this can help troubleshooting by
5409 having the logs instantly available.
5410
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005411 You may want to reference some environment variables in the
5412 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005413
Willy Tarreau18324f52014-06-27 18:10:07 +02005414 <length> is an optional maximum line length. Log lines larger than this
5415 value will be truncated before being sent. The reason is that
5416 syslog servers act differently on log line length. All servers
5417 support the default value of 1024, but some servers simply drop
5418 larger lines while others do log them. If a server supports long
5419 lines, it may make sense to set this value here in order to avoid
5420 truncating long lines. Similarly, if a server drops long lines,
5421 it is preferable to truncate them before sending them. Accepted
5422 values are 80 to 65535 inclusive. The default value of 1024 is
5423 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005424 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005425
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005426 <ranges> A list of comma-separated ranges to identify the logs to sample.
5427 This is used to balance the load of the logs to send to the log
5428 server. The limits of the ranges cannot be null. They are numbered
5429 from 1. The size or period (in number of logs) of the sample must
5430 be set with <sample_size> parameter.
5431
5432 <sample_size>
5433 The size of the sample in number of logs to consider when balancing
5434 their logging loads. It is used to balance the load of the logs to
5435 send to the syslog server. This size must be greater or equal to the
5436 maximum of the high limits of the ranges.
5437 (see also <ranges> parameter).
5438
Willy Tarreauadb345d2018-11-12 07:56:13 +01005439 <format> is the log format used when generating syslog messages. It may be
5440 one of the following :
5441
5442 rfc3164 The RFC3164 syslog message format. This is the default.
5443 (https://tools.ietf.org/html/rfc3164)
5444
5445 rfc5424 The RFC5424 syslog message format.
5446 (https://tools.ietf.org/html/rfc5424)
5447
Willy Tarreaue8746a02018-11-12 08:45:00 +01005448 short A message containing only a level between angle brackets such as
5449 '<3>', followed by the text. The PID, date, time, process name
5450 and system name are omitted. This is designed to be used with a
5451 local log server. This format is compatible with what the
5452 systemd logger consumes.
5453
Willy Tarreauc1b06452018-11-12 11:57:56 +01005454 raw A message containing only the text. The level, PID, date, time,
5455 process name and system name are omitted. This is designed to
5456 be used in containers or during development, where the severity
5457 only depends on the file descriptor used (stdout/stderr).
5458
Willy Tarreau2769aa02007-12-27 18:26:09 +01005459 <facility> must be one of the 24 standard syslog facilities :
5460
Willy Tarreaue8746a02018-11-12 08:45:00 +01005461 kern user mail daemon auth syslog lpr news
5462 uucp cron auth2 ftp ntp audit alert cron2
5463 local0 local1 local2 local3 local4 local5 local6 local7
5464
Willy Tarreauc1b06452018-11-12 11:57:56 +01005465 Note that the facility is ignored for the "short" and "raw"
5466 formats, but still required as a positional field. It is
5467 recommended to use "daemon" in this case to make it clear that
5468 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005469
5470 <level> is optional and can be specified to filter outgoing messages. By
5471 default, all messages are sent. If a level is specified, only
5472 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005473 will be sent. An optional minimum level can be specified. If it
5474 is set, logs emitted with a more severe level than this one will
5475 be capped to this level. This is used to avoid sending "emerg"
5476 messages on all terminals on some default syslog configurations.
5477 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005478
5479 emerg alert crit err warning notice info debug
5480
William Lallemand0f99e342011-10-12 17:50:54 +02005481 It is important to keep in mind that it is the frontend which decides what to
5482 log from a connection, and that in case of content switching, the log entries
5483 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005484
5485 However, backend log declaration define how and where servers status changes
5486 will be logged. Level "notice" will be used to indicate a server going up,
5487 "warning" will be used for termination signals and definitive service
5488 termination, and "alert" will be used for when a server goes down.
5489
5490 Note : According to RFC3164, messages are truncated to 1024 bytes before
5491 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005492
5493 Example :
5494 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005495 log stdout format short daemon # send log to systemd
5496 log stdout format raw daemon # send everything to stdout
5497 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005498 log 127.0.0.1:514 local0 notice # only send important events
5499 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005500 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005501
Willy Tarreau2769aa02007-12-27 18:26:09 +01005502
William Lallemand48940402012-01-30 16:47:22 +01005503log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005504 Specifies the log format string to use for traffic logs
5505 May be used in sections: defaults | frontend | listen | backend
5506 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005507
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005508 This directive specifies the log format string that will be used for all logs
5509 resulting from traffic passing through the frontend using this line. If the
5510 directive is used in a defaults section, all subsequent frontends will use
5511 the same log format. Please see section 8.2.4 which covers the log format
5512 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005513
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005514 "log-format" directive overrides previous "option tcplog", "log-format" and
5515 "option httplog" directives.
5516
Dragan Dosen7ad31542015-09-28 17:16:47 +02005517log-format-sd <string>
5518 Specifies the RFC5424 structured-data log format string
5519 May be used in sections: defaults | frontend | listen | backend
5520 yes | yes | yes | no
5521
5522 This directive specifies the RFC5424 structured-data log format string that
5523 will be used for all logs resulting from traffic passing through the frontend
5524 using this line. If the directive is used in a defaults section, all
5525 subsequent frontends will use the same log format. Please see section 8.2.4
5526 which covers the log format string in depth.
5527
5528 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5529 about the RFC5424 structured-data part.
5530
5531 Note : This log format string will be used only for loggers that have set
5532 log format to "rfc5424".
5533
5534 Example :
5535 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5536
5537
Willy Tarreau094af4e2015-01-07 15:03:42 +01005538log-tag <string>
5539 Specifies the log tag to use for all outgoing logs
5540 May be used in sections: defaults | frontend | listen | backend
5541 yes | yes | yes | yes
5542
5543 Sets the tag field in the syslog header to this string. It defaults to the
5544 log-tag set in the global section, otherwise the program name as launched
5545 from the command line, which usually is "haproxy". Sometimes it can be useful
5546 to differentiate between multiple processes running on the same host, or to
5547 differentiate customer instances running in the same process. In the backend,
5548 logs about servers up/down will use this tag. As a hint, it can be convenient
5549 to set a log-tag related to a hosted customer in a defaults section then put
5550 all the frontends and backends for that customer, then start another customer
5551 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005552
Willy Tarreauc35362a2014-04-25 13:58:37 +02005553max-keep-alive-queue <value>
5554 Set the maximum server queue size for maintaining keep-alive connections
5555 May be used in sections: defaults | frontend | listen | backend
5556 yes | no | yes | yes
5557
5558 HTTP keep-alive tries to reuse the same server connection whenever possible,
5559 but sometimes it can be counter-productive, for example if a server has a lot
5560 of connections while other ones are idle. This is especially true for static
5561 servers.
5562
5563 The purpose of this setting is to set a threshold on the number of queued
5564 connections at which haproxy stops trying to reuse the same server and prefers
5565 to find another one. The default value, -1, means there is no limit. A value
5566 of zero means that keep-alive requests will never be queued. For very close
5567 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005568 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005569 use a value of 10 or less). For remote servers suffering from a high latency,
5570 higher values might be needed to cover for the latency and/or the cost of
5571 picking a different server.
5572
5573 Note that this has no impact on responses which are maintained to the same
5574 server consecutively to a 401 response. They will still go to the same server
5575 even if they have to be queued.
5576
5577 See also : "option http-server-close", "option prefer-last-server", server
5578 "maxconn" and cookie persistence.
5579
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01005580max-session-srv-conns <nb>
5581 Set the maximum number of outgoing connections we can keep idling for a given
5582 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
5583 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02005584
Willy Tarreau2769aa02007-12-27 18:26:09 +01005585maxconn <conns>
5586 Fix the maximum number of concurrent connections on a frontend
5587 May be used in sections : defaults | frontend | listen | backend
5588 yes | yes | yes | no
5589 Arguments :
5590 <conns> is the maximum number of concurrent connections the frontend will
5591 accept to serve. Excess connections will be queued by the system
5592 in the socket's listen queue and will be served once a connection
5593 closes.
5594
5595 If the system supports it, it can be useful on big sites to raise this limit
5596 very high so that haproxy manages connection queues, instead of leaving the
5597 clients with unanswered connection attempts. This value should not exceed the
5598 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005599 of tune.bufsize (16kB by default) each, as well as some other data resulting
5600 in about 33 kB of RAM being consumed per established connection. That means
5601 that a medium system equipped with 1GB of RAM can withstand around
5602 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005603
5604 Also, when <conns> is set to large values, it is possible that the servers
5605 are not sized to accept such loads, and for this reason it is generally wise
5606 to assign them some reasonable connection limits.
5607
Willy Tarreauc8d5b952019-02-27 17:25:52 +01005608 When this value is set to zero, which is the default, the global "maxconn"
5609 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02005610
Willy Tarreau2769aa02007-12-27 18:26:09 +01005611 See also : "server", global section's "maxconn", "fullconn"
5612
5613
5614mode { tcp|http|health }
5615 Set the running mode or protocol of the instance
5616 May be used in sections : defaults | frontend | listen | backend
5617 yes | yes | yes | yes
5618 Arguments :
5619 tcp The instance will work in pure TCP mode. A full-duplex connection
5620 will be established between clients and servers, and no layer 7
5621 examination will be performed. This is the default mode. It
5622 should be used for SSL, SSH, SMTP, ...
5623
5624 http The instance will work in HTTP mode. The client request will be
5625 analyzed in depth before connecting to any server. Any request
5626 which is not RFC-compliant will be rejected. Layer 7 filtering,
5627 processing and switching will be possible. This is the mode which
5628 brings HAProxy most of its value.
5629
5630 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005631 to incoming connections and close the connection. Alternatively,
5632 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5633 instead. Nothing will be logged in either case. This mode is used
5634 to reply to external components health checks. This mode is
5635 deprecated and should not be used anymore as it is possible to do
5636 the same and even better by combining TCP or HTTP modes with the
5637 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005638
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005639 When doing content switching, it is mandatory that the frontend and the
5640 backend are in the same mode (generally HTTP), otherwise the configuration
5641 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005642
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005643 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005644 defaults http_instances
5645 mode http
5646
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005647 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005648
Willy Tarreau0ba27502007-12-24 16:55:16 +01005649
Cyril Bontéf0c60612010-02-06 14:44:47 +01005650monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005651 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005652 May be used in sections : defaults | frontend | listen | backend
5653 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005654 Arguments :
5655 if <cond> the monitor request will fail if the condition is satisfied,
5656 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005657 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005658 are met, for instance a low number of servers both in a
5659 backend and its backup.
5660
5661 unless <cond> the monitor request will succeed only if the condition is
5662 satisfied, and will fail otherwise. Such a condition may be
5663 based on a test on the presence of a minimum number of active
5664 servers in a list of backends.
5665
5666 This statement adds a condition which can force the response to a monitor
5667 request to report a failure. By default, when an external component queries
5668 the URI dedicated to monitoring, a 200 response is returned. When one of the
5669 conditions above is met, haproxy will return 503 instead of 200. This is
5670 very useful to report a site failure to an external component which may base
5671 routing advertisements between multiple sites on the availability reported by
5672 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005673 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5674 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005675
5676 Example:
5677 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005678 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005679 acl site_dead nbsrv(dynamic) lt 2
5680 acl site_dead nbsrv(static) lt 2
5681 monitor-uri /site_alive
5682 monitor fail if site_dead
5683
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005684 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005685
5686
5687monitor-net <source>
5688 Declare a source network which is limited to monitor requests
5689 May be used in sections : defaults | frontend | listen | backend
5690 yes | yes | yes | no
5691 Arguments :
5692 <source> is the source IPv4 address or network which will only be able to
5693 get monitor responses to any request. It can be either an IPv4
5694 address, a host name, or an address followed by a slash ('/')
5695 followed by a mask.
5696
5697 In TCP mode, any connection coming from a source matching <source> will cause
5698 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005699 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005700 forwarding the connection to a remote server.
5701
5702 In HTTP mode, a connection coming from a source matching <source> will be
5703 accepted, the following response will be sent without waiting for a request,
5704 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5705 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005706 running without forwarding the request to a backend server. Note that this
5707 response is sent in raw format, without any transformation. This is important
5708 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005709
Willy Tarreau82569f92012-09-27 23:48:56 +02005710 Monitor requests are processed very early, just after tcp-request connection
5711 ACLs which are the only ones able to block them. These connections are short
5712 lived and never wait for any data from the client. They cannot be logged, and
5713 it is the intended purpose. They are only used to report HAProxy's health to
5714 an upper component, nothing more. Please note that "monitor fail" rules do
5715 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005716
Willy Tarreau95cd2832010-03-04 23:36:33 +01005717 Last, please note that only one "monitor-net" statement can be specified in
5718 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005719
Willy Tarreau2769aa02007-12-27 18:26:09 +01005720 Example :
5721 # addresses .252 and .253 are just probing us.
5722 frontend www
5723 monitor-net 192.168.0.252/31
5724
5725 See also : "monitor fail", "monitor-uri"
5726
5727
5728monitor-uri <uri>
5729 Intercept a URI used by external components' monitor requests
5730 May be used in sections : defaults | frontend | listen | backend
5731 yes | yes | yes | no
5732 Arguments :
5733 <uri> is the exact URI which we want to intercept to return HAProxy's
5734 health status instead of forwarding the request.
5735
5736 When an HTTP request referencing <uri> will be received on a frontend,
5737 HAProxy will not forward it nor log it, but instead will return either
5738 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5739 conditions defined with "monitor fail". This is normally enough for any
5740 front-end HTTP probe to detect that the service is UP and running without
5741 forwarding the request to a backend server. Note that the HTTP method, the
5742 version and all headers are ignored, but the request must at least be valid
5743 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5744
Willy Tarreau721d8e02017-12-01 18:25:08 +01005745 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005746 and even before any "http-request". The only rulesets applied before are the
5747 tcp-request ones. They cannot be logged either, and it is the intended
5748 purpose. They are only used to report HAProxy's health to an upper component,
5749 nothing more. However, it is possible to add any number of conditions using
5750 "monitor fail" and ACLs so that the result can be adjusted to whatever check
5751 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005752
5753 Example :
5754 # Use /haproxy_test to report haproxy's status
5755 frontend www
5756 mode http
5757 monitor-uri /haproxy_test
5758
5759 See also : "monitor fail", "monitor-net"
5760
Willy Tarreau0ba27502007-12-24 16:55:16 +01005761
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005762option abortonclose
5763no option abortonclose
5764 Enable or disable early dropping of aborted requests pending in queues.
5765 May be used in sections : defaults | frontend | listen | backend
5766 yes | no | yes | yes
5767 Arguments : none
5768
5769 In presence of very high loads, the servers will take some time to respond.
5770 The per-instance connection queue will inflate, and the response time will
5771 increase respective to the size of the queue times the average per-session
5772 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005773 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005774 the queue, and slowing down other users, and the servers as well, because the
5775 request will eventually be served, then aborted at the first error
5776 encountered while delivering the response.
5777
5778 As there is no way to distinguish between a full STOP and a simple output
5779 close on the client side, HTTP agents should be conservative and consider
5780 that the client might only have closed its output channel while waiting for
5781 the response. However, this introduces risks of congestion when lots of users
5782 do the same, and is completely useless nowadays because probably no client at
5783 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005784 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005785 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005786 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005787 of being the single component to break rare but valid traffic is extremely
5788 low, which adds to the temptation to be able to abort a session early while
5789 still not served and not pollute the servers.
5790
Davor Ocelice9ed2812017-12-25 17:49:28 +01005791 In HAProxy, the user can choose the desired behavior using the option
5792 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005793 compliant and aborted requests will be served. But when the option is
5794 specified, a session with an incoming channel closed will be aborted while
5795 it is still possible, either pending in the queue for a connection slot, or
5796 during the connection establishment if the server has not yet acknowledged
5797 the connection request. This considerably reduces the queue size and the load
5798 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005799 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005800
5801 If this option has been enabled in a "defaults" section, it can be disabled
5802 in a specific instance by prepending the "no" keyword before it.
5803
5804 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5805
5806
Willy Tarreau4076a152009-04-02 15:18:36 +02005807option accept-invalid-http-request
5808no option accept-invalid-http-request
5809 Enable or disable relaxing of HTTP request parsing
5810 May be used in sections : defaults | frontend | listen | backend
5811 yes | yes | yes | no
5812 Arguments : none
5813
Willy Tarreau91852eb2015-05-01 13:26:00 +02005814 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005815 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005816 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005817 forbidden characters are essentially used to build attacks exploiting server
5818 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5819 server will emit invalid header names for whatever reason (configuration,
5820 implementation) and the issue will not be immediately fixed. In such a case,
5821 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005822 even if that does not make sense, by specifying this option. Similarly, the
5823 list of characters allowed to appear in a URI is well defined by RFC3986, and
5824 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5825 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005826 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005827 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005828 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5829 to pass through (no version specified) and multiple digits for both the major
5830 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005831
5832 This option should never be enabled by default as it hides application bugs
5833 and open security breaches. It should only be deployed after a problem has
5834 been confirmed.
5835
5836 When this option is enabled, erroneous header names will still be accepted in
5837 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005838 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5839 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005840 also helps confirming that the issue has been solved.
5841
5842 If this option has been enabled in a "defaults" section, it can be disabled
5843 in a specific instance by prepending the "no" keyword before it.
5844
5845 See also : "option accept-invalid-http-response" and "show errors" on the
5846 stats socket.
5847
5848
5849option accept-invalid-http-response
5850no option accept-invalid-http-response
5851 Enable or disable relaxing of HTTP response parsing
5852 May be used in sections : defaults | frontend | listen | backend
5853 yes | no | yes | yes
5854 Arguments : none
5855
Willy Tarreau91852eb2015-05-01 13:26:00 +02005856 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005857 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005858 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005859 forbidden characters are essentially used to build attacks exploiting server
5860 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5861 server will emit invalid header names for whatever reason (configuration,
5862 implementation) and the issue will not be immediately fixed. In such a case,
5863 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005864 even if that does not make sense, by specifying this option. This option also
5865 relaxes the test on the HTTP version format, it allows multiple digits for
5866 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005867
5868 This option should never be enabled by default as it hides application bugs
5869 and open security breaches. It should only be deployed after a problem has
5870 been confirmed.
5871
5872 When this option is enabled, erroneous header names will still be accepted in
5873 responses, but the complete response will be captured in order to permit
5874 later analysis using the "show errors" request on the UNIX stats socket.
5875 Doing this also helps confirming that the issue has been solved.
5876
5877 If this option has been enabled in a "defaults" section, it can be disabled
5878 in a specific instance by prepending the "no" keyword before it.
5879
5880 See also : "option accept-invalid-http-request" and "show errors" on the
5881 stats socket.
5882
5883
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005884option allbackups
5885no option allbackups
5886 Use either all backup servers at a time or only the first one
5887 May be used in sections : defaults | frontend | listen | backend
5888 yes | no | yes | yes
5889 Arguments : none
5890
5891 By default, the first operational backup server gets all traffic when normal
5892 servers are all down. Sometimes, it may be preferred to use multiple backups
5893 at once, because one will not be enough. When "option allbackups" is enabled,
5894 the load balancing will be performed among all backup servers when all normal
5895 ones are unavailable. The same load balancing algorithm will be used and the
5896 servers' weights will be respected. Thus, there will not be any priority
5897 order between the backup servers anymore.
5898
5899 This option is mostly used with static server farms dedicated to return a
5900 "sorry" page when an application is completely offline.
5901
5902 If this option has been enabled in a "defaults" section, it can be disabled
5903 in a specific instance by prepending the "no" keyword before it.
5904
5905
5906option checkcache
5907no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005908 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005909 May be used in sections : defaults | frontend | listen | backend
5910 yes | no | yes | yes
5911 Arguments : none
5912
5913 Some high-level frameworks set application cookies everywhere and do not
5914 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005915 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005916 high risk of session crossing or stealing between users traversing the same
5917 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005918 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005919
5920 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005921 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005922 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005923 response to check if there's a risk of caching a cookie on a client-side
5924 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005925 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005926 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005927 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5928 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005929 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005930 - all those that result from a request using a method other than GET, HEAD,
5931 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005932 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005933 - those with a 'Pragma: no-cache' header
5934 - those with a 'Cache-control: private' header
5935 - those with a 'Cache-control: no-store' header
5936 - those with a 'Cache-control: max-age=0' header
5937 - those with a 'Cache-control: s-maxage=0' header
5938 - those with a 'Cache-control: no-cache' header
5939 - those with a 'Cache-control: no-cache="set-cookie"' header
5940 - those with a 'Cache-control: no-cache="set-cookie,' header
5941 (allowing other fields after set-cookie)
5942
5943 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005944 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
5945 gateway". The session state shows "PH--" meaning that the proxy blocked the
5946 response during headers processing. Additionally, an alert will be sent in
5947 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005948
5949 Due to the high impact on the application, the application should be tested
5950 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005951 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005952 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005953
5954 If this option has been enabled in a "defaults" section, it can be disabled
5955 in a specific instance by prepending the "no" keyword before it.
5956
5957
5958option clitcpka
5959no option clitcpka
5960 Enable or disable the sending of TCP keepalive packets on the client side
5961 May be used in sections : defaults | frontend | listen | backend
5962 yes | yes | yes | no
5963 Arguments : none
5964
5965 When there is a firewall or any session-aware component between a client and
5966 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005967 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005968 components decides to expire a session which has remained idle for too long.
5969
5970 Enabling socket-level TCP keep-alives makes the system regularly send packets
5971 to the other end of the connection, leaving it active. The delay between
5972 keep-alive probes is controlled by the system only and depends both on the
5973 operating system and its tuning parameters.
5974
5975 It is important to understand that keep-alive packets are neither emitted nor
5976 received at the application level. It is only the network stacks which sees
5977 them. For this reason, even if one side of the proxy already uses keep-alives
5978 to maintain its connection alive, those keep-alive packets will not be
5979 forwarded to the other side of the proxy.
5980
5981 Please note that this has nothing to do with HTTP keep-alive.
5982
5983 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5984 client side of a connection, which should help when session expirations are
5985 noticed between HAProxy and a client.
5986
5987 If this option has been enabled in a "defaults" section, it can be disabled
5988 in a specific instance by prepending the "no" keyword before it.
5989
5990 See also : "option srvtcpka", "option tcpka"
5991
5992
Willy Tarreau0ba27502007-12-24 16:55:16 +01005993option contstats
5994 Enable continuous traffic statistics updates
5995 May be used in sections : defaults | frontend | listen | backend
5996 yes | yes | yes | no
5997 Arguments : none
5998
5999 By default, counters used for statistics calculation are incremented
6000 only when a session finishes. It works quite well when serving small
6001 objects, but with big ones (for example large images or archives) or
6002 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01006003 a hedgehog. With this option enabled counters get incremented frequently
6004 along the session, typically every 5 seconds, which is often enough to
6005 produce clean graphs. Recounting touches a hotpath directly so it is not
6006 not enabled by default, as it can cause a lot of wakeups for very large
6007 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01006008
6009
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006010option dontlog-normal
6011no option dontlog-normal
6012 Enable or disable logging of normal, successful connections
6013 May be used in sections : defaults | frontend | listen | backend
6014 yes | yes | yes | no
6015 Arguments : none
6016
6017 There are large sites dealing with several thousand connections per second
6018 and for which logging is a major pain. Some of them are even forced to turn
6019 logs off and cannot debug production issues. Setting this option ensures that
6020 normal connections, those which experience no error, no timeout, no retry nor
6021 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
6022 mode, the response status code is checked and return codes 5xx will still be
6023 logged.
6024
6025 It is strongly discouraged to use this option as most of the time, the key to
6026 complex issues is in the normal logs which will not be logged here. If you
6027 need to separate logs, see the "log-separate-errors" option instead.
6028
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006029 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006030 logging.
6031
6032
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006033option dontlognull
6034no option dontlognull
6035 Enable or disable logging of null connections
6036 May be used in sections : defaults | frontend | listen | backend
6037 yes | yes | yes | no
6038 Arguments : none
6039
6040 In certain environments, there are components which will regularly connect to
6041 various systems to ensure that they are still alive. It can be the case from
6042 another load balancer as well as from monitoring systems. By default, even a
6043 simple port probe or scan will produce a log. If those connections pollute
6044 the logs too much, it is possible to enable option "dontlognull" to indicate
6045 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02006046 which typically corresponds to those probes. Note that errors will still be
6047 returned to the client and accounted for in the stats. If this is not what is
6048 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006049
6050 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01006051 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006052 would not be logged.
6053
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
Willy Tarreau0f228a02015-05-01 15:37:53 +02006057 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
6058 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006059
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006060
Willy Tarreau87cf5142011-08-19 22:57:24 +02006061option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006062 Enable insertion of the X-Forwarded-For header to requests sent to servers
6063 May be used in sections : defaults | frontend | listen | backend
6064 yes | yes | yes | yes
6065 Arguments :
6066 <network> is an optional argument used to disable this option for sources
6067 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02006068 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01006069 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006070
6071 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
6072 their client address. This is sometimes annoying when the client's IP address
6073 is expected in server logs. To solve this problem, the well-known HTTP header
6074 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
6075 This header contains a value representing the client's IP address. Since this
6076 header is always appended at the end of the existing header list, the server
6077 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02006078 the server's manual to find how to enable use of this standard header. Note
6079 that only the last occurrence of the header must be used, since it is really
6080 possible that the client has already brought one.
6081
Willy Tarreaud72758d2010-01-12 10:42:19 +01006082 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02006083 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01006084 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01006085 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006086 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02006087 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01006088
6089 Sometimes, a same HAProxy instance may be shared between a direct client
6090 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6091 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6092 header for a known source address or network by adding the "except" keyword
6093 followed by the network address. In this case, any source IP matching the
6094 network will not cause an addition of this header. Most common uses are with
6095 private networks or 127.0.0.1.
6096
Willy Tarreau87cf5142011-08-19 22:57:24 +02006097 Alternatively, the keyword "if-none" states that the header will only be
6098 added if it is not present. This should only be used in perfectly trusted
6099 environment, as this might cause a security issue if headers reaching haproxy
6100 are under the control of the end-user.
6101
Willy Tarreauc27debf2008-01-06 08:57:02 +01006102 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02006103 least one of them uses it, the header will be added. Note that the backend's
6104 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02006105 both are defined. In the case of the "if-none" argument, if at least one of
6106 the frontend or the backend does not specify it, it wants the addition to be
6107 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006108
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006109 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01006110 # Public HTTP address also used by stunnel on the same machine
6111 frontend www
6112 mode http
6113 option forwardfor except 127.0.0.1 # stunnel already adds the header
6114
Ross Westaf72a1d2008-08-03 10:51:45 +02006115 # Those servers want the IP Address in X-Client
6116 backend www
6117 mode http
6118 option forwardfor header X-Client
6119
Willy Tarreau87cf5142011-08-19 22:57:24 +02006120 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006121 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01006122
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006123
Christopher Faulet98fbe952019-07-22 16:18:24 +02006124option h1-case-adjust-bogus-client
6125no option h1-case-adjust-bogus-client
6126 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
6127 May be used in sections : defaults | frontend | listen | backend
6128 yes | yes | yes | no
6129 Arguments : none
6130
6131 There is no standard case for header names because, as stated in RFC7230,
6132 they are case-insensitive. So applications must handle them in a case-
6133 insensitive manner. But some bogus applications violate the standards and
6134 erroneously rely on the cases most commonly used by browsers. This problem
6135 becomes critical with HTTP/2 because all header names must be exchanged in
6136 lower case, and HAProxy follows the same convention. All header names are
6137 sent in lower case to clients and servers, regardless of the HTTP version.
6138
6139 When HAProxy receives an HTTP/1 response, its header names are converted to
6140 lower case and manipulated and sent this way to the clients. If a client is
6141 known to violate the HTTP standards and to fail to process a response coming
6142 from HAProxy, it is possible to transform the lower case header names to a
6143 different format when the response is formatted and sent to the client, by
6144 enabling this option and specifying the list of headers to be reformatted
6145 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6146 must only be a temporary workaround for the time it takes the client to be
6147 fixed, because clients which require such workarounds might be vulnerable to
6148 content smuggling attacks and must absolutely be fixed.
6149
6150 Please note that this option will not affect standards-compliant clients.
6151
6152 If this option has been enabled in a "defaults" section, it can be disabled
6153 in a specific instance by prepending the "no" keyword before it.
6154
6155 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
6156 "h1-case-adjust-file".
6157
6158
6159option h1-case-adjust-bogus-server
6160no option h1-case-adjust-bogus-server
6161 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
6162 May be used in sections : defaults | frontend | listen | backend
6163 yes | no | yes | yes
6164 Arguments : none
6165
6166 There is no standard case for header names because, as stated in RFC7230,
6167 they are case-insensitive. So applications must handle them in a case-
6168 insensitive manner. But some bogus applications violate the standards and
6169 erroneously rely on the cases most commonly used by browsers. This problem
6170 becomes critical with HTTP/2 because all header names must be exchanged in
6171 lower case, and HAProxy follows the same convention. All header names are
6172 sent in lower case to clients and servers, regardless of the HTTP version.
6173
6174 When HAProxy receives an HTTP/1 request, its header names are converted to
6175 lower case and manipulated and sent this way to the servers. If a server is
6176 known to violate the HTTP standards and to fail to process a request coming
6177 from HAProxy, it is possible to transform the lower case header names to a
6178 different format when the request is formatted and sent to the server, by
6179 enabling this option and specifying the list of headers to be reformatted
6180 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6181 must only be a temporary workaround for the time it takes the server to be
6182 fixed, because servers which require such workarounds might be vulnerable to
6183 content smuggling attacks and must absolutely be fixed.
6184
6185 Please note that this option will not affect standards-compliant servers.
6186
6187 If this option has been enabled in a "defaults" section, it can be disabled
6188 in a specific instance by prepending the "no" keyword before it.
6189
6190 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
6191 "h1-case-adjust-file".
6192
6193
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006194option http-buffer-request
6195no option http-buffer-request
6196 Enable or disable waiting for whole HTTP request body before proceeding
6197 May be used in sections : defaults | frontend | listen | backend
6198 yes | yes | yes | yes
6199 Arguments : none
6200
6201 It is sometimes desirable to wait for the body of an HTTP request before
6202 taking a decision. This is what is being done by "balance url_param" for
6203 example. The first use case is to buffer requests from slow clients before
6204 connecting to the server. Another use case consists in taking the routing
6205 decision based on the request body's contents. This option placed in a
6206 frontend or backend forces the HTTP processing to wait until either the whole
6207 body is received, or the request buffer is full, or the first chunk is
6208 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01006209 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006210 the frontend and the backend, so this should definitely not be used by
6211 default.
6212
Baptiste Assmanneccdf432015-10-28 13:49:01 +01006213 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006214
6215
Willy Tarreau0f228a02015-05-01 15:37:53 +02006216option http-ignore-probes
6217no option http-ignore-probes
6218 Enable or disable logging of null connections and request timeouts
6219 May be used in sections : defaults | frontend | listen | backend
6220 yes | yes | yes | no
6221 Arguments : none
6222
6223 Recently some browsers started to implement a "pre-connect" feature
6224 consisting in speculatively connecting to some recently visited web sites
6225 just in case the user would like to visit them. This results in many
6226 connections being established to web sites, which end up in 408 Request
6227 Timeout if the timeout strikes first, or 400 Bad Request when the browser
6228 decides to close them first. These ones pollute the log and feed the error
6229 counters. There was already "option dontlognull" but it's insufficient in
6230 this case. Instead, this option does the following things :
6231 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01006232 was received over a connection before it was closed;
6233 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02006234 - prevent any error counter from being incremented
6235
6236 That way the empty connection is silently ignored. Note that it is better
6237 not to use this unless it is clear that it is needed, because it will hide
6238 real problems. The most common reason for not receiving a request and seeing
6239 a 408 is due to an MTU inconsistency between the client and an intermediary
6240 element such as a VPN, which blocks too large packets. These issues are
6241 generally seen with POST requests as well as GET with large cookies. The logs
6242 are often the only way to detect them.
6243
6244 If this option has been enabled in a "defaults" section, it can be disabled
6245 in a specific instance by prepending the "no" keyword before it.
6246
6247 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
6248
6249
Willy Tarreau16bfb022010-01-16 19:48:41 +01006250option http-keep-alive
6251no option http-keep-alive
6252 Enable or disable HTTP keep-alive from client to server
6253 May be used in sections : defaults | frontend | listen | backend
6254 yes | yes | yes | yes
6255 Arguments : none
6256
Willy Tarreau70dffda2014-01-30 03:07:23 +01006257 By default HAProxy operates in keep-alive mode with regards to persistent
6258 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02006259 leaves the connection idle on both sides between the end of a response and
6260 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006261 as "option http-server-close" or "option httpclose". This option allows to
6262 set back the keep-alive mode, which can be useful when another mode was used
6263 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01006264
6265 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
6266 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01006267 network) and the fastest session reuse on the server side at the expense
6268 of maintaining idle connections to the servers. In general, it is possible
6269 with this option to achieve approximately twice the request rate that the
6270 "http-server-close" option achieves on small objects. There are mainly two
6271 situations where this option may be useful :
6272
6273 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01006274 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01006275
6276 - when the cost of establishing the connection to the server is significant
6277 compared to the cost of retrieving the associated object from the server.
6278
6279 This last case can happen when the server is a fast static server of cache.
6280 In this case, the server will need to be properly tuned to support high enough
6281 connection counts because connections will last until the client sends another
6282 request.
6283
6284 If the client request has to go to another backend or another server due to
6285 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006286 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6287 available to try optimize server selection so that if the server currently
6288 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006289
Willy Tarreau16bfb022010-01-16 19:48:41 +01006290 At the moment, logs will not indicate whether requests came from the same
6291 session or not. The accept date reported in the logs corresponds to the end
6292 of the previous request, and the request time corresponds to the time spent
6293 waiting for a new request. The keep-alive request time is still bound to the
6294 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6295 not set.
6296
Christopher Faulet159e6672019-07-16 15:09:52 +02006297 This option disables and replaces any previous "option httpclose" or "option
6298 http-server-close". When backend and frontend options differ, all of these 4
6299 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006300
Christopher Faulet315b39c2018-09-21 16:26:19 +02006301 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006302 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01006303 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006304
6305
Willy Tarreau96e31212011-05-30 18:10:30 +02006306option http-no-delay
6307no option http-no-delay
6308 Instruct the system to favor low interactive delays over performance in HTTP
6309 May be used in sections : defaults | frontend | listen | backend
6310 yes | yes | yes | yes
6311 Arguments : none
6312
6313 In HTTP, each payload is unidirectional and has no notion of interactivity.
6314 Any agent is expected to queue data somewhat for a reasonably low delay.
6315 There are some very rare server-to-server applications that abuse the HTTP
6316 protocol and expect the payload phase to be highly interactive, with many
6317 interleaved data chunks in both directions within a single request. This is
6318 absolutely not supported by the HTTP specification and will not work across
6319 most proxies or servers. When such applications attempt to do this through
6320 haproxy, it works but they will experience high delays due to the network
6321 optimizations which favor performance by instructing the system to wait for
6322 enough data to be available in order to only send full packets. Typical
6323 delays are around 200 ms per round trip. Note that this only happens with
6324 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6325 affected.
6326
6327 When "option http-no-delay" is present in either the frontend or the backend
6328 used by a connection, all such optimizations will be disabled in order to
6329 make the exchanges as fast as possible. Of course this offers no guarantee on
6330 the functionality, as it may break at any other place. But if it works via
6331 HAProxy, it will work as fast as possible. This option should never be used
6332 by default, and should never be used at all unless such a buggy application
6333 is discovered. The impact of using this option is an increase of bandwidth
6334 usage and CPU usage, which may significantly lower performance in high
6335 latency environments.
6336
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006337 See also : "option http-buffer-request"
6338
Willy Tarreau96e31212011-05-30 18:10:30 +02006339
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006340option http-pretend-keepalive
6341no option http-pretend-keepalive
6342 Define whether haproxy will announce keepalive to the server or not
6343 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006344 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006345 Arguments : none
6346
Christopher Faulet315b39c2018-09-21 16:26:19 +02006347 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006348 adds a "Connection: close" header to the request forwarded to the server.
6349 Unfortunately, when some servers see this header, they automatically refrain
6350 from using the chunked encoding for responses of unknown length, while this
6351 is totally unrelated. The immediate effect is that this prevents haproxy from
6352 maintaining the client connection alive. A second effect is that a client or
6353 a cache could receive an incomplete response without being aware of it, and
6354 consider the response complete.
6355
6356 By setting "option http-pretend-keepalive", haproxy will make the server
6357 believe it will keep the connection alive. The server will then not fall back
6358 to the abnormal undesired above. When haproxy gets the whole response, it
6359 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006360 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006361 connection is correctly closed on the server side.
6362
6363 It is recommended not to enable this option by default, because most servers
6364 will more efficiently close the connection themselves after the last packet,
6365 and release its buffers slightly earlier. Also, the added packet on the
6366 network could slightly reduce the overall peak performance. However it is
6367 worth noting that when this option is enabled, haproxy will have slightly
6368 less work to do. So if haproxy is the bottleneck on the whole architecture,
6369 enabling this option might save a few CPU cycles.
6370
Christopher Faulet98db9762018-09-21 10:25:19 +02006371 This option may be set in backend and listen sections. Using it in a frontend
6372 section will be ignored and a warning will be reported during startup. It is
6373 a backend related option, so there is no real reason to set it on a
6374 frontend. This option may be combined with "option httpclose", which will
6375 cause keepalive to be announced to the server and close to be announced to
6376 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006377
6378 If this option has been enabled in a "defaults" section, it can be disabled
6379 in a specific instance by prepending the "no" keyword before it.
6380
Christopher Faulet315b39c2018-09-21 16:26:19 +02006381 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006382 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006383
Willy Tarreauc27debf2008-01-06 08:57:02 +01006384
Willy Tarreaub608feb2010-01-02 22:47:18 +01006385option http-server-close
6386no option http-server-close
6387 Enable or disable HTTP connection closing on the server side
6388 May be used in sections : defaults | frontend | listen | backend
6389 yes | yes | yes | yes
6390 Arguments : none
6391
Willy Tarreau70dffda2014-01-30 03:07:23 +01006392 By default HAProxy operates in keep-alive mode with regards to persistent
6393 connections: for each connection it processes each request and response, and
6394 leaves the connection idle on both sides between the end of a response and
6395 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006396 as "option http-server-close" or "option httpclose". Setting "option
6397 http-server-close" enables HTTP connection-close mode on the server side
6398 while keeping the ability to support HTTP keep-alive and pipelining on the
6399 client side. This provides the lowest latency on the client side (slow
6400 network) and the fastest session reuse on the server side to save server
6401 resources, similarly to "option httpclose". It also permits non-keepalive
6402 capable servers to be served in keep-alive mode to the clients if they
6403 conform to the requirements of RFC7230. Please note that some servers do not
6404 always conform to those requirements when they see "Connection: close" in the
6405 request. The effect will be that keep-alive will never be used. A workaround
6406 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006407
6408 At the moment, logs will not indicate whether requests came from the same
6409 session or not. The accept date reported in the logs corresponds to the end
6410 of the previous request, and the request time corresponds to the time spent
6411 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006412 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6413 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006414
6415 This option may be set both in a frontend and in a backend. It is enabled if
6416 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02006417 It disables and replaces any previous "option httpclose" or "option
6418 http-keep-alive". Please check section 4 ("Proxies") to see how this option
6419 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006420
6421 If this option has been enabled in a "defaults" section, it can be disabled
6422 in a specific instance by prepending the "no" keyword before it.
6423
Christopher Faulet315b39c2018-09-21 16:26:19 +02006424 See also : "option httpclose", "option http-pretend-keepalive",
6425 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006426
Willy Tarreau88d349d2010-01-25 12:15:43 +01006427option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006428no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006429 Make use of non-standard Proxy-Connection header instead of Connection
6430 May be used in sections : defaults | frontend | listen | backend
6431 yes | yes | yes | no
6432 Arguments : none
6433
Lukas Tribus23953682017-04-28 13:24:30 +00006434 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006435 Connection header to indicate their wish of persistent or non-persistent
6436 connections, both browsers and proxies ignore this header for proxied
6437 connections and make use of the undocumented, non-standard Proxy-Connection
6438 header instead. The issue begins when trying to put a load balancer between
6439 browsers and such proxies, because there will be a difference between what
6440 haproxy understands and what the client and the proxy agree on.
6441
6442 By setting this option in a frontend, haproxy can automatically switch to use
6443 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006444 defined here as one where the URI begins with neither a '/' nor a '*'. This
6445 is incompatible with the HTTP tunnel mode. Note that this option can only be
6446 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006447
Willy Tarreau844a7e72010-01-31 21:46:18 +01006448 Also, when this option is set, a request which requires authentication will
6449 automatically switch to use proxy authentication headers if it is itself a
6450 proxied request. That makes it possible to check or enforce authentication in
6451 front of an existing proxy.
6452
Willy Tarreau88d349d2010-01-25 12:15:43 +01006453 This option should normally never be used, except in front of a proxy.
6454
Christopher Faulet315b39c2018-09-21 16:26:19 +02006455 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006456
Willy Tarreaud63335a2010-02-26 12:56:52 +01006457option httpchk
6458option httpchk <uri>
6459option httpchk <method> <uri>
6460option httpchk <method> <uri> <version>
6461 Enable HTTP protocol to check on the servers health
6462 May be used in sections : defaults | frontend | listen | backend
6463 yes | no | yes | yes
6464 Arguments :
6465 <method> is the optional HTTP method used with the requests. When not set,
6466 the "OPTIONS" method is used, as it generally requires low server
6467 processing and is easy to filter out from the logs. Any method
6468 may be used, though it is not recommended to invent non-standard
6469 ones.
6470
6471 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6472 which is accessible by default on almost any server, but may be
6473 changed to any other URI. Query strings are permitted.
6474
6475 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6476 but some servers might behave incorrectly in HTTP 1.0, so turning
6477 it to HTTP/1.1 may sometimes help. Note that the Host field is
6478 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6479 after "\r\n" following the version string.
6480
6481 By default, server health checks only consist in trying to establish a TCP
6482 connection. When "option httpchk" is specified, a complete HTTP request is
6483 sent once the TCP connection is established, and responses 2xx and 3xx are
6484 considered valid, while all other ones indicate a server failure, including
6485 the lack of any response.
6486
6487 The port and interval are specified in the server configuration.
6488
6489 This option does not necessarily require an HTTP backend, it also works with
6490 plain TCP backends. This is particularly useful to check simple scripts bound
6491 to some dedicated ports using the inetd daemon.
6492
6493 Examples :
6494 # Relay HTTPS traffic to Apache instance and check service availability
6495 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6496 backend https_relay
6497 mode tcp
6498 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6499 server apache1 192.168.1.1:443 check port 80
6500
Simon Hormanafc47ee2013-11-25 10:46:35 +09006501 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6502 "option pgsql-check", "http-check" and the "check", "port" and
6503 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006504
6505
Willy Tarreauc27debf2008-01-06 08:57:02 +01006506option httpclose
6507no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006508 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006509 May be used in sections : defaults | frontend | listen | backend
6510 yes | yes | yes | yes
6511 Arguments : none
6512
Willy Tarreau70dffda2014-01-30 03:07:23 +01006513 By default HAProxy operates in keep-alive mode with regards to persistent
6514 connections: for each connection it processes each request and response, and
6515 leaves the connection idle on both sides between the end of a response and
6516 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006517 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006518
Christopher Faulet315b39c2018-09-21 16:26:19 +02006519 If "option httpclose" is set, HAProxy will close connections with the server
6520 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05006521 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02006522 and will add one if missing. Any "Connection" header different from "close"
6523 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006524
Christopher Faulet315b39c2018-09-21 16:26:19 +02006525 This option may also be combined with "option http-pretend-keepalive", which
6526 will disable sending of the "Connection: close" header, but will still cause
6527 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006528
6529 This option may be set both in a frontend and in a backend. It is enabled if
6530 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02006531 It disables and replaces any previous "option http-server-close" or "option
6532 http-keep-alive". Please check section 4 ("Proxies") to see how this option
6533 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006534
6535 If this option has been enabled in a "defaults" section, it can be disabled
6536 in a specific instance by prepending the "no" keyword before it.
6537
Christopher Faulet315b39c2018-09-21 16:26:19 +02006538 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006539
6540
Emeric Brun3a058f32009-06-30 18:26:00 +02006541option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006542 Enable logging of HTTP request, session state and timers
6543 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006544 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006545 Arguments :
6546 clf if the "clf" argument is added, then the output format will be
6547 the CLF format instead of HAProxy's default HTTP format. You can
6548 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006549 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006550 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006551
6552 By default, the log output format is very poor, as it only contains the
6553 source and destination addresses, and the instance name. By specifying
6554 "option httplog", each log line turns into a much richer format including,
6555 but not limited to, the HTTP request, the connection timers, the session
6556 status, the connections numbers, the captured headers and cookies, the
6557 frontend, backend and server name, and of course the source address and
6558 ports.
6559
PiBa-NLbd556bf2014-12-11 21:31:54 +01006560 Specifying only "option httplog" will automatically clear the 'clf' mode
6561 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006562
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006563 "option httplog" overrides any previous "log-format" directive.
6564
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006565 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006566
Willy Tarreau55165fe2009-05-10 12:02:55 +02006567
6568option http_proxy
6569no option http_proxy
6570 Enable or disable plain HTTP proxy mode
6571 May be used in sections : defaults | frontend | listen | backend
6572 yes | yes | yes | yes
6573 Arguments : none
6574
6575 It sometimes happens that people need a pure HTTP proxy which understands
6576 basic proxy requests without caching nor any fancy feature. In this case,
6577 it may be worth setting up an HAProxy instance with the "option http_proxy"
6578 set. In this mode, no server is declared, and the connection is forwarded to
6579 the IP address and port found in the URL after the "http://" scheme.
6580
6581 No host address resolution is performed, so this only works when pure IP
6582 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006583 it will probably be used only by experts who know they need exactly it. This
6584 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006585
6586 If this option has been enabled in a "defaults" section, it can be disabled
6587 in a specific instance by prepending the "no" keyword before it.
6588
6589 Example :
6590 # this backend understands HTTP proxy requests and forwards them directly.
6591 backend direct_forward
6592 option httpclose
6593 option http_proxy
6594
6595 See also : "option httpclose"
6596
Willy Tarreau211ad242009-10-03 21:45:07 +02006597
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006598option independent-streams
6599no option independent-streams
6600 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006601 May be used in sections : defaults | frontend | listen | backend
6602 yes | yes | yes | yes
6603 Arguments : none
6604
6605 By default, when data is sent over a socket, both the write timeout and the
6606 read timeout for that socket are refreshed, because we consider that there is
6607 activity on that socket, and we have no other means of guessing if we should
6608 receive data or not.
6609
Davor Ocelice9ed2812017-12-25 17:49:28 +01006610 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006611 exists a situation where it is desirable to disable it, and only refresh the
6612 read timeout if there are incoming data. This happens on sessions with large
6613 timeouts and low amounts of exchanged data such as telnet session. If the
6614 server suddenly disappears, the output data accumulates in the system's
6615 socket buffers, both timeouts are correctly refreshed, and there is no way
6616 to know the server does not receive them, so we don't timeout. However, when
6617 the underlying protocol always echoes sent data, it would be enough by itself
6618 to detect the issue using the read timeout. Note that this problem does not
6619 happen with more verbose protocols because data won't accumulate long in the
6620 socket buffers.
6621
6622 When this option is set on the frontend, it will disable read timeout updates
6623 on data sent to the client. There probably is little use of this case. When
6624 the option is set on the backend, it will disable read timeout updates on
6625 data sent to the server. Doing so will typically break large HTTP posts from
6626 slow lines, so use it with caution.
6627
Willy Tarreauce887fd2012-05-12 12:50:00 +02006628 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006629
6630
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006631option ldap-check
6632 Use LDAPv3 health checks for server testing
6633 May be used in sections : defaults | frontend | listen | backend
6634 yes | no | yes | yes
6635 Arguments : none
6636
6637 It is possible to test that the server correctly talks LDAPv3 instead of just
6638 testing that it accepts the TCP connection. When this option is set, an
6639 LDAPv3 anonymous simple bind message is sent to the server, and the response
6640 is analyzed to find an LDAPv3 bind response message.
6641
6642 The server is considered valid only when the LDAP response contains success
6643 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6644
6645 Logging of bind requests is server dependent see your documentation how to
6646 configure it.
6647
6648 Example :
6649 option ldap-check
6650
6651 See also : "option httpchk"
6652
6653
Simon Horman98637e52014-06-20 12:30:16 +09006654option external-check
6655 Use external processes for server health checks
6656 May be used in sections : defaults | frontend | listen | backend
6657 yes | no | yes | yes
6658
6659 It is possible to test the health of a server using an external command.
6660 This is achieved by running the executable set using "external-check
6661 command".
6662
6663 Requires the "external-check" global to be set.
6664
6665 See also : "external-check", "external-check command", "external-check path"
6666
6667
Willy Tarreau211ad242009-10-03 21:45:07 +02006668option log-health-checks
6669no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006670 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006671 May be used in sections : defaults | frontend | listen | backend
6672 yes | no | yes | yes
6673 Arguments : none
6674
Willy Tarreaubef1b322014-05-13 21:01:39 +02006675 By default, failed health check are logged if server is UP and successful
6676 health checks are logged if server is DOWN, so the amount of additional
6677 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006678
Willy Tarreaubef1b322014-05-13 21:01:39 +02006679 When this option is enabled, any change of the health check status or to
6680 the server's health will be logged, so that it becomes possible to know
6681 that a server was failing occasional checks before crashing, or exactly when
6682 it failed to respond a valid HTTP status, then when the port started to
6683 reject connections, then when the server stopped responding at all.
6684
Davor Ocelice9ed2812017-12-25 17:49:28 +01006685 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006686 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006687
Willy Tarreaubef1b322014-05-13 21:01:39 +02006688 See also: "option httpchk", "option ldap-check", "option mysql-check",
6689 "option pgsql-check", "option redis-check", "option smtpchk",
6690 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006691
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006692
6693option log-separate-errors
6694no option log-separate-errors
6695 Change log level for non-completely successful connections
6696 May be used in sections : defaults | frontend | listen | backend
6697 yes | yes | yes | no
6698 Arguments : none
6699
6700 Sometimes looking for errors in logs is not easy. This option makes haproxy
6701 raise the level of logs containing potentially interesting information such
6702 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6703 level changes from "info" to "err". This makes it possible to log them
6704 separately to a different file with most syslog daemons. Be careful not to
6705 remove them from the original file, otherwise you would lose ordering which
6706 provides very important information.
6707
6708 Using this option, large sites dealing with several thousand connections per
6709 second may log normal traffic to a rotating buffer and only archive smaller
6710 error logs.
6711
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006712 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006713 logging.
6714
Willy Tarreauc27debf2008-01-06 08:57:02 +01006715
6716option logasap
6717no option logasap
6718 Enable or disable early logging of HTTP requests
6719 May be used in sections : defaults | frontend | listen | backend
6720 yes | yes | yes | no
6721 Arguments : none
6722
6723 By default, HTTP requests are logged upon termination so that the total
6724 transfer time and the number of bytes appear in the logs. When large objects
6725 are being transferred, it may take a while before the request appears in the
6726 logs. Using "option logasap", the request gets logged as soon as the server
6727 sends the complete headers. The only missing information in the logs will be
6728 the total number of bytes which will indicate everything except the amount
6729 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006730 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006731 "Content-Length" response header so that the logs at least indicate how many
6732 bytes are expected to be transferred.
6733
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006734 Examples :
6735 listen http_proxy 0.0.0.0:80
6736 mode http
6737 option httplog
6738 option logasap
6739 log 192.168.2.200 local3
6740
6741 >>> Feb 6 12:14:14 localhost \
6742 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6743 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6744 "GET /image.iso HTTP/1.0"
6745
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006746 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006747 logging.
6748
6749
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006750option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006751 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006752 May be used in sections : defaults | frontend | listen | backend
6753 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006754 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006755 <username> This is the username which will be used when connecting to MySQL
6756 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006757 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006758
6759 If you specify a username, the check consists of sending two MySQL packet,
6760 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006761 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006762 Error packet. It is a basic but useful test which does not produce error nor
6763 aborted connect on the server. However, it requires adding an authorization
6764 in the MySQL table, like this :
6765
6766 USE mysql;
6767 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6768 FLUSH PRIVILEGES;
6769
6770 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006771 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006772 Error packet, we don't send anything in this mode. It was reported that it
6773 can generate lockout if check is too frequent and/or if there is not enough
6774 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6775 value as if a connection is established successfully within fewer than MySQL
6776 "max_connect_errors" attempts after a previous connection was interrupted,
6777 the error count for the host is cleared to zero. If HAProxy's server get
6778 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6779
6780 Remember that this does not check database presence nor database consistency.
6781 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006782
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006783 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006784
6785 Most often, an incoming MySQL server needs to see the client's IP address for
6786 various purposes, including IP privilege matching and connection logging.
6787 When possible, it is often wise to masquerade the client's IP address when
6788 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006789 which requires the transparent proxy feature to be compiled in, and the MySQL
6790 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006791
6792 See also: "option httpchk"
6793
6794
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006795option nolinger
6796no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006797 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006798 May be used in sections: defaults | frontend | listen | backend
6799 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006800 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006801
Davor Ocelice9ed2812017-12-25 17:49:28 +01006802 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006803 physically disconnected), the session timeouts triggers and the session is
6804 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6805 using some resources and possibly limiting the ability to establish newer
6806 connections.
6807
6808 When this happens, it is possible to activate "option nolinger" which forces
6809 the system to immediately remove any socket's pending data on close. Thus,
6810 the session is instantly purged from the system's tables. This usually has
6811 side effects such as increased number of TCP resets due to old retransmits
6812 getting immediately rejected. Some firewalls may sometimes complain about
6813 this too.
6814
6815 For this reason, it is not recommended to use this option when not absolutely
6816 needed. You know that you need it when you have thousands of FIN_WAIT1
6817 sessions on your system (TIME_WAIT ones do not count).
6818
6819 This option may be used both on frontends and backends, depending on the side
6820 where it is required. Use it on the frontend for clients, and on the backend
6821 for servers.
6822
6823 If this option has been enabled in a "defaults" section, it can be disabled
6824 in a specific instance by prepending the "no" keyword before it.
6825
6826
Willy Tarreau55165fe2009-05-10 12:02:55 +02006827option originalto [ except <network> ] [ header <name> ]
6828 Enable insertion of the X-Original-To header to requests sent to servers
6829 May be used in sections : defaults | frontend | listen | backend
6830 yes | yes | yes | yes
6831 Arguments :
6832 <network> is an optional argument used to disable this option for sources
6833 matching <network>
6834 <name> an optional argument to specify a different "X-Original-To"
6835 header name.
6836
6837 Since HAProxy can work in transparent mode, every request from a client can
6838 be redirected to the proxy and HAProxy itself can proxy every request to a
6839 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6840 be lost. This is annoying when you want access rules based on destination ip
6841 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6842 added by HAProxy to all requests sent to the server. This header contains a
6843 value representing the original destination IP address. Since this must be
6844 configured to always use the last occurrence of this header only. Note that
6845 only the last occurrence of the header must be used, since it is really
6846 possible that the client has already brought one.
6847
6848 The keyword "header" may be used to supply a different header name to replace
6849 the default "X-Original-To". This can be useful where you might already
6850 have a "X-Original-To" header from a different application, and you need
6851 preserve it. Also if your backend server doesn't use the "X-Original-To"
6852 header and requires different one.
6853
6854 Sometimes, a same HAProxy instance may be shared between a direct client
6855 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6856 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6857 header for a known source address or network by adding the "except" keyword
6858 followed by the network address. In this case, any source IP matching the
6859 network will not cause an addition of this header. Most common uses are with
6860 private networks or 127.0.0.1.
6861
6862 This option may be specified either in the frontend or in the backend. If at
6863 least one of them uses it, the header will be added. Note that the backend's
6864 setting of the header subargument takes precedence over the frontend's if
6865 both are defined.
6866
Willy Tarreau55165fe2009-05-10 12:02:55 +02006867 Examples :
6868 # Original Destination address
6869 frontend www
6870 mode http
6871 option originalto except 127.0.0.1
6872
6873 # Those servers want the IP Address in X-Client-Dst
6874 backend www
6875 mode http
6876 option originalto header X-Client-Dst
6877
Christopher Faulet315b39c2018-09-21 16:26:19 +02006878 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02006879
6880
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006881option persist
6882no option persist
6883 Enable or disable forced persistence on down servers
6884 May be used in sections: defaults | frontend | listen | backend
6885 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006886 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006887
6888 When an HTTP request reaches a backend with a cookie which references a dead
6889 server, by default it is redispatched to another server. It is possible to
6890 force the request to be sent to the dead server first using "option persist"
6891 if absolutely needed. A common use case is when servers are under extreme
6892 load and spend their time flapping. In this case, the users would still be
6893 directed to the server they opened the session on, in the hope they would be
6894 correctly served. It is recommended to use "option redispatch" in conjunction
6895 with this option so that in the event it would not be possible to connect to
6896 the server at all (server definitely dead), the client would finally be
6897 redirected to another valid server.
6898
6899 If this option has been enabled in a "defaults" section, it can be disabled
6900 in a specific instance by prepending the "no" keyword before it.
6901
Willy Tarreau4de91492010-01-22 19:10:05 +01006902 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006903
6904
Willy Tarreau0c122822013-12-15 18:49:01 +01006905option pgsql-check [ user <username> ]
6906 Use PostgreSQL health checks for server testing
6907 May be used in sections : defaults | frontend | listen | backend
6908 yes | no | yes | yes
6909 Arguments :
6910 <username> This is the username which will be used when connecting to
6911 PostgreSQL server.
6912
6913 The check sends a PostgreSQL StartupMessage and waits for either
6914 Authentication request or ErrorResponse message. It is a basic but useful
6915 test which does not produce error nor aborted connect on the server.
6916 This check is identical with the "mysql-check".
6917
6918 See also: "option httpchk"
6919
6920
Willy Tarreau9420b122013-12-15 18:58:25 +01006921option prefer-last-server
6922no option prefer-last-server
6923 Allow multiple load balanced requests to remain on the same server
6924 May be used in sections: defaults | frontend | listen | backend
6925 yes | no | yes | yes
6926 Arguments : none
6927
6928 When the load balancing algorithm in use is not deterministic, and a previous
6929 request was sent to a server to which haproxy still holds a connection, it is
6930 sometimes desirable that subsequent requests on a same session go to the same
6931 server as much as possible. Note that this is different from persistence, as
6932 we only indicate a preference which haproxy tries to apply without any form
6933 of warranty. The real use is for keep-alive connections sent to servers. When
6934 this option is used, haproxy will try to reuse the same connection that is
6935 attached to the server instead of rebalancing to another server, causing a
6936 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006937 not make much sense to use this in combination with hashing algorithms. Note,
6938 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02006939 to a proxy which sends a 407 (authentication required), when the load
6940 balancing algorithm is not deterministic. This is mandatory for use with the
6941 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01006942 troubleshooting some faulty applications. Option prefer-last-server might be
6943 desirable in these environments as well, to avoid redistributing the traffic
6944 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006945
6946 If this option has been enabled in a "defaults" section, it can be disabled
6947 in a specific instance by prepending the "no" keyword before it.
6948
6949 See also: "option http-keep-alive"
6950
6951
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006952option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006953option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006954no option redispatch
6955 Enable or disable session redistribution in case of connection failure
6956 May be used in sections: defaults | frontend | listen | backend
6957 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006958 Arguments :
6959 <interval> The optional integer value that controls how often redispatches
6960 occur when retrying connections. Positive value P indicates a
6961 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006962 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006963 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006964 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006965 positive value of 1 would indicate a redispatch on every retry,
6966 and a positive value of 3 would indicate a redispatch on every
6967 third retry. You can disable redispatches with a value of 0.
6968
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006969
6970 In HTTP mode, if a server designated by a cookie is down, clients may
6971 definitely stick to it because they cannot flush the cookie, so they will not
6972 be able to access the service anymore.
6973
Willy Tarreau59884a62019-01-02 14:48:31 +01006974 Specifying "option redispatch" will allow the proxy to break cookie or
6975 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006976
Joseph Lynch726ab712015-05-11 23:25:34 -07006977 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006978 connection failures. Of course, it requires having "retries" set to a nonzero
6979 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006980
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006981 If this option has been enabled in a "defaults" section, it can be disabled
6982 in a specific instance by prepending the "no" keyword before it.
6983
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006984 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006985
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006986
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006987option redis-check
6988 Use redis health checks for server testing
6989 May be used in sections : defaults | frontend | listen | backend
6990 yes | no | yes | yes
6991 Arguments : none
6992
6993 It is possible to test that the server correctly talks REDIS protocol instead
6994 of just testing that it accepts the TCP connection. When this option is set,
6995 a PING redis command is sent to the server, and the response is analyzed to
6996 find the "+PONG" response message.
6997
6998 Example :
6999 option redis-check
7000
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007001 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007002
7003
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007004option smtpchk
7005option smtpchk <hello> <domain>
7006 Use SMTP health checks for server testing
7007 May be used in sections : defaults | frontend | listen | backend
7008 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01007009 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007010 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02007011 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007012 values will be turned into the default command ("HELO").
7013
7014 <domain> is the domain name to present to the server. It may only be
7015 specified (and is mandatory) if the hello command has been
7016 specified. By default, "localhost" is used.
7017
7018 When "option smtpchk" is set, the health checks will consist in TCP
7019 connections followed by an SMTP command. By default, this command is
7020 "HELO localhost". The server's return code is analyzed and only return codes
7021 starting with a "2" will be considered as valid. All other responses,
7022 including a lack of response will constitute an error and will indicate a
7023 dead server.
7024
7025 This test is meant to be used with SMTP servers or relays. Depending on the
7026 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01007027 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007028 25 is often easier than adjusting the configuration.
7029
7030 Most often, an incoming SMTP server needs to see the client's IP address for
7031 various purposes, including spam filtering, anti-spoofing and logging. When
7032 possible, it is often wise to masquerade the client's IP address when
7033 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007034 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007035
7036 Example :
7037 option smtpchk HELO mydomain.org
7038
7039 See also : "option httpchk", "source"
7040
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007041
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02007042option socket-stats
7043no option socket-stats
7044
7045 Enable or disable collecting & providing separate statistics for each socket.
7046 May be used in sections : defaults | frontend | listen | backend
7047 yes | yes | yes | no
7048
7049 Arguments : none
7050
7051
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007052option splice-auto
7053no option splice-auto
7054 Enable or disable automatic kernel acceleration on sockets in both directions
7055 May be used in sections : defaults | frontend | listen | backend
7056 yes | yes | yes | yes
7057 Arguments : none
7058
7059 When this option is enabled either on a frontend or on a backend, haproxy
7060 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007061 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007062 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007063 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007064 are not much aggressive in order to limit excessive use of splicing. This
7065 option requires splicing to be enabled at compile time, and may be globally
7066 disabled with the global option "nosplice". Since splice uses pipes, using it
7067 requires that there are enough spare pipes.
7068
7069 Important note: kernel-based TCP splicing is a Linux-specific feature which
7070 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
7071 transfer data between sockets without copying these data to user-space, thus
7072 providing noticeable performance gains and CPU cycles savings. Since many
7073 early implementations are buggy, corrupt data and/or are inefficient, this
7074 feature is not enabled by default, and it should be used with extreme care.
7075 While it is not possible to detect the correctness of an implementation,
7076 2.6.29 is the first version offering a properly working implementation. In
7077 case of doubt, splicing may be globally disabled using the global "nosplice"
7078 keyword.
7079
7080 Example :
7081 option splice-auto
7082
7083 If this option has been enabled in a "defaults" section, it can be disabled
7084 in a specific instance by prepending the "no" keyword before it.
7085
7086 See also : "option splice-request", "option splice-response", and global
7087 options "nosplice" and "maxpipes"
7088
7089
7090option splice-request
7091no option splice-request
7092 Enable or disable automatic kernel acceleration on sockets for requests
7093 May be used in sections : defaults | frontend | listen | backend
7094 yes | yes | yes | yes
7095 Arguments : none
7096
7097 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007098 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007099 the client to the server. It might still use the recv/send scheme if there
7100 are no spare pipes left. This option requires splicing to be enabled at
7101 compile time, and may be globally disabled with the global option "nosplice".
7102 Since splice uses pipes, using it requires that there are enough spare pipes.
7103
7104 Important note: see "option splice-auto" for usage limitations.
7105
7106 Example :
7107 option splice-request
7108
7109 If this option has been enabled in a "defaults" section, it can be disabled
7110 in a specific instance by prepending the "no" keyword before it.
7111
7112 See also : "option splice-auto", "option splice-response", and global options
7113 "nosplice" and "maxpipes"
7114
7115
7116option splice-response
7117no option splice-response
7118 Enable or disable automatic kernel acceleration on sockets for responses
7119 May be used in sections : defaults | frontend | listen | backend
7120 yes | yes | yes | yes
7121 Arguments : none
7122
7123 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007124 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007125 the server to the client. It might still use the recv/send scheme if there
7126 are no spare pipes left. This option requires splicing to be enabled at
7127 compile time, and may be globally disabled with the global option "nosplice".
7128 Since splice uses pipes, using it requires that there are enough spare pipes.
7129
7130 Important note: see "option splice-auto" for usage limitations.
7131
7132 Example :
7133 option splice-response
7134
7135 If this option has been enabled in a "defaults" section, it can be disabled
7136 in a specific instance by prepending the "no" keyword before it.
7137
7138 See also : "option splice-auto", "option splice-request", and global options
7139 "nosplice" and "maxpipes"
7140
7141
Christopher Fauletba7bc162016-11-07 21:07:38 +01007142option spop-check
7143 Use SPOP health checks for server testing
7144 May be used in sections : defaults | frontend | listen | backend
7145 no | no | no | yes
7146 Arguments : none
7147
7148 It is possible to test that the server correctly talks SPOP protocol instead
7149 of just testing that it accepts the TCP connection. When this option is set,
7150 a HELLO handshake is performed between HAProxy and the server, and the
7151 response is analyzed to check no error is reported.
7152
7153 Example :
7154 option spop-check
7155
7156 See also : "option httpchk"
7157
7158
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007159option srvtcpka
7160no option srvtcpka
7161 Enable or disable the sending of TCP keepalive packets on the server side
7162 May be used in sections : defaults | frontend | listen | backend
7163 yes | no | yes | yes
7164 Arguments : none
7165
7166 When there is a firewall or any session-aware component between a client and
7167 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007168 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007169 components decides to expire a session which has remained idle for too long.
7170
7171 Enabling socket-level TCP keep-alives makes the system regularly send packets
7172 to the other end of the connection, leaving it active. The delay between
7173 keep-alive probes is controlled by the system only and depends both on the
7174 operating system and its tuning parameters.
7175
7176 It is important to understand that keep-alive packets are neither emitted nor
7177 received at the application level. It is only the network stacks which sees
7178 them. For this reason, even if one side of the proxy already uses keep-alives
7179 to maintain its connection alive, those keep-alive packets will not be
7180 forwarded to the other side of the proxy.
7181
7182 Please note that this has nothing to do with HTTP keep-alive.
7183
7184 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
7185 server side of a connection, which should help when session expirations are
7186 noticed between HAProxy and a server.
7187
7188 If this option has been enabled in a "defaults" section, it can be disabled
7189 in a specific instance by prepending the "no" keyword before it.
7190
7191 See also : "option clitcpka", "option tcpka"
7192
7193
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007194option ssl-hello-chk
7195 Use SSLv3 client hello health checks for server testing
7196 May be used in sections : defaults | frontend | listen | backend
7197 yes | no | yes | yes
7198 Arguments : none
7199
7200 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
7201 possible to test that the server correctly talks SSL instead of just testing
7202 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
7203 SSLv3 client hello messages are sent once the connection is established to
7204 the server, and the response is analyzed to find an SSL server hello message.
7205 The server is considered valid only when the response contains this server
7206 hello message.
7207
7208 All servers tested till there correctly reply to SSLv3 client hello messages,
7209 and most servers tested do not even log the requests containing only hello
7210 messages, which is appreciable.
7211
Willy Tarreau763a95b2012-10-04 23:15:39 +02007212 Note that this check works even when SSL support was not built into haproxy
7213 because it forges the SSL message. When SSL support is available, it is best
7214 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007215
Willy Tarreau763a95b2012-10-04 23:15:39 +02007216 See also: "option httpchk", "check-ssl"
7217
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007218
Willy Tarreaued179852013-12-16 01:07:00 +01007219option tcp-check
7220 Perform health checks using tcp-check send/expect sequences
7221 May be used in sections: defaults | frontend | listen | backend
7222 yes | no | yes | yes
7223
7224 This health check method is intended to be combined with "tcp-check" command
7225 lists in order to support send/expect types of health check sequences.
7226
7227 TCP checks currently support 4 modes of operations :
7228 - no "tcp-check" directive : the health check only consists in a connection
7229 attempt, which remains the default mode.
7230
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007231 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007232 used to send a string along with a connection opening. With some
7233 protocols, it helps sending a "QUIT" message for example that prevents
7234 the server from logging a connection error for each health check. The
7235 check result will still be based on the ability to open the connection
7236 only.
7237
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007238 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007239 The connection is opened and haproxy waits for the server to present some
7240 contents which must validate some rules. The check result will be based
7241 on the matching between the contents and the rules. This is suited for
7242 POP, IMAP, SMTP, FTP, SSH, TELNET.
7243
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007244 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007245 used to test a hello-type protocol. HAProxy sends a message, the server
7246 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007247 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007248 suited for protocols which require a binding or a request/response model.
7249 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7250 already all have their dedicated checks with a deeper understanding of
7251 the respective protocols.
7252 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007253 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007254
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007255 A fifth mode can be used to insert comments in different steps of the
7256 script.
7257
7258 For each tcp-check rule you create, you can add a "comment" directive,
7259 followed by a string. This string will be reported in the log and stderr
7260 in debug mode. It is useful to make user-friendly error reporting.
7261 The "comment" is of course optional.
7262
7263
Willy Tarreaued179852013-12-16 01:07:00 +01007264 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007265 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007266 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007267 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007268
Davor Ocelice9ed2812017-12-25 17:49:28 +01007269 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007270 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007271 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007272
7273 # look for the redis master server after ensuring it speaks well
7274 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007275 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007276 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007277 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007278 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007279 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007280 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007281 tcp-check send info\ replication\r\n
7282 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007283 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007284 tcp-check send QUIT\r\n
7285 tcp-check expect string +OK
7286
Davor Ocelice9ed2812017-12-25 17:49:28 +01007287 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007288 (send many headers before analyzing)
7289 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007290 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007291 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7292 tcp-check send Host:\ www.mydomain.com\r\n
7293 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7294 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007295 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007296
7297
7298 See also : "tcp-check expect", "tcp-check send"
7299
7300
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007301option tcp-smart-accept
7302no option tcp-smart-accept
7303 Enable or disable the saving of one ACK packet during the accept sequence
7304 May be used in sections : defaults | frontend | listen | backend
7305 yes | yes | yes | no
7306 Arguments : none
7307
7308 When an HTTP connection request comes in, the system acknowledges it on
7309 behalf of HAProxy, then the client immediately sends its request, and the
7310 system acknowledges it too while it is notifying HAProxy about the new
7311 connection. HAProxy then reads the request and responds. This means that we
7312 have one TCP ACK sent by the system for nothing, because the request could
7313 very well be acknowledged by HAProxy when it sends its response.
7314
7315 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7316 sending this useless ACK on platforms which support it (currently at least
7317 Linux). It must not cause any problem, because the system will send it anyway
7318 after 40 ms if the response takes more time than expected to come.
7319
7320 During complex network debugging sessions, it may be desirable to disable
7321 this optimization because delayed ACKs can make troubleshooting more complex
7322 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007323 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007324
7325 It is also possible to force it for non-HTTP proxies by simply specifying
7326 "option tcp-smart-accept". For instance, it can make sense with some services
7327 such as SMTP where the server speaks first.
7328
7329 It is recommended to avoid forcing this option in a defaults section. In case
7330 of doubt, consider setting it back to automatic values by prepending the
7331 "default" keyword before it, or disabling it using the "no" keyword.
7332
Willy Tarreaud88edf22009-06-14 15:48:17 +02007333 See also : "option tcp-smart-connect"
7334
7335
7336option tcp-smart-connect
7337no option tcp-smart-connect
7338 Enable or disable the saving of one ACK packet during the connect sequence
7339 May be used in sections : defaults | frontend | listen | backend
7340 yes | no | yes | yes
7341 Arguments : none
7342
7343 On certain systems (at least Linux), HAProxy can ask the kernel not to
7344 immediately send an empty ACK upon a connection request, but to directly
7345 send the buffer request instead. This saves one packet on the network and
7346 thus boosts performance. It can also be useful for some servers, because they
7347 immediately get the request along with the incoming connection.
7348
7349 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7350 It is not enabled by default because it makes network troubleshooting more
7351 complex.
7352
7353 It only makes sense to enable it with protocols where the client speaks first
7354 such as HTTP. In other situations, if there is no data to send in place of
7355 the ACK, a normal ACK is sent.
7356
7357 If this option has been enabled in a "defaults" section, it can be disabled
7358 in a specific instance by prepending the "no" keyword before it.
7359
7360 See also : "option tcp-smart-accept"
7361
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007362
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007363option tcpka
7364 Enable or disable the sending of TCP keepalive packets on both sides
7365 May be used in sections : defaults | frontend | listen | backend
7366 yes | yes | yes | yes
7367 Arguments : none
7368
7369 When there is a firewall or any session-aware component between a client and
7370 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007371 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007372 components decides to expire a session which has remained idle for too long.
7373
7374 Enabling socket-level TCP keep-alives makes the system regularly send packets
7375 to the other end of the connection, leaving it active. The delay between
7376 keep-alive probes is controlled by the system only and depends both on the
7377 operating system and its tuning parameters.
7378
7379 It is important to understand that keep-alive packets are neither emitted nor
7380 received at the application level. It is only the network stacks which sees
7381 them. For this reason, even if one side of the proxy already uses keep-alives
7382 to maintain its connection alive, those keep-alive packets will not be
7383 forwarded to the other side of the proxy.
7384
7385 Please note that this has nothing to do with HTTP keep-alive.
7386
7387 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7388 the client and server sides of a connection. Note that this is meaningful
7389 only in "defaults" or "listen" sections. If this option is used in a
7390 frontend, only the client side will get keep-alives, and if this option is
7391 used in a backend, only the server side will get keep-alives. For this
7392 reason, it is strongly recommended to explicitly use "option clitcpka" and
7393 "option srvtcpka" when the configuration is split between frontends and
7394 backends.
7395
7396 See also : "option clitcpka", "option srvtcpka"
7397
Willy Tarreau844e3c52008-01-11 16:28:18 +01007398
7399option tcplog
7400 Enable advanced logging of TCP connections with session state and timers
7401 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007402 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007403 Arguments : none
7404
7405 By default, the log output format is very poor, as it only contains the
7406 source and destination addresses, and the instance name. By specifying
7407 "option tcplog", each log line turns into a much richer format including, but
7408 not limited to, the connection timers, the session status, the connections
7409 numbers, the frontend, backend and server name, and of course the source
7410 address and ports. This option is useful for pure TCP proxies in order to
7411 find which of the client or server disconnects or times out. For normal HTTP
7412 proxies, it's better to use "option httplog" which is even more complete.
7413
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007414 "option tcplog" overrides any previous "log-format" directive.
7415
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007416 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007417
7418
Willy Tarreau844e3c52008-01-11 16:28:18 +01007419option transparent
7420no option transparent
7421 Enable client-side transparent proxying
7422 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007423 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007424 Arguments : none
7425
7426 This option was introduced in order to provide layer 7 persistence to layer 3
7427 load balancers. The idea is to use the OS's ability to redirect an incoming
7428 connection for a remote address to a local process (here HAProxy), and let
7429 this process know what address was initially requested. When this option is
7430 used, sessions without cookies will be forwarded to the original destination
7431 IP address of the incoming request (which should match that of another
7432 equipment), while requests with cookies will still be forwarded to the
7433 appropriate server.
7434
7435 Note that contrary to a common belief, this option does NOT make HAProxy
7436 present the client's IP to the server when establishing the connection.
7437
Willy Tarreaua1146052011-03-01 09:51:54 +01007438 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007439 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007440
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007441
Simon Horman98637e52014-06-20 12:30:16 +09007442external-check command <command>
7443 Executable to run when performing an external-check
7444 May be used in sections : defaults | frontend | listen | backend
7445 yes | no | yes | yes
7446
7447 Arguments :
7448 <command> is the external command to run
7449
Simon Horman98637e52014-06-20 12:30:16 +09007450 The arguments passed to the to the command are:
7451
Cyril Bonté777be862014-12-02 21:21:35 +01007452 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007453
Cyril Bonté777be862014-12-02 21:21:35 +01007454 The <proxy_address> and <proxy_port> are derived from the first listener
7455 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7456 listener the proxy_address will be the path of the socket and the
7457 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7458 possible to determine a listener, and both <proxy_address> and <proxy_port>
7459 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007460
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007461 Some values are also provided through environment variables.
7462
7463 Environment variables :
7464 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7465 applicable, for example in a "backend" section).
7466
7467 HAPROXY_PROXY_ID The backend id.
7468
7469 HAPROXY_PROXY_NAME The backend name.
7470
7471 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7472 applicable, for example in a "backend" section or
7473 for a UNIX socket).
7474
7475 HAPROXY_SERVER_ADDR The server address.
7476
7477 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7478
7479 HAPROXY_SERVER_ID The server id.
7480
7481 HAPROXY_SERVER_MAXCONN The server max connections.
7482
7483 HAPROXY_SERVER_NAME The server name.
7484
7485 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7486 socket).
7487
7488 PATH The PATH environment variable used when executing
7489 the command may be set using "external-check path".
7490
William Lallemand4d03e432019-06-14 15:35:37 +02007491 See also "2.3. Environment variables" for other variables.
7492
Simon Horman98637e52014-06-20 12:30:16 +09007493 If the command executed and exits with a zero status then the check is
7494 considered to have passed, otherwise the check is considered to have
7495 failed.
7496
7497 Example :
7498 external-check command /bin/true
7499
7500 See also : "external-check", "option external-check", "external-check path"
7501
7502
7503external-check path <path>
7504 The value of the PATH environment variable used when running an external-check
7505 May be used in sections : defaults | frontend | listen | backend
7506 yes | no | yes | yes
7507
7508 Arguments :
7509 <path> is the path used when executing external command to run
7510
7511 The default path is "".
7512
7513 Example :
7514 external-check path "/usr/bin:/bin"
7515
7516 See also : "external-check", "option external-check",
7517 "external-check command"
7518
7519
Emeric Brun647caf12009-06-30 17:57:00 +02007520persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007521persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007522 Enable RDP cookie-based persistence
7523 May be used in sections : defaults | frontend | listen | backend
7524 yes | no | yes | yes
7525 Arguments :
7526 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007527 default cookie name "msts" will be used. There currently is no
7528 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007529
7530 This statement enables persistence based on an RDP cookie. The RDP cookie
7531 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007532 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007533 and if an RDP cookie is found, it is decoded. If it matches a known server
7534 which is still UP (or if "option persist" is set), then the connection is
7535 forwarded to this server.
7536
7537 Note that this only makes sense in a TCP backend, but for this to work, the
7538 frontend must have waited long enough to ensure that an RDP cookie is present
7539 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007540 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007541 a single "listen" section.
7542
Willy Tarreau61e28f22010-05-16 22:31:05 +02007543 Also, it is important to understand that the terminal server will emit this
7544 RDP cookie only if it is configured for "token redirection mode", which means
7545 that the "IP address redirection" option is disabled.
7546
Emeric Brun647caf12009-06-30 17:57:00 +02007547 Example :
7548 listen tse-farm
7549 bind :3389
7550 # wait up to 5s for an RDP cookie in the request
7551 tcp-request inspect-delay 5s
7552 tcp-request content accept if RDP_COOKIE
7553 # apply RDP cookie persistence
7554 persist rdp-cookie
7555 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007556 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007557 balance rdp-cookie
7558 server srv1 1.1.1.1:3389
7559 server srv2 1.1.1.2:3389
7560
Simon Hormanab814e02011-06-24 14:50:20 +09007561 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7562 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007563
7564
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007565rate-limit sessions <rate>
7566 Set a limit on the number of new sessions accepted per second on a frontend
7567 May be used in sections : defaults | frontend | listen | backend
7568 yes | yes | yes | no
7569 Arguments :
7570 <rate> The <rate> parameter is an integer designating the maximum number
7571 of new sessions per second to accept on the frontend.
7572
7573 When the frontend reaches the specified number of new sessions per second, it
7574 stops accepting new connections until the rate drops below the limit again.
7575 During this time, the pending sessions will be kept in the socket's backlog
7576 (in system buffers) and haproxy will not even be aware that sessions are
7577 pending. When applying very low limit on a highly loaded service, it may make
7578 sense to increase the socket's backlog using the "backlog" keyword.
7579
7580 This feature is particularly efficient at blocking connection-based attacks
7581 or service abuse on fragile servers. Since the session rate is measured every
7582 millisecond, it is extremely accurate. Also, the limit applies immediately,
7583 no delay is needed at all to detect the threshold.
7584
7585 Example : limit the connection rate on SMTP to 10 per second max
7586 listen smtp
7587 mode tcp
7588 bind :25
7589 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007590 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007591
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007592 Note : when the maximum rate is reached, the frontend's status is not changed
7593 but its sockets appear as "WAITING" in the statistics if the
7594 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007595
7596 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7597
7598
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007599redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7600redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7601redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007602 Return an HTTP redirection if/unless a condition is matched
7603 May be used in sections : defaults | frontend | listen | backend
7604 no | yes | yes | yes
7605
7606 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007607 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007608
Willy Tarreau0140f252008-11-19 21:07:09 +01007609 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007610 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007611 the HTTP "Location" header. When used in an "http-request" rule,
7612 <loc> value follows the log-format rules and can include some
7613 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007614
7615 <pfx> With "redirect prefix", the "Location" header is built from the
7616 concatenation of <pfx> and the complete URI path, including the
7617 query string, unless the "drop-query" option is specified (see
7618 below). As a special case, if <pfx> equals exactly "/", then
7619 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007620 redirect to the same URL (for instance, to insert a cookie). When
7621 used in an "http-request" rule, <pfx> value follows the log-format
7622 rules and can include some dynamic values (see Custom Log Format
7623 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007624
7625 <sch> With "redirect scheme", then the "Location" header is built by
7626 concatenating <sch> with "://" then the first occurrence of the
7627 "Host" header, and then the URI path, including the query string
7628 unless the "drop-query" option is specified (see below). If no
7629 path is found or if the path is "*", then "/" is used instead. If
7630 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007631 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007632 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007633 HTTPS. When used in an "http-request" rule, <sch> value follows
7634 the log-format rules and can include some dynamic values (see
7635 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007636
7637 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007638 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7639 with 302 used by default if no code is specified. 301 means
7640 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007641 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007642 cache the redirection. 303 is equivalent to 302 except that the
7643 browser will fetch the location with a GET method. 307 is just
7644 like 302 but makes it clear that the same method must be reused.
7645 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007646
7647 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007648 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007649
7650 - "drop-query"
7651 When this keyword is used in a prefix-based redirection, then the
7652 location will be set without any possible query-string, which is useful
7653 for directing users to a non-secure page for instance. It has no effect
7654 with a location-type redirect.
7655
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007656 - "append-slash"
7657 This keyword may be used in conjunction with "drop-query" to redirect
7658 users who use a URL not ending with a '/' to the same one with the '/'.
7659 It can be useful to ensure that search engines will only see one URL.
7660 For this, a return code 301 is preferred.
7661
Willy Tarreau0140f252008-11-19 21:07:09 +01007662 - "set-cookie NAME[=value]"
7663 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7664 to the response. This is sometimes used to indicate that a user has
7665 been seen, for instance to protect against some types of DoS. No other
7666 cookie option is added, so the cookie will be a session cookie. Note
7667 that for a browser, a sole cookie name without an equal sign is
7668 different from a cookie with an equal sign.
7669
7670 - "clear-cookie NAME[=]"
7671 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7672 with the "Max-Age" attribute set to zero. This will tell the browser to
7673 delete this cookie. It is useful for instance on logout pages. It is
7674 important to note that clearing the cookie "NAME" will not remove a
7675 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7676 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007677
7678 Example: move the login URL only to HTTPS.
7679 acl clear dst_port 80
7680 acl secure dst_port 8080
7681 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007682 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007683 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007684 acl cookie_set hdr_sub(cookie) SEEN=1
7685
7686 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007687 redirect prefix https://mysite.com if login_page !secure
7688 redirect prefix http://mysite.com drop-query if login_page !uid_given
7689 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007690 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007691
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007692 Example: send redirects for request for articles without a '/'.
7693 acl missing_slash path_reg ^/article/[^/]*$
7694 redirect code 301 prefix / drop-query append-slash if missing_slash
7695
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007696 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007697 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007698
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007699 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007700 http-request redirect code 301 location \
7701 http://www.%[hdr(host)]%[capture.req.uri] \
7702 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007703
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007704 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007705
Willy Tarreau303c0352008-01-17 19:01:39 +01007706
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007707retries <value>
7708 Set the number of retries to perform on a server after a connection failure
7709 May be used in sections: defaults | frontend | listen | backend
7710 yes | no | yes | yes
7711 Arguments :
7712 <value> is the number of times a connection attempt should be retried on
7713 a server when a connection either is refused or times out. The
7714 default value is 3.
7715
7716 It is important to understand that this value applies to the number of
7717 connection attempts, not full requests. When a connection has effectively
7718 been established to a server, there will be no more retry.
7719
7720 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007721 a turn-around timer of min("timeout connect", one second) is applied before
7722 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007723
7724 When "option redispatch" is set, the last retry may be performed on another
7725 server even if a cookie references a different server.
7726
7727 See also : "option redispatch"
7728
7729
Olivier Houcharda254a372019-04-05 15:30:12 +02007730retry-on [list of keywords]
7731 Specify when to attempt to automatically retry a failed request
7732 May be used in sections: defaults | frontend | listen | backend
7733 yes | no | yes | yes
7734 Arguments :
7735 <keywords> is a list of keywords or HTTP status codes, each representing a
7736 type of failure event on which an attempt to retry the request
7737 is desired. Please read the notes at the bottom before changing
7738 this setting. The following keywords are supported :
7739
7740 none never retry
7741
7742 conn-failure retry when the connection or the SSL handshake failed
7743 and the request could not be sent. This is the default.
7744
7745 empty-response retry when the server connection was closed after part
7746 of the request was sent, and nothing was received from
7747 the server. This type of failure may be caused by the
7748 request timeout on the server side, poor network
7749 condition, or a server crash or restart while
7750 processing the request.
7751
Olivier Houcharde3249a92019-05-03 23:01:47 +02007752 junk-response retry when the server returned something not looking
7753 like a complete HTTP response. This includes partial
7754 responses headers as well as non-HTTP contents. It
7755 usually is a bad idea to retry on such events, which
7756 may be caused a configuration issue (wrong server port)
7757 or by the request being harmful to the server (buffer
7758 overflow attack for example).
7759
Olivier Houcharda254a372019-04-05 15:30:12 +02007760 response-timeout the server timeout stroke while waiting for the server
7761 to respond to the request. This may be caused by poor
7762 network condition, the reuse of an idle connection
7763 which has expired on the path, or by the request being
7764 extremely expensive to process. It generally is a bad
7765 idea to retry on such events on servers dealing with
7766 heavy database processing (full scans, etc) as it may
7767 amplify denial of service attacks.
7768
Olivier Houchard865d8392019-05-03 22:46:27 +02007769 0rtt-rejected retry requests which were sent over early data and were
7770 rejected by the server. These requests are generally
7771 considered to be safe to retry.
7772
Olivier Houcharda254a372019-04-05 15:30:12 +02007773 <status> any HTTP status code among "404" (Not Found), "408"
7774 (Request Timeout), "425" (Too Early), "500" (Server
7775 Error), "501" (Not Implemented), "502" (Bad Gateway),
7776 "503" (Service Unavailable), "504" (Gateway Timeout).
7777
Olivier Houchardddf0e032019-05-10 18:05:40 +02007778 all-retryable-errors
7779 retry request for any error that are considered
7780 retryable. This currently activates "conn-failure",
7781 "empty-response", "junk-response", "response-timeout",
7782 "0rtt-rejected", "500", "502", "503", and "504".
7783
Olivier Houcharda254a372019-04-05 15:30:12 +02007784 Using this directive replaces any previous settings with the new ones; it is
7785 not cumulative.
7786
7787 Please note that using anything other than "none" and "conn-failure" requires
7788 to allocate a buffer and copy the whole request into it, so it has memory and
7789 performance impacts. Requests not fitting in a single buffer will never be
7790 retried (see the global tune.bufsize setting).
7791
7792 You have to make sure the application has a replay protection mechanism built
7793 in such as a unique transaction IDs passed in requests, or that replaying the
7794 same request has no consequence, or it is very dangerous to use any retry-on
7795 value beside "conn-failure" and "none". Static file servers and caches are
7796 generally considered safe against any type of retry. Using a status code can
7797 be useful to quickly leave a server showing an abnormal behavior (out of
7798 memory, file system issues, etc), but in this case it may be a good idea to
7799 immediately redispatch the connection to another server (please see "option
7800 redispatch" for this). Last, it is important to understand that most causes
7801 of failures are the requests themselves and that retrying a request causing a
7802 server to misbehave will often make the situation even worse for this server,
7803 or for the whole service in case of redispatch.
7804
7805 Unless you know exactly how the application deals with replayed requests, you
7806 should not use this directive.
7807
7808 The default is "conn-failure".
7809
7810 See also: "retries", "option redispatch", "tune.bufsize"
7811
David du Colombier486df472011-03-17 10:40:26 +01007812server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007813 Declare a server in a backend
7814 May be used in sections : defaults | frontend | listen | backend
7815 no | no | yes | yes
7816 Arguments :
7817 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007818 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007819 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007820
David du Colombier486df472011-03-17 10:40:26 +01007821 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7822 resolvable hostname is supported, but this name will be resolved
7823 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7824 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007825 address as the one from the client connection. This is useful in
7826 transparent proxy architectures where the client's connection is
7827 intercepted and haproxy must forward to the original destination
7828 address. This is more or less what the "transparent" keyword does
7829 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007830 to report statistics. Optionally, an address family prefix may be
7831 used before the address to force the family regardless of the
7832 address format, which can be useful to specify a path to a unix
7833 socket with no slash ('/'). Currently supported prefixes are :
7834 - 'ipv4@' -> address is always IPv4
7835 - 'ipv6@' -> address is always IPv6
7836 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007837 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02007838 - 'sockpair@' -> address is the FD of a connected unix
7839 socket or of a socketpair. During a connection, the
7840 backend creates a pair of connected sockets, and passes
7841 one of them over the FD. The bind part will use the
7842 received socket as the client FD. Should be used
7843 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02007844 You may want to reference some environment variables in the
7845 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007846 variables. The "init-addr" setting can be used to modify the way
7847 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007848
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007849 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007850 be sent to this port. If unset, the same port the client
7851 connected to will be used. The port may also be prefixed by a "+"
7852 or a "-". In this case, the server's port will be determined by
7853 adding this value to the client's port.
7854
7855 <param*> is a list of parameters for this server. The "server" keywords
7856 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007857 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007858
7859 Examples :
7860 server first 10.1.1.1:1080 cookie first check inter 1000
7861 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007862 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007863 server backup "${SRV_BACKUP}:1080" backup
7864 server www1_dc1 "${LAN_DC1}.101:80"
7865 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007866
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007867 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7868 sun_path length is used for the address length. Some other programs
7869 such as socat use the string length only by default. Pass the option
7870 ",unix-tightsocklen=0" to any abstract socket definition in socat to
7871 make it compatible with HAProxy's.
7872
Mark Lamourinec2247f02012-01-04 13:02:01 -05007873 See also: "default-server", "http-send-name-header" and section 5 about
7874 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007875
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007876server-state-file-name [<file>]
7877 Set the server state file to read, load and apply to servers available in
7878 this backend. It only applies when the directive "load-server-state-from-file"
7879 is set to "local". When <file> is not provided or if this directive is not
7880 set, then backend name is used. If <file> starts with a slash '/', then it is
7881 considered as an absolute path. Otherwise, <file> is concatenated to the
7882 global directive "server-state-file-base".
7883
7884 Example: the minimal configuration below would make HAProxy look for the
7885 state server file '/etc/haproxy/states/bk':
7886
7887 global
7888 server-state-file-base /etc/haproxy/states
7889
7890 backend bk
7891 load-server-state-from-file
7892
7893 See also: "server-state-file-base", "load-server-state-from-file", and
7894 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007895
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02007896server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
7897 Set a template to initialize servers with shared parameters.
7898 The names of these servers are built from <prefix> and <num | range> parameters.
7899 May be used in sections : defaults | frontend | listen | backend
7900 no | no | yes | yes
7901
7902 Arguments:
7903 <prefix> A prefix for the server names to be built.
7904
7905 <num | range>
7906 If <num> is provided, this template initializes <num> servers
7907 with 1 up to <num> as server name suffixes. A range of numbers
7908 <num_low>-<num_high> may also be used to use <num_low> up to
7909 <num_high> as server name suffixes.
7910
7911 <fqdn> A FQDN for all the servers this template initializes.
7912
7913 <port> Same meaning as "server" <port> argument (see "server" keyword).
7914
7915 <params*>
7916 Remaining server parameters among all those supported by "server"
7917 keyword.
7918
7919 Examples:
7920 # Initializes 3 servers with srv1, srv2 and srv3 as names,
7921 # google.com as FQDN, and health-check enabled.
7922 server-template srv 1-3 google.com:80 check
7923
7924 # or
7925 server-template srv 3 google.com:80 check
7926
7927 # would be equivalent to:
7928 server srv1 google.com:80 check
7929 server srv2 google.com:80 check
7930 server srv3 google.com:80 check
7931
7932
7933
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007934source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02007935source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007936source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007937 Set the source address for outgoing connections
7938 May be used in sections : defaults | frontend | listen | backend
7939 yes | no | yes | yes
7940 Arguments :
7941 <addr> is the IPv4 address HAProxy will bind to before connecting to a
7942 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01007943
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007944 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01007945 the most appropriate address to reach its destination. Optionally
7946 an address family prefix may be used before the address to force
7947 the family regardless of the address format, which can be useful
7948 to specify a path to a unix socket with no slash ('/'). Currently
7949 supported prefixes are :
7950 - 'ipv4@' -> address is always IPv4
7951 - 'ipv6@' -> address is always IPv6
7952 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007953 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02007954 You may want to reference some environment variables in the
7955 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007956
7957 <port> is an optional port. It is normally not needed but may be useful
7958 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02007959 the system will select a free port. Note that port ranges are not
7960 supported in the backend. If you want to force port ranges, you
7961 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007962
7963 <addr2> is the IP address to present to the server when connections are
7964 forwarded in full transparent proxy mode. This is currently only
7965 supported on some patched Linux kernels. When this address is
7966 specified, clients connecting to the server will be presented
7967 with this address, while health checks will still use the address
7968 <addr>.
7969
7970 <port2> is the optional port to present to the server when connections
7971 are forwarded in full transparent proxy mode (see <addr2> above).
7972 The default value of zero means the system will select a free
7973 port.
7974
Willy Tarreaubce70882009-09-07 11:51:47 +02007975 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
7976 This is the name of a comma-separated header list which can
7977 contain multiple IP addresses. By default, the last occurrence is
7978 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01007979 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02007980 by previous proxy, typically Stunnel. In order to use another
7981 occurrence from the last one, please see the <occ> parameter
7982 below. When the header (or occurrence) is not found, no binding
7983 is performed so that the proxy's default IP address is used. Also
7984 keep in mind that the header name is case insensitive, as for any
7985 HTTP header.
7986
7987 <occ> is the occurrence number of a value to be used in a multi-value
7988 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007989 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02007990 address. Positive values indicate a position from the first
7991 occurrence, 1 being the first one. Negative values indicate
7992 positions relative to the last one, -1 being the last one. This
7993 is helpful for situations where an X-Forwarded-For header is set
7994 at the entry point of an infrastructure and must be used several
7995 proxy layers away. When this value is not specified, -1 is
7996 assumed. Passing a zero here disables the feature.
7997
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007998 <name> is an optional interface name to which to bind to for outgoing
7999 traffic. On systems supporting this features (currently, only
8000 Linux), this allows one to bind all traffic to the server to
8001 this interface even if it is not the one the system would select
8002 based on routing tables. This should be used with extreme care.
8003 Note that using this option requires root privileges.
8004
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008005 The "source" keyword is useful in complex environments where a specific
8006 address only is allowed to connect to the servers. It may be needed when a
8007 private address must be used through a public gateway for instance, and it is
8008 known that the system cannot determine the adequate source address by itself.
8009
8010 An extension which is available on certain patched Linux kernels may be used
8011 through the "usesrc" optional keyword. It makes it possible to connect to the
8012 servers with an IP address which does not belong to the system itself. This
8013 is called "full transparent proxy mode". For this to work, the destination
8014 servers have to route their traffic back to this address through the machine
8015 running HAProxy, and IP forwarding must generally be enabled on this machine.
8016
8017 In this "full transparent proxy" mode, it is possible to force a specific IP
8018 address to be presented to the servers. This is not much used in fact. A more
8019 common use is to tell HAProxy to present the client's IP address. For this,
8020 there are two methods :
8021
8022 - present the client's IP and port addresses. This is the most transparent
8023 mode, but it can cause problems when IP connection tracking is enabled on
8024 the machine, because a same connection may be seen twice with different
8025 states. However, this solution presents the huge advantage of not
8026 limiting the system to the 64k outgoing address+port couples, because all
8027 of the client ranges may be used.
8028
8029 - present only the client's IP address and select a spare port. This
8030 solution is still quite elegant but slightly less transparent (downstream
8031 firewalls logs will not match upstream's). It also presents the downside
8032 of limiting the number of concurrent connections to the usual 64k ports.
8033 However, since the upstream and downstream ports are different, local IP
8034 connection tracking on the machine will not be upset by the reuse of the
8035 same session.
8036
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008037 This option sets the default source for all servers in the backend. It may
8038 also be specified in a "defaults" section. Finer source address specification
8039 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008040 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008041
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008042 In order to work, "usesrc" requires root privileges.
8043
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008044 Examples :
8045 backend private
8046 # Connect to the servers using our 192.168.1.200 source address
8047 source 192.168.1.200
8048
8049 backend transparent_ssl1
8050 # Connect to the SSL farm from the client's source address
8051 source 192.168.1.200 usesrc clientip
8052
8053 backend transparent_ssl2
8054 # Connect to the SSL farm from the client's source address and port
8055 # not recommended if IP conntrack is present on the local machine.
8056 source 192.168.1.200 usesrc client
8057
8058 backend transparent_ssl3
8059 # Connect to the SSL farm from the client's source address. It
8060 # is more conntrack-friendly.
8061 source 192.168.1.200 usesrc clientip
8062
8063 backend transparent_smtp
8064 # Connect to the SMTP farm from the client's source address/port
8065 # with Tproxy version 4.
8066 source 0.0.0.0 usesrc clientip
8067
Willy Tarreaubce70882009-09-07 11:51:47 +02008068 backend transparent_http
8069 # Connect to the servers using the client's IP as seen by previous
8070 # proxy.
8071 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8072
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008073 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008074 the Linux kernel on www.balabit.com, the "bind" keyword.
8075
Willy Tarreau844e3c52008-01-11 16:28:18 +01008076
Cyril Bonté66c327d2010-10-12 00:14:37 +02008077stats admin { if | unless } <cond>
8078 Enable statistics admin level if/unless a condition is matched
8079 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008080 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008081
8082 This statement enables the statistics admin level if/unless a condition is
8083 matched.
8084
8085 The admin level allows to enable/disable servers from the web interface. By
8086 default, statistics page is read-only for security reasons.
8087
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008088 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8089 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008090 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008091
Cyril Bonté23b39d92011-02-10 22:54:44 +01008092 Currently, the POST request is limited to the buffer size minus the reserved
8093 buffer space, which means that if the list of servers is too long, the
8094 request won't be processed. It is recommended to alter few servers at a
8095 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008096
8097 Example :
8098 # statistics admin level only for localhost
8099 backend stats_localhost
8100 stats enable
8101 stats admin if LOCALHOST
8102
8103 Example :
8104 # statistics admin level always enabled because of the authentication
8105 backend stats_auth
8106 stats enable
8107 stats auth admin:AdMiN123
8108 stats admin if TRUE
8109
8110 Example :
8111 # statistics admin level depends on the authenticated user
8112 userlist stats-auth
8113 group admin users admin
8114 user admin insecure-password AdMiN123
8115 group readonly users haproxy
8116 user haproxy insecure-password haproxy
8117
8118 backend stats_auth
8119 stats enable
8120 acl AUTH http_auth(stats-auth)
8121 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8122 stats http-request auth unless AUTH
8123 stats admin if AUTH_ADMIN
8124
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008125 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8126 "bind-process", section 3.4 about userlists and section 7 about
8127 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008128
8129
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008130stats auth <user>:<passwd>
8131 Enable statistics with authentication and grant access to an account
8132 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008133 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008134 Arguments :
8135 <user> is a user name to grant access to
8136
8137 <passwd> is the cleartext password associated to this user
8138
8139 This statement enables statistics with default settings, and restricts access
8140 to declared users only. It may be repeated as many times as necessary to
8141 allow as many users as desired. When a user tries to access the statistics
8142 without a valid account, a "401 Forbidden" response will be returned so that
8143 the browser asks the user to provide a valid user and password. The real
8144 which will be returned to the browser is configurable using "stats realm".
8145
8146 Since the authentication method is HTTP Basic Authentication, the passwords
8147 circulate in cleartext on the network. Thus, it was decided that the
8148 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008149 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008150
8151 It is also possible to reduce the scope of the proxies which appear in the
8152 report using "stats scope".
8153
8154 Though this statement alone is enough to enable statistics reporting, it is
8155 recommended to set all other settings in order to avoid relying on default
8156 unobvious parameters.
8157
8158 Example :
8159 # public access (limited to this backend only)
8160 backend public_www
8161 server srv1 192.168.0.1:80
8162 stats enable
8163 stats hide-version
8164 stats scope .
8165 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008166 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008167 stats auth admin1:AdMiN123
8168 stats auth admin2:AdMiN321
8169
8170 # internal monitoring access (unlimited)
8171 backend private_monitoring
8172 stats enable
8173 stats uri /admin?stats
8174 stats refresh 5s
8175
8176 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8177
8178
8179stats enable
8180 Enable statistics reporting with default settings
8181 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008182 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008183 Arguments : none
8184
8185 This statement enables statistics reporting with default settings defined
8186 at build time. Unless stated otherwise, these settings are used :
8187 - stats uri : /haproxy?stats
8188 - stats realm : "HAProxy Statistics"
8189 - stats auth : no authentication
8190 - stats scope : no restriction
8191
8192 Though this statement alone is enough to enable statistics reporting, it is
8193 recommended to set all other settings in order to avoid relying on default
8194 unobvious parameters.
8195
8196 Example :
8197 # public access (limited to this backend only)
8198 backend public_www
8199 server srv1 192.168.0.1:80
8200 stats enable
8201 stats hide-version
8202 stats scope .
8203 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008204 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008205 stats auth admin1:AdMiN123
8206 stats auth admin2:AdMiN321
8207
8208 # internal monitoring access (unlimited)
8209 backend private_monitoring
8210 stats enable
8211 stats uri /admin?stats
8212 stats refresh 5s
8213
8214 See also : "stats auth", "stats realm", "stats uri"
8215
8216
Willy Tarreaud63335a2010-02-26 12:56:52 +01008217stats hide-version
8218 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008219 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008220 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008221 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008222
Willy Tarreaud63335a2010-02-26 12:56:52 +01008223 By default, the stats page reports some useful status information along with
8224 the statistics. Among them is HAProxy's version. However, it is generally
8225 considered dangerous to report precise version to anyone, as it can help them
8226 target known weaknesses with specific attacks. The "stats hide-version"
8227 statement removes the version from the statistics report. This is recommended
8228 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008229
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008230 Though this statement alone is enough to enable statistics reporting, it is
8231 recommended to set all other settings in order to avoid relying on default
8232 unobvious parameters.
8233
Willy Tarreaud63335a2010-02-26 12:56:52 +01008234 Example :
8235 # public access (limited to this backend only)
8236 backend public_www
8237 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008238 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008239 stats hide-version
8240 stats scope .
8241 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008242 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008243 stats auth admin1:AdMiN123
8244 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008245
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008246 # internal monitoring access (unlimited)
8247 backend private_monitoring
8248 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008249 stats uri /admin?stats
8250 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008251
Willy Tarreaud63335a2010-02-26 12:56:52 +01008252 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008253
Willy Tarreau983e01e2010-01-11 18:42:06 +01008254
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008255stats http-request { allow | deny | auth [realm <realm>] }
8256 [ { if | unless } <condition> ]
8257 Access control for statistics
8258
8259 May be used in sections: defaults | frontend | listen | backend
8260 no | no | yes | yes
8261
8262 As "http-request", these set of options allow to fine control access to
8263 statistics. Each option may be followed by if/unless and acl.
8264 First option with matched condition (or option without condition) is final.
8265 For "deny" a 403 error will be returned, for "allow" normal processing is
8266 performed, for "auth" a 401/407 error code is returned so the client
8267 should be asked to enter a username and password.
8268
8269 There is no fixed limit to the number of http-request statements per
8270 instance.
8271
8272 See also : "http-request", section 3.4 about userlists and section 7
8273 about ACL usage.
8274
8275
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008276stats realm <realm>
8277 Enable statistics and set authentication realm
8278 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008279 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008280 Arguments :
8281 <realm> is the name of the HTTP Basic Authentication realm reported to
8282 the browser. The browser uses it to display it in the pop-up
8283 inviting the user to enter a valid username and password.
8284
8285 The realm is read as a single word, so any spaces in it should be escaped
8286 using a backslash ('\').
8287
8288 This statement is useful only in conjunction with "stats auth" since it is
8289 only related to authentication.
8290
8291 Though this statement alone is enough to enable statistics reporting, it is
8292 recommended to set all other settings in order to avoid relying on default
8293 unobvious parameters.
8294
8295 Example :
8296 # public access (limited to this backend only)
8297 backend public_www
8298 server srv1 192.168.0.1:80
8299 stats enable
8300 stats hide-version
8301 stats scope .
8302 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008303 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008304 stats auth admin1:AdMiN123
8305 stats auth admin2:AdMiN321
8306
8307 # internal monitoring access (unlimited)
8308 backend private_monitoring
8309 stats enable
8310 stats uri /admin?stats
8311 stats refresh 5s
8312
8313 See also : "stats auth", "stats enable", "stats uri"
8314
8315
8316stats refresh <delay>
8317 Enable statistics with automatic refresh
8318 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008319 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008320 Arguments :
8321 <delay> is the suggested refresh delay, specified in seconds, which will
8322 be returned to the browser consulting the report page. While the
8323 browser is free to apply any delay, it will generally respect it
8324 and refresh the page this every seconds. The refresh interval may
8325 be specified in any other non-default time unit, by suffixing the
8326 unit after the value, as explained at the top of this document.
8327
8328 This statement is useful on monitoring displays with a permanent page
8329 reporting the load balancer's activity. When set, the HTML report page will
8330 include a link "refresh"/"stop refresh" so that the user can select whether
8331 he wants automatic refresh of the page or not.
8332
8333 Though this statement alone is enough to enable statistics reporting, it is
8334 recommended to set all other settings in order to avoid relying on default
8335 unobvious parameters.
8336
8337 Example :
8338 # public access (limited to this backend only)
8339 backend public_www
8340 server srv1 192.168.0.1:80
8341 stats enable
8342 stats hide-version
8343 stats scope .
8344 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008345 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008346 stats auth admin1:AdMiN123
8347 stats auth admin2:AdMiN321
8348
8349 # internal monitoring access (unlimited)
8350 backend private_monitoring
8351 stats enable
8352 stats uri /admin?stats
8353 stats refresh 5s
8354
8355 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8356
8357
8358stats scope { <name> | "." }
8359 Enable statistics and limit access scope
8360 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008361 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008362 Arguments :
8363 <name> is the name of a listen, frontend or backend section to be
8364 reported. The special name "." (a single dot) designates the
8365 section in which the statement appears.
8366
8367 When this statement is specified, only the sections enumerated with this
8368 statement will appear in the report. All other ones will be hidden. This
8369 statement may appear as many times as needed if multiple sections need to be
8370 reported. Please note that the name checking is performed as simple string
8371 comparisons, and that it is never checked that a give section name really
8372 exists.
8373
8374 Though this statement alone is enough to enable statistics reporting, it is
8375 recommended to set all other settings in order to avoid relying on default
8376 unobvious parameters.
8377
8378 Example :
8379 # public access (limited to this backend only)
8380 backend public_www
8381 server srv1 192.168.0.1:80
8382 stats enable
8383 stats hide-version
8384 stats scope .
8385 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008386 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008387 stats auth admin1:AdMiN123
8388 stats auth admin2:AdMiN321
8389
8390 # internal monitoring access (unlimited)
8391 backend private_monitoring
8392 stats enable
8393 stats uri /admin?stats
8394 stats refresh 5s
8395
8396 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8397
Willy Tarreaud63335a2010-02-26 12:56:52 +01008398
Willy Tarreauc9705a12010-07-27 20:05:50 +02008399stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008400 Enable reporting of a description on the statistics page.
8401 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008402 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008403
Willy Tarreauc9705a12010-07-27 20:05:50 +02008404 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008405 description from global section is automatically used instead.
8406
8407 This statement is useful for users that offer shared services to their
8408 customers, where node or description should be different for each customer.
8409
8410 Though this statement alone is enough to enable statistics reporting, it is
8411 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008412 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008413
8414 Example :
8415 # internal monitoring access (unlimited)
8416 backend private_monitoring
8417 stats enable
8418 stats show-desc Master node for Europe, Asia, Africa
8419 stats uri /admin?stats
8420 stats refresh 5s
8421
8422 See also: "show-node", "stats enable", "stats uri" and "description" in
8423 global section.
8424
8425
8426stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008427 Enable reporting additional information on the statistics page
8428 May be used in sections : defaults | frontend | listen | backend
8429 yes | yes | yes | yes
8430 Arguments : none
8431
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008432 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008433 - cap: capabilities (proxy)
8434 - mode: one of tcp, http or health (proxy)
8435 - id: SNMP ID (proxy, socket, server)
8436 - IP (socket, server)
8437 - cookie (backend, server)
8438
8439 Though this statement alone is enough to enable statistics reporting, it is
8440 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008441 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008442
8443 See also: "stats enable", "stats uri".
8444
8445
8446stats show-node [ <name> ]
8447 Enable reporting of a host name on the statistics page.
8448 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008449 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008450 Arguments:
8451 <name> is an optional name to be reported. If unspecified, the
8452 node name from global section is automatically used instead.
8453
8454 This statement is useful for users that offer shared services to their
8455 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008456 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008457
8458 Though this statement alone is enough to enable statistics reporting, it is
8459 recommended to set all other settings in order to avoid relying on default
8460 unobvious parameters.
8461
8462 Example:
8463 # internal monitoring access (unlimited)
8464 backend private_monitoring
8465 stats enable
8466 stats show-node Europe-1
8467 stats uri /admin?stats
8468 stats refresh 5s
8469
8470 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8471 section.
8472
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008473
8474stats uri <prefix>
8475 Enable statistics and define the URI prefix to access them
8476 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008477 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008478 Arguments :
8479 <prefix> is the prefix of any URI which will be redirected to stats. This
8480 prefix may contain a question mark ('?') to indicate part of a
8481 query string.
8482
8483 The statistics URI is intercepted on the relayed traffic, so it appears as a
8484 page within the normal application. It is strongly advised to ensure that the
8485 selected URI will never appear in the application, otherwise it will never be
8486 possible to reach it in the application.
8487
8488 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008489 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008490 It is generally a good idea to include a question mark in the URI so that
8491 intermediate proxies refrain from caching the results. Also, since any string
8492 beginning with the prefix will be accepted as a stats request, the question
8493 mark helps ensuring that no valid URI will begin with the same words.
8494
8495 It is sometimes very convenient to use "/" as the URI prefix, and put that
8496 statement in a "listen" instance of its own. That makes it easy to dedicate
8497 an address or a port to statistics only.
8498
8499 Though this statement alone is enough to enable statistics reporting, it is
8500 recommended to set all other settings in order to avoid relying on default
8501 unobvious parameters.
8502
8503 Example :
8504 # public access (limited to this backend only)
8505 backend public_www
8506 server srv1 192.168.0.1:80
8507 stats enable
8508 stats hide-version
8509 stats scope .
8510 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008511 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008512 stats auth admin1:AdMiN123
8513 stats auth admin2:AdMiN321
8514
8515 # internal monitoring access (unlimited)
8516 backend private_monitoring
8517 stats enable
8518 stats uri /admin?stats
8519 stats refresh 5s
8520
8521 See also : "stats auth", "stats enable", "stats realm"
8522
8523
Willy Tarreaud63335a2010-02-26 12:56:52 +01008524stick match <pattern> [table <table>] [{if | unless} <cond>]
8525 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008526 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008527 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008528
8529 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008530 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008531 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008532 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008533 stickiness table. This rule is mandatory.
8534
8535 <table> is an optional stickiness table name. If unspecified, the same
8536 backend's table is used. A stickiness table is declared using
8537 the "stick-table" statement.
8538
8539 <cond> is an optional matching condition. It makes it possible to match
8540 on a certain criterion only when other conditions are met (or
8541 not met). For instance, it could be used to match on a source IP
8542 address except when a request passes through a known proxy, in
8543 which case we'd match on a header containing that IP address.
8544
8545 Some protocols or applications require complex stickiness rules and cannot
8546 always simply rely on cookies nor hashing. The "stick match" statement
8547 describes a rule to extract the stickiness criterion from an incoming request
8548 or connection. See section 7 for a complete list of possible patterns and
8549 transformation rules.
8550
8551 The table has to be declared using the "stick-table" statement. It must be of
8552 a type compatible with the pattern. By default it is the one which is present
8553 in the same backend. It is possible to share a table with other backends by
8554 referencing it using the "table" keyword. If another table is referenced,
8555 the server's ID inside the backends are used. By default, all server IDs
8556 start at 1 in each backend, so the server ordering is enough. But in case of
8557 doubt, it is highly recommended to force server IDs using their "id" setting.
8558
8559 It is possible to restrict the conditions where a "stick match" statement
8560 will apply, using "if" or "unless" followed by a condition. See section 7 for
8561 ACL based conditions.
8562
8563 There is no limit on the number of "stick match" statements. The first that
8564 applies and matches will cause the request to be directed to the same server
8565 as was used for the request which created the entry. That way, multiple
8566 matches can be used as fallbacks.
8567
8568 The stick rules are checked after the persistence cookies, so they will not
8569 affect stickiness if a cookie has already been used to select a server. That
8570 way, it becomes very easy to insert cookies and match on IP addresses in
8571 order to maintain stickiness between HTTP and HTTPS.
8572
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008573 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8574 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008575 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008576
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008577 Example :
8578 # forward SMTP users to the same server they just used for POP in the
8579 # last 30 minutes
8580 backend pop
8581 mode tcp
8582 balance roundrobin
8583 stick store-request src
8584 stick-table type ip size 200k expire 30m
8585 server s1 192.168.1.1:110
8586 server s2 192.168.1.1:110
8587
8588 backend smtp
8589 mode tcp
8590 balance roundrobin
8591 stick match src table pop
8592 server s1 192.168.1.1:25
8593 server s2 192.168.1.1:25
8594
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008595 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008596 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008597
8598
8599stick on <pattern> [table <table>] [{if | unless} <condition>]
8600 Define a request pattern to associate a user to a server
8601 May be used in sections : defaults | frontend | listen | backend
8602 no | no | yes | yes
8603
8604 Note : This form is exactly equivalent to "stick match" followed by
8605 "stick store-request", all with the same arguments. Please refer
8606 to both keywords for details. It is only provided as a convenience
8607 for writing more maintainable configurations.
8608
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008609 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8610 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008611 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008612
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008613 Examples :
8614 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008615 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008616
8617 # ...is strictly equivalent to this one :
8618 stick match src table pop if !localhost
8619 stick store-request src table pop if !localhost
8620
8621
8622 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8623 # well as HTTP without cookie. Share the same table between both accesses.
8624 backend http
8625 mode http
8626 balance roundrobin
8627 stick on src table https
8628 cookie SRV insert indirect nocache
8629 server s1 192.168.1.1:80 cookie s1
8630 server s2 192.168.1.1:80 cookie s2
8631
8632 backend https
8633 mode tcp
8634 balance roundrobin
8635 stick-table type ip size 200k expire 30m
8636 stick on src
8637 server s1 192.168.1.1:443
8638 server s2 192.168.1.1:443
8639
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008640 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008641
8642
8643stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8644 Define a request pattern used to create an entry in a stickiness table
8645 May be used in sections : defaults | frontend | listen | backend
8646 no | no | yes | yes
8647
8648 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008649 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008650 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008651 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008652 server is selected.
8653
8654 <table> is an optional stickiness table name. If unspecified, the same
8655 backend's table is used. A stickiness table is declared using
8656 the "stick-table" statement.
8657
8658 <cond> is an optional storage condition. It makes it possible to store
8659 certain criteria only when some conditions are met (or not met).
8660 For instance, it could be used to store the source IP address
8661 except when the request passes through a known proxy, in which
8662 case we'd store a converted form of a header containing that IP
8663 address.
8664
8665 Some protocols or applications require complex stickiness rules and cannot
8666 always simply rely on cookies nor hashing. The "stick store-request" statement
8667 describes a rule to decide what to extract from the request and when to do
8668 it, in order to store it into a stickiness table for further requests to
8669 match it using the "stick match" statement. Obviously the extracted part must
8670 make sense and have a chance to be matched in a further request. Storing a
8671 client's IP address for instance often makes sense. Storing an ID found in a
8672 URL parameter also makes sense. Storing a source port will almost never make
8673 any sense because it will be randomly matched. See section 7 for a complete
8674 list of possible patterns and transformation rules.
8675
8676 The table has to be declared using the "stick-table" statement. It must be of
8677 a type compatible with the pattern. By default it is the one which is present
8678 in the same backend. It is possible to share a table with other backends by
8679 referencing it using the "table" keyword. If another table is referenced,
8680 the server's ID inside the backends are used. By default, all server IDs
8681 start at 1 in each backend, so the server ordering is enough. But in case of
8682 doubt, it is highly recommended to force server IDs using their "id" setting.
8683
8684 It is possible to restrict the conditions where a "stick store-request"
8685 statement will apply, using "if" or "unless" followed by a condition. This
8686 condition will be evaluated while parsing the request, so any criteria can be
8687 used. See section 7 for ACL based conditions.
8688
8689 There is no limit on the number of "stick store-request" statements, but
8690 there is a limit of 8 simultaneous stores per request or response. This
8691 makes it possible to store up to 8 criteria, all extracted from either the
8692 request or the response, regardless of the number of rules. Only the 8 first
8693 ones which match will be kept. Using this, it is possible to feed multiple
8694 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008695 another protocol or access method. Using multiple store-request rules with
8696 the same table is possible and may be used to find the best criterion to rely
8697 on, by arranging the rules by decreasing preference order. Only the first
8698 extracted criterion for a given table will be stored. All subsequent store-
8699 request rules referencing the same table will be skipped and their ACLs will
8700 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008701
8702 The "store-request" rules are evaluated once the server connection has been
8703 established, so that the table will contain the real server that processed
8704 the request.
8705
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008706 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8707 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008708 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008709
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008710 Example :
8711 # forward SMTP users to the same server they just used for POP in the
8712 # last 30 minutes
8713 backend pop
8714 mode tcp
8715 balance roundrobin
8716 stick store-request src
8717 stick-table type ip size 200k expire 30m
8718 server s1 192.168.1.1:110
8719 server s2 192.168.1.1:110
8720
8721 backend smtp
8722 mode tcp
8723 balance roundrobin
8724 stick match src table pop
8725 server s1 192.168.1.1:25
8726 server s2 192.168.1.1:25
8727
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008728 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008729 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008730
8731
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008732stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008733 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8734 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008735 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008736 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008737 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008738
8739 Arguments :
8740 ip a table declared with "type ip" will only store IPv4 addresses.
8741 This form is very compact (about 50 bytes per entry) and allows
8742 very fast entry lookup and stores with almost no overhead. This
8743 is mainly used to store client source IP addresses.
8744
David du Colombier9a6d3c92011-03-17 10:40:24 +01008745 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8746 This form is very compact (about 60 bytes per entry) and allows
8747 very fast entry lookup and stores with almost no overhead. This
8748 is mainly used to store client source IP addresses.
8749
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008750 integer a table declared with "type integer" will store 32bit integers
8751 which can represent a client identifier found in a request for
8752 instance.
8753
8754 string a table declared with "type string" will store substrings of up
8755 to <len> characters. If the string provided by the pattern
8756 extractor is larger than <len>, it will be truncated before
8757 being stored. During matching, at most <len> characters will be
8758 compared between the string in the table and the extracted
8759 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008760 to 32 characters.
8761
8762 binary a table declared with "type binary" will store binary blocks
8763 of <len> bytes. If the block provided by the pattern
8764 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008765 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008766 is shorter than <len>, it will be padded by 0. When not
8767 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008768
8769 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008770 "string" type table (See type "string" above). Or the number
8771 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008772 changing this parameter as memory usage will proportionally
8773 increase.
8774
8775 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008776 value directly impacts memory usage. Count approximately
8777 50 bytes per entry, plus the size of a string if any. The size
8778 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008779
8780 [nopurge] indicates that we refuse to purge older entries when the table
8781 is full. When not specified and the table is full when haproxy
8782 wants to store an entry in it, it will flush a few of the oldest
8783 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008784 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008785 be desirable to refuse new entries instead of purging the older
8786 ones. That may be the case when the amount of data to store is
8787 far above the hardware limits and we prefer not to offer access
8788 to new clients than to reject the ones already connected. When
8789 using this parameter, be sure to properly set the "expire"
8790 parameter (see below).
8791
Emeric Brunf099e792010-09-27 12:05:28 +02008792 <peersect> is the name of the peers section to use for replication. Entries
8793 which associate keys to server IDs are kept synchronized with
8794 the remote peers declared in this section. All entries are also
8795 automatically learned from the local peer (old process) during a
8796 soft restart.
8797
Willy Tarreau1abc6732015-05-01 19:21:02 +02008798 NOTE : each peers section may be referenced only by tables
8799 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008800
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008801 <expire> defines the maximum duration of an entry in the table since it
8802 was last created, refreshed or matched. The expiration delay is
8803 defined using the standard time format, similarly as the various
8804 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008805 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008806 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008807 be removed once full. Be sure not to use the "nopurge" parameter
8808 if not expiration delay is specified.
8809
Willy Tarreau08d5f982010-06-06 13:34:54 +02008810 <data_type> is used to store additional information in the stick-table. This
8811 may be used by ACLs in order to control various criteria related
8812 to the activity of the client matching the stick-table. For each
8813 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008814 that the additional data can fit. Several data types may be
8815 stored with an entry. Multiple data types may be specified after
8816 the "store" keyword, as a comma-separated list. Alternatively,
8817 it is possible to repeat the "store" keyword followed by one or
8818 several data types. Except for the "server_id" type which is
8819 automatically detected and enabled, all data types must be
8820 explicitly declared to be stored. If an ACL references a data
8821 type which is not stored, the ACL will simply not match. Some
8822 data types require an argument which must be passed just after
8823 the type between parenthesis. See below for the supported data
8824 types and their arguments.
8825
8826 The data types that can be stored with an entry are the following :
8827 - server_id : this is an integer which holds the numeric ID of the server a
8828 request was assigned to. It is used by the "stick match", "stick store",
8829 and "stick on" rules. It is automatically enabled when referenced.
8830
8831 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
8832 integer which may be used for anything. Most of the time it will be used
8833 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01008834 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02008835
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008836 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
8837 over a period. It is a positive 32-bit integer integer which may be used
8838 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01008839 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008840 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01008841 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008842
Frédéric Lécaille6778b272018-01-29 15:22:53 +01008843 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
8844 integer which may be used for anything. Most of the time it will be used
8845 to put a special tag on some entries, for instance to note that a
8846 specific behavior was detected and must be known for future matches.
8847
8848 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
8849 over a period. It is a positive 32-bit integer integer which may be used
8850 for anything. Just like <gpc1>, it counts events, but instead of keeping
8851 a cumulative number, it maintains the rate at which the counter is
8852 incremented. Most of the time it will be used to measure the frequency of
8853 occurrence of certain events (e.g. requests to a specific URL).
8854
Willy Tarreauc9705a12010-07-27 20:05:50 +02008855 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
8856 the absolute number of connections received from clients which matched
8857 this entry. It does not mean the connections were accepted, just that
8858 they were received.
8859
8860 - conn_cur : Current Connections. It is a positive 32-bit integer which
8861 stores the concurrent connection counts for the entry. It is incremented
8862 once an incoming connection matches the entry, and decremented once the
8863 connection leaves. That way it is possible to know at any time the exact
8864 number of concurrent connections for an entry.
8865
8866 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8867 integer parameter <period> which indicates in milliseconds the length
8868 of the period over which the average is measured. It reports the average
8869 incoming connection rate over that period, in connections per period. The
8870 result is an integer which can be matched using ACLs.
8871
8872 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
8873 the absolute number of sessions received from clients which matched this
8874 entry. A session is a connection that was accepted by the layer 4 rules.
8875
8876 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8877 integer parameter <period> which indicates in milliseconds the length
8878 of the period over which the average is measured. It reports the average
8879 incoming session rate over that period, in sessions per period. The
8880 result is an integer which can be matched using ACLs.
8881
8882 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
8883 counts the absolute number of HTTP requests received from clients which
8884 matched this entry. It does not matter whether they are valid requests or
8885 not. Note that this is different from sessions when keep-alive is used on
8886 the client side.
8887
8888 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8889 integer parameter <period> which indicates in milliseconds the length
8890 of the period over which the average is measured. It reports the average
8891 HTTP request rate over that period, in requests per period. The result is
8892 an integer which can be matched using ACLs. It does not matter whether
8893 they are valid requests or not. Note that this is different from sessions
8894 when keep-alive is used on the client side.
8895
8896 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
8897 counts the absolute number of HTTP requests errors induced by clients
8898 which matched this entry. Errors are counted on invalid and truncated
8899 requests, as well as on denied or tarpitted requests, and on failed
8900 authentications. If the server responds with 4xx, then the request is
8901 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01008902 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02008903
8904 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8905 integer parameter <period> which indicates in milliseconds the length
8906 of the period over which the average is measured. It reports the average
8907 HTTP request error rate over that period, in requests per period (see
8908 http_err_cnt above for what is accounted as an error). The result is an
8909 integer which can be matched using ACLs.
8910
8911 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008912 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02008913 which matched this entry. Headers are included in the count. This may be
8914 used to limit abuse of upload features on photo or video servers.
8915
8916 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8917 integer parameter <period> which indicates in milliseconds the length
8918 of the period over which the average is measured. It reports the average
8919 incoming bytes rate over that period, in bytes per period. It may be used
8920 to detect users which upload too much and too fast. Warning: with large
8921 uploads, it is possible that the amount of uploaded data will be counted
8922 once upon termination, thus causing spikes in the average transfer speed
8923 instead of having a smooth one. This may partially be smoothed with
8924 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
8925 recommended for better fairness.
8926
8927 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008928 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02008929 matched this entry. Headers are included in the count. This may be used
8930 to limit abuse of bots sucking the whole site.
8931
8932 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
8933 an integer parameter <period> which indicates in milliseconds the length
8934 of the period over which the average is measured. It reports the average
8935 outgoing bytes rate over that period, in bytes per period. It may be used
8936 to detect users which download too much and too fast. Warning: with large
8937 transfers, it is possible that the amount of transferred data will be
8938 counted once upon termination, thus causing spikes in the average
8939 transfer speed instead of having a smooth one. This may partially be
8940 smoothed with "option contstats" though this is not perfect yet. Use of
8941 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02008942
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008943 There is only one stick-table per proxy. At the moment of writing this doc,
8944 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008945 to be required, simply create a dummy backend with a stick-table in it and
8946 reference it.
8947
8948 It is important to understand that stickiness based on learning information
8949 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01008950 lost upon restart unless peers are properly configured to transfer such
8951 information upon restart (recommended). In general it can be good as a
8952 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008953
Willy Tarreauc9705a12010-07-27 20:05:50 +02008954 Last, memory requirements may be important when storing many data types.
8955 Indeed, storing all indicators above at once in each entry requires 116 bytes
8956 per entry, or 116 MB for a 1-million entries table. This is definitely not
8957 something that can be ignored.
8958
8959 Example:
8960 # Keep track of counters of up to 1 million IP addresses over 5 minutes
8961 # and store a general purpose counter and the average connection rate
8962 # computed over a sliding window of 30 seconds.
8963 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
8964
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008965 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01008966 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008967
8968
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008969stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01008970 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008971 May be used in sections : defaults | frontend | listen | backend
8972 no | no | yes | yes
8973
8974 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008975 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008976 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01008977 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008978 server is selected.
8979
8980 <table> is an optional stickiness table name. If unspecified, the same
8981 backend's table is used. A stickiness table is declared using
8982 the "stick-table" statement.
8983
8984 <cond> is an optional storage condition. It makes it possible to store
8985 certain criteria only when some conditions are met (or not met).
8986 For instance, it could be used to store the SSL session ID only
8987 when the response is a SSL server hello.
8988
8989 Some protocols or applications require complex stickiness rules and cannot
8990 always simply rely on cookies nor hashing. The "stick store-response"
8991 statement describes a rule to decide what to extract from the response and
8992 when to do it, in order to store it into a stickiness table for further
8993 requests to match it using the "stick match" statement. Obviously the
8994 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008995 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008996 See section 7 for a complete list of possible patterns and transformation
8997 rules.
8998
8999 The table has to be declared using the "stick-table" statement. It must be of
9000 a type compatible with the pattern. By default it is the one which is present
9001 in the same backend. It is possible to share a table with other backends by
9002 referencing it using the "table" keyword. If another table is referenced,
9003 the server's ID inside the backends are used. By default, all server IDs
9004 start at 1 in each backend, so the server ordering is enough. But in case of
9005 doubt, it is highly recommended to force server IDs using their "id" setting.
9006
9007 It is possible to restrict the conditions where a "stick store-response"
9008 statement will apply, using "if" or "unless" followed by a condition. This
9009 condition will be evaluated while parsing the response, so any criteria can
9010 be used. See section 7 for ACL based conditions.
9011
9012 There is no limit on the number of "stick store-response" statements, but
9013 there is a limit of 8 simultaneous stores per request or response. This
9014 makes it possible to store up to 8 criteria, all extracted from either the
9015 request or the response, regardless of the number of rules. Only the 8 first
9016 ones which match will be kept. Using this, it is possible to feed multiple
9017 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009018 another protocol or access method. Using multiple store-response rules with
9019 the same table is possible and may be used to find the best criterion to rely
9020 on, by arranging the rules by decreasing preference order. Only the first
9021 extracted criterion for a given table will be stored. All subsequent store-
9022 response rules referencing the same table will be skipped and their ACLs will
9023 not be evaluated. However, even if a store-request rule references a table, a
9024 store-response rule may also use the same table. This means that each table
9025 may learn exactly one element from the request and one element from the
9026 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009027
9028 The table will contain the real server that processed the request.
9029
9030 Example :
9031 # Learn SSL session ID from both request and response and create affinity.
9032 backend https
9033 mode tcp
9034 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009035 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009036 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009037
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009038 acl clienthello req_ssl_hello_type 1
9039 acl serverhello rep_ssl_hello_type 2
9040
9041 # use tcp content accepts to detects ssl client and server hello.
9042 tcp-request inspect-delay 5s
9043 tcp-request content accept if clienthello
9044
9045 # no timeout on response inspect delay by default.
9046 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009047
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009048 # SSL session ID (SSLID) may be present on a client or server hello.
9049 # Its length is coded on 1 byte at offset 43 and its value starts
9050 # at offset 44.
9051
9052 # Match and learn on request if client hello.
9053 stick on payload_lv(43,1) if clienthello
9054
9055 # Learn on response if server hello.
9056 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009057
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009058 server s1 192.168.1.1:443
9059 server s2 192.168.1.1:443
9060
9061 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9062 extraction.
9063
9064
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009065tcp-check connect [params*]
9066 Opens a new connection
9067 May be used in sections: defaults | frontend | listen | backend
9068 no | no | yes | yes
9069
9070 When an application lies on more than a single TCP port or when HAProxy
9071 load-balance many services in a single backend, it makes sense to probe all
9072 the services individually before considering a server as operational.
9073
9074 When there are no TCP port configured on the server line neither server port
9075 directive, then the 'tcp-check connect port <port>' must be the first step
9076 of the sequence.
9077
9078 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9079 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9080 do.
9081
9082 Parameters :
9083 They are optional and can be used to describe how HAProxy should open and
9084 use the TCP connection.
9085
9086 port if not set, check port or server port is used.
9087 It tells HAProxy where to open the connection to.
9088 <port> must be a valid TCP port source integer, from 1 to 65535.
9089
9090 send-proxy send a PROXY protocol string
9091
9092 ssl opens a ciphered connection
9093
9094 Examples:
9095 # check HTTP and HTTPs services on a server.
9096 # first open port 80 thanks to server line port directive, then
9097 # tcp-check opens port 443, ciphered and run a request on it:
9098 option tcp-check
9099 tcp-check connect
9100 tcp-check send GET\ /\ HTTP/1.0\r\n
9101 tcp-check send Host:\ haproxy.1wt.eu\r\n
9102 tcp-check send \r\n
9103 tcp-check expect rstring (2..|3..)
9104 tcp-check connect port 443 ssl
9105 tcp-check send GET\ /\ HTTP/1.0\r\n
9106 tcp-check send Host:\ haproxy.1wt.eu\r\n
9107 tcp-check send \r\n
9108 tcp-check expect rstring (2..|3..)
9109 server www 10.0.0.1 check port 80
9110
9111 # check both POP and IMAP from a single server:
9112 option tcp-check
9113 tcp-check connect port 110
9114 tcp-check expect string +OK\ POP3\ ready
9115 tcp-check connect port 143
9116 tcp-check expect string *\ OK\ IMAP4\ ready
9117 server mail 10.0.0.1 check
9118
9119 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9120
9121
9122tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009123 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009124 May be used in sections: defaults | frontend | listen | backend
9125 no | no | yes | yes
9126
9127 Arguments :
9128 <match> is a keyword indicating how to look for a specific pattern in the
9129 response. The keyword may be one of "string", "rstring" or
9130 binary.
9131 The keyword may be preceded by an exclamation mark ("!") to negate
9132 the match. Spaces are allowed between the exclamation mark and the
9133 keyword. See below for more details on the supported keywords.
9134
9135 <pattern> is the pattern to look for. It may be a string or a regular
9136 expression. If the pattern contains spaces, they must be escaped
9137 with the usual backslash ('\').
9138 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009139 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009140 two digits will represent a byte. The hexadecimal digits may be
9141 used upper or lower case.
9142
9143
9144 The available matches are intentionally similar to their http-check cousins :
9145
9146 string <string> : test the exact string matches in the response buffer.
9147 A health check response will be considered valid if the
9148 response's buffer contains this exact string. If the
9149 "string" keyword is prefixed with "!", then the response
9150 will be considered invalid if the body contains this
9151 string. This can be used to look for a mandatory pattern
9152 in a protocol response, or to detect a failure when a
9153 specific error appears in a protocol banner.
9154
9155 rstring <regex> : test a regular expression on the response buffer.
9156 A health check response will be considered valid if the
9157 response's buffer matches this expression. If the
9158 "rstring" keyword is prefixed with "!", then the response
9159 will be considered invalid if the body matches the
9160 expression.
9161
9162 binary <hexstring> : test the exact string in its hexadecimal form matches
9163 in the response buffer. A health check response will
9164 be considered valid if the response's buffer contains
9165 this exact hexadecimal string.
9166 Purpose is to match data on binary protocols.
9167
9168 It is important to note that the responses will be limited to a certain size
9169 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9170 Thus, too large responses may not contain the mandatory pattern when using
9171 "string", "rstring" or binary. If a large response is absolutely required, it
9172 is possible to change the default max size by setting the global variable.
9173 However, it is worth keeping in mind that parsing very large responses can
9174 waste some CPU cycles, especially when regular expressions are used, and that
9175 it is always better to focus the checks on smaller resources. Also, in its
9176 current state, the check will not find any string nor regex past a null
9177 character in the response. Similarly it is not possible to request matching
9178 the null character.
9179
9180 Examples :
9181 # perform a POP check
9182 option tcp-check
9183 tcp-check expect string +OK\ POP3\ ready
9184
9185 # perform an IMAP check
9186 option tcp-check
9187 tcp-check expect string *\ OK\ IMAP4\ ready
9188
9189 # look for the redis master server
9190 option tcp-check
9191 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009192 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009193 tcp-check send info\ replication\r\n
9194 tcp-check expect string role:master
9195 tcp-check send QUIT\r\n
9196 tcp-check expect string +OK
9197
9198
9199 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9200 "tcp-check send-binary", "http-check expect", tune.chksize
9201
9202
9203tcp-check send <data>
9204 Specify a string to be sent as a question during a generic health check
9205 May be used in sections: defaults | frontend | listen | backend
9206 no | no | yes | yes
9207
9208 <data> : the data to be sent as a question during a generic health check
9209 session. For now, <data> must be a string.
9210
9211 Examples :
9212 # look for the redis master server
9213 option tcp-check
9214 tcp-check send info\ replication\r\n
9215 tcp-check expect string role:master
9216
9217 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9218 "tcp-check send-binary", tune.chksize
9219
9220
Davor Ocelice9ed2812017-12-25 17:49:28 +01009221tcp-check send-binary <hexstring>
9222 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009223 tcp health check
9224 May be used in sections: defaults | frontend | listen | backend
9225 no | no | yes | yes
9226
9227 <data> : the data to be sent as a question during a generic health check
9228 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009229 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009230 response buffer. A health check response will be considered
9231 valid if the response's buffer contains this exact
9232 hexadecimal string.
9233 Purpose is to send binary data to ask on binary protocols.
9234
9235 Examples :
9236 # redis check in binary
9237 option tcp-check
9238 tcp-check send-binary 50494e470d0a # PING\r\n
9239 tcp-check expect binary 2b504F4e47 # +PONG
9240
9241
9242 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9243 "tcp-check send", tune.chksize
9244
9245
Willy Tarreaue9656522010-08-17 15:40:09 +02009246tcp-request connection <action> [{if | unless} <condition>]
9247 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009248 May be used in sections : defaults | frontend | listen | backend
9249 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009250 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009251 <action> defines the action to perform if the condition applies. See
9252 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009253
Willy Tarreaue9656522010-08-17 15:40:09 +02009254 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009255
9256 Immediately after acceptance of a new incoming connection, it is possible to
9257 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009258 or dropped or have its counters tracked. Those conditions cannot make use of
9259 any data contents because the connection has not been read from yet, and the
9260 buffers are not yet allocated. This is used to selectively and very quickly
9261 accept or drop connections from various sources with a very low overhead. If
9262 some contents need to be inspected in order to take the decision, the
9263 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009264
Willy Tarreaue9656522010-08-17 15:40:09 +02009265 The "tcp-request connection" rules are evaluated in their exact declaration
9266 order. If no rule matches or if there is no rule, the default action is to
9267 accept the incoming connection. There is no specific limit to the number of
9268 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009269
Willy Tarreaua9083d02015-05-08 15:27:59 +02009270 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009271 - accept :
9272 accepts the connection if the condition is true (when used with "if")
9273 or false (when used with "unless"). The first such rule executed ends
9274 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009275
Willy Tarreaue9656522010-08-17 15:40:09 +02009276 - reject :
9277 rejects the connection if the condition is true (when used with "if")
9278 or false (when used with "unless"). The first such rule executed ends
9279 the rules evaluation. Rejected connections do not even become a
9280 session, which is why they are accounted separately for in the stats,
9281 as "denied connections". They are not considered for the session
9282 rate-limit and are not logged either. The reason is that these rules
9283 should only be used to filter extremely high connection rates such as
9284 the ones encountered during a massive DDoS attack. Under these extreme
9285 conditions, the simple action of logging each event would make the
9286 system collapse and would considerably lower the filtering capacity. If
9287 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009288 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009289
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009290 - expect-proxy layer4 :
9291 configures the client-facing connection to receive a PROXY protocol
9292 header before any byte is read from the socket. This is equivalent to
9293 having the "accept-proxy" keyword on the "bind" line, except that using
9294 the TCP rule allows the PROXY protocol to be accepted only for certain
9295 IP address ranges using an ACL. This is convenient when multiple layers
9296 of load balancers are passed through by traffic coming from public
9297 hosts.
9298
Bertrand Jacquin90759682016-06-06 15:35:39 +01009299 - expect-netscaler-cip layer4 :
9300 configures the client-facing connection to receive a NetScaler Client
9301 IP insertion protocol header before any byte is read from the socket.
9302 This is equivalent to having the "accept-netscaler-cip" keyword on the
9303 "bind" line, except that using the TCP rule allows the PROXY protocol
9304 to be accepted only for certain IP address ranges using an ACL. This
9305 is convenient when multiple layers of load balancers are passed
9306 through by traffic coming from public hosts.
9307
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009308 - capture <sample> len <length> :
9309 This only applies to "tcp-request content" rules. It captures sample
9310 expression <sample> from the request buffer, and converts it to a
9311 string of at most <len> characters. The resulting string is stored into
9312 the next request "capture" slot, so it will possibly appear next to
9313 some captured HTTP headers. It will then automatically appear in the
9314 logs, and it will be possible to extract it using sample fetch rules to
9315 feed it into headers or anything. The length should be limited given
9316 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009317 session life. Please check section 7.3 (Fetching samples) and "capture
9318 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009319
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009320 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009321 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009322 rules do not stop evaluation and do not change default action. The
9323 number of counters that may be simultaneously tracked by the same
9324 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -05009325 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009326 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9327 tracking of the counters of the specified table as the first set. The
9328 first "track-sc1" rule executed enables tracking of the counters of the
9329 specified table as the second set. The first "track-sc2" rule executed
9330 enables tracking of the counters of the specified table as the third
9331 set. It is a recommended practice to use the first set of counters for
9332 the per-frontend counters and the second set for the per-backend ones.
9333 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009334
Willy Tarreaue9656522010-08-17 15:40:09 +02009335 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009336 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009337 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009338 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009339 and used to select which table entry to update the counters.
9340 Note that "tcp-request connection" cannot use content-based
9341 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009342
Willy Tarreaue9656522010-08-17 15:40:09 +02009343 <table> is an optional table to be used instead of the default one,
9344 which is the stick-table declared in the current proxy. All
9345 the counters for the matches and updates for the key will
9346 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009347
Willy Tarreaue9656522010-08-17 15:40:09 +02009348 Once a "track-sc*" rule is executed, the key is looked up in the table
9349 and if it is not found, an entry is allocated for it. Then a pointer to
9350 that entry is kept during all the session's life, and this entry's
9351 counters are updated as often as possible, every time the session's
9352 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009353 Counters are only updated for events that happen after the tracking has
9354 been started. For example, connection counters will not be updated when
9355 tracking layer 7 information, since the connection event happens before
9356 layer7 information is extracted.
9357
Willy Tarreaue9656522010-08-17 15:40:09 +02009358 If the entry tracks concurrent connection counters, one connection is
9359 counted for as long as the entry is tracked, and the entry will not
9360 expire during that time. Tracking counters also provides a performance
9361 advantage over just checking the keys, because only one table lookup is
9362 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009363
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009364 - sc-inc-gpc0(<sc-id>):
9365 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9366 counter designated by <sc-id>. If an error occurs, this action silently
9367 fails and the actions evaluation continues.
9368
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009369 - sc-inc-gpc1(<sc-id>):
9370 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9371 counter designated by <sc-id>. If an error occurs, this action silently
9372 fails and the actions evaluation continues.
9373
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009374 - sc-set-gpt0(<sc-id>) <int>:
9375 This action sets the GPT0 tag according to the sticky counter designated
9376 by <sc-id> and the value of <int>. The expected result is a boolean. If
9377 an error occurs, this action silently fails and the actions evaluation
9378 continues.
9379
William Lallemand2e785f22016-05-25 01:48:42 +02009380 - set-src <expr> :
9381 Is used to set the source IP address to the value of specified
9382 expression. Useful if you want to mask source IP for privacy.
9383 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009384 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +02009385
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009386 Arguments:
9387 <expr> Is a standard HAProxy expression formed by a sample-fetch
9388 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +02009389
9390 Example:
William Lallemand2e785f22016-05-25 01:48:42 +02009391 tcp-request connection set-src src,ipmask(24)
9392
Willy Tarreau0c630532016-10-21 17:52:58 +02009393 When possible, set-src preserves the original source port as long as the
9394 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009395
William Lallemand44be6402016-05-25 01:51:35 +02009396 - set-src-port <expr> :
9397 Is used to set the source port address to the value of specified
9398 expression.
9399
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009400 Arguments:
9401 <expr> Is a standard HAProxy expression formed by a sample-fetch
9402 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +02009403
9404 Example:
William Lallemand44be6402016-05-25 01:51:35 +02009405 tcp-request connection set-src-port int(4000)
9406
Willy Tarreau0c630532016-10-21 17:52:58 +02009407 When possible, set-src-port preserves the original source address as long
9408 as the address family supports a port, otherwise it forces the source
9409 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009410
William Lallemand13e9b0c2016-05-25 02:34:07 +02009411 - set-dst <expr> :
9412 Is used to set the destination IP address to the value of specified
9413 expression. Useful if you want to mask IP for privacy in log.
9414 If you want to provide an IP from a HTTP header use "http-request
9415 set-dst". If you want to connect to the new address/port, use
9416 '0.0.0.0:0' as a server address in the backend.
9417
9418 <expr> Is a standard HAProxy expression formed by a sample-fetch
9419 followed by some converters.
9420
9421 Example:
9422
9423 tcp-request connection set-dst dst,ipmask(24)
9424 tcp-request connection set-dst ipv4(10.0.0.1)
9425
Willy Tarreau0c630532016-10-21 17:52:58 +02009426 When possible, set-dst preserves the original destination port as long as
9427 the address family allows it, otherwise the destination port is set to 0.
9428
William Lallemand13e9b0c2016-05-25 02:34:07 +02009429 - set-dst-port <expr> :
9430 Is used to set the destination port address to the value of specified
9431 expression. If you want to connect to the new address/port, use
9432 '0.0.0.0:0' as a server address in the backend.
9433
9434
9435 <expr> Is a standard HAProxy expression formed by a sample-fetch
9436 followed by some converters.
9437
9438 Example:
9439
9440 tcp-request connection set-dst-port int(4000)
9441
Willy Tarreau0c630532016-10-21 17:52:58 +02009442 When possible, set-dst-port preserves the original destination address as
9443 long as the address family supports a port, otherwise it forces the
9444 destination address to IPv4 "0.0.0.0" before rewriting the port.
9445
Willy Tarreau2d392c22015-08-24 01:43:45 +02009446 - "silent-drop" :
9447 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009448 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009449 to prevent the client from being notified. The effect it then that the
9450 client still sees an established connection while there's none on
9451 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9452 except that it doesn't use any local resource at all on the machine
9453 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009454 slow down stronger attackers. It is important to understand the impact
9455 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009456 client and HAProxy (firewalls, proxies, load balancers) will also keep
9457 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009458 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009459 TCP_REPAIR socket option is used to block the emission of a TCP
9460 reset. On other systems, the socket's TTL is reduced to 1 so that the
9461 TCP reset doesn't pass the first router, though it's still delivered to
9462 local networks. Do not use it unless you fully understand how it works.
9463
Willy Tarreaue9656522010-08-17 15:40:09 +02009464 Note that the "if/unless" condition is optional. If no condition is set on
9465 the action, it is simply performed unconditionally. That can be useful for
9466 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009467
Willy Tarreaue9656522010-08-17 15:40:09 +02009468 Example: accept all connections from white-listed hosts, reject too fast
9469 connection without counting them, and track accepted connections.
9470 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009471
Willy Tarreaue9656522010-08-17 15:40:09 +02009472 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009473 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009474 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009475
Willy Tarreaue9656522010-08-17 15:40:09 +02009476 Example: accept all connections from white-listed hosts, count all other
9477 connections and reject too fast ones. This results in abusive ones
9478 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009479
Willy Tarreaue9656522010-08-17 15:40:09 +02009480 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009481 tcp-request connection track-sc0 src
9482 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009483
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009484 Example: enable the PROXY protocol for traffic coming from all known proxies.
9485
9486 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9487
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009488 See section 7 about ACL usage.
9489
Willy Tarreau4f614292016-10-21 17:49:36 +02009490 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009491
9492
Willy Tarreaue9656522010-08-17 15:40:09 +02009493tcp-request content <action> [{if | unless} <condition>]
9494 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009495 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009496 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009497 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009498 <action> defines the action to perform if the condition applies. See
9499 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009500
Willy Tarreaue9656522010-08-17 15:40:09 +02009501 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009502
Davor Ocelice9ed2812017-12-25 17:49:28 +01009503 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009504 called "TCP content inspection". During this stage, ACL-based rules are
9505 evaluated every time the request contents are updated, until either an
9506 "accept" or a "reject" rule matches, or the TCP request inspection delay
9507 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009508
Willy Tarreaue9656522010-08-17 15:40:09 +02009509 The first difference between these rules and "tcp-request connection" rules
9510 is that "tcp-request content" rules can make use of contents to take a
9511 decision. Most often, these decisions will consider a protocol recognition or
9512 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009513 both frontends and backends. In case of HTTP keep-alive with the client, all
9514 tcp-request content rules are evaluated again, so haproxy keeps a record of
9515 what sticky counters were assigned by a "tcp-request connection" versus a
9516 "tcp-request content" rule, and flushes all the content-related ones after
9517 processing an HTTP request, so that they may be evaluated again by the rules
9518 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009519 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009520 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009521
Willy Tarreaue9656522010-08-17 15:40:09 +02009522 Content-based rules are evaluated in their exact declaration order. If no
9523 rule matches or if there is no rule, the default action is to accept the
9524 contents. There is no specific limit to the number of rules which may be
9525 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009526
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009527 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009528 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +01009529 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009530 - reject : the request is rejected and the connection is closed
9531 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009532 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009533 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009534 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009535 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009536 - sc-set-gpt0(<sc-id>) <int>
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +02009537 - set-dst <expr>
9538 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009539 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009540 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009541 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009542 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009543
Willy Tarreaue9656522010-08-17 15:40:09 +02009544 They have the same meaning as their counter-parts in "tcp-request connection"
9545 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +01009546 For "do-resolve" action, please check the "http-request do-resolve"
9547 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +02009548
Willy Tarreauf3338342014-01-28 21:40:28 +01009549 While there is nothing mandatory about it, it is recommended to use the
9550 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9551 content" rules in the frontend, and track-sc2 for "tcp-request content"
9552 rules in the backend, because that makes the configuration more readable
9553 and easier to troubleshoot, but this is just a guideline and all counters
9554 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009555
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009556 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009557 the action, it is simply performed unconditionally. That can be useful for
9558 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009559
Willy Tarreaue9656522010-08-17 15:40:09 +02009560 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009561 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9562 contents of a buffer before extracting the required data. If the buffered
9563 contents do not parse as a valid HTTP message, then the ACL does not match.
9564 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009565 processing, so there is no risk of parsing something differently. In an HTTP
9566 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9567 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009568
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009569 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009570 are present when the rule is processed. The rule processing engine is able to
9571 wait until the inspect delay expires when the data to be tracked is not yet
9572 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009573
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +02009574 The "set-dst" and "set-dst-port" are used to set respectively the destination
9575 IP and port. More information on how to use it at "http-request set-dst".
9576
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009577 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009578 declared inline. For "tcp-request session" rules, only session-level
9579 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009580
Daniel Schneller0b547052016-03-21 20:46:57 +01009581 <var-name> The name of the variable starts with an indication about
9582 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009583 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009584 "sess" : the variable is shared with the whole session
9585 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009586 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009587 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009588 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009589 "res" : the variable is shared only during response
9590 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009591 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009592 The name may only contain characters 'a-z', 'A-Z', '0-9',
9593 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009594
9595 <expr> Is a standard HAProxy expression formed by a sample-fetch
9596 followed by some converters.
9597
Christopher Faulet85d79c92016-11-09 16:54:56 +01009598 The "unset-var" is used to unset a variable. See above for details about
9599 <var-name>.
9600
Patrick Hemmer268a7072018-05-11 12:52:31 -04009601 The "set-priority-class" is used to set the queue priority class of the
9602 current request. The value must be a sample expression which converts to an
9603 integer in the range -2047..2047. Results outside this range will be
9604 truncated. The priority class determines the order in which queued requests
9605 are processed. Lower values have higher priority.
9606
9607 The "set-priority-offset" is used to set the queue priority timestamp offset
9608 of the current request. The value must be a sample expression which converts
9609 to an integer in the range -524287..524287. Results outside this range will be
9610 truncated. When a request is queued, it is ordered first by the priority
9611 class, then by the current timestamp adjusted by the given offset in
9612 milliseconds. Lower values have higher priority.
9613 Note that the resulting timestamp is is only tracked with enough precision for
9614 524,287ms (8m44s287ms). If the request is queued long enough to where the
9615 adjusted timestamp exceeds this value, it will be misidentified as highest
9616 priority. Thus it is important to set "timeout queue" to a value, where when
9617 combined with the offset, does not exceed this limit.
9618
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009619 The "send-spoe-group" is used to trigger sending of a group of SPOE
9620 messages. To do so, the SPOE engine used to send messages must be defined, as
9621 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9622 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9623 the SPOE agent name must be used.
9624
9625 <engine-name> The SPOE engine name.
9626
9627 <group-name> The SPOE group name as specified in the engine configuration.
9628
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009629 Example:
9630
9631 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009632 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009633
Willy Tarreau62644772008-07-16 18:36:06 +02009634 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009635 # Accept HTTP requests containing a Host header saying "example.com"
9636 # and reject everything else.
9637 acl is_host_com hdr(Host) -i example.com
9638 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009639 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009640 tcp-request content reject
9641
9642 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009643 # reject SMTP connection if client speaks first
9644 tcp-request inspect-delay 30s
9645 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009646 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009647
9648 # Forward HTTPS connection only if client speaks
9649 tcp-request inspect-delay 30s
9650 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009651 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009652 tcp-request content reject
9653
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009654 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009655 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009656 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009657 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009658 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9659 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009660
9661 Example:
9662 # track request counts per "base" (concatenation of Host+URL)
9663 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009664 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009665
Willy Tarreaue9656522010-08-17 15:40:09 +02009666 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009667 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009668
9669 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009670 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009671 # protecting all our sites
9672 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009673 tcp-request connection track-sc0 src
9674 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009675 ...
9676 use_backend http_dynamic if { path_end .php }
9677
9678 backend http_dynamic
9679 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009680 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009681 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009682 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009683 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009684 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009685 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009686
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009687 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009688
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009689 See also : "tcp-request connection", "tcp-request session",
9690 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009691
9692
9693tcp-request inspect-delay <timeout>
9694 Set the maximum allowed time to wait for data during content inspection
9695 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009696 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009697 Arguments :
9698 <timeout> is the timeout value specified in milliseconds by default, but
9699 can be in any other unit if the number is suffixed by the unit,
9700 as explained at the top of this document.
9701
9702 People using haproxy primarily as a TCP relay are often worried about the
9703 risk of passing any type of protocol to a server without any analysis. In
9704 order to be able to analyze the request contents, we must first withhold
9705 the data then analyze them. This statement simply enables withholding of
9706 data for at most the specified amount of time.
9707
Willy Tarreaufb356202010-08-03 14:02:05 +02009708 TCP content inspection applies very early when a connection reaches a
9709 frontend, then very early when the connection is forwarded to a backend. This
9710 means that a connection may experience a first delay in the frontend and a
9711 second delay in the backend if both have tcp-request rules.
9712
Willy Tarreau62644772008-07-16 18:36:06 +02009713 Note that when performing content inspection, haproxy will evaluate the whole
9714 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009715 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009716 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009717 contents are definitive. If no delay is set, haproxy will not wait at all
9718 and will immediately apply a verdict based on the available information.
9719 Obviously this is unlikely to be very useful and might even be racy, so such
9720 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009721
9722 As soon as a rule matches, the request is released and continues as usual. If
9723 the timeout is reached and no rule matches, the default policy will be to let
9724 it pass through unaffected.
9725
9726 For most protocols, it is enough to set it to a few seconds, as most clients
9727 send the full request immediately upon connection. Add 3 or more seconds to
9728 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009729 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009730 before the server (e.g. SMTP), or to wait for a client to talk before passing
9731 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009732 least the inspection delay, otherwise it will expire first. If the client
9733 closes the connection or if the buffer is full, the delay immediately expires
9734 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009735
Willy Tarreau55165fe2009-05-10 12:02:55 +02009736 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009737 "timeout client".
9738
9739
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009740tcp-response content <action> [{if | unless} <condition>]
9741 Perform an action on a session response depending on a layer 4-7 condition
9742 May be used in sections : defaults | frontend | listen | backend
9743 no | no | yes | yes
9744 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009745 <action> defines the action to perform if the condition applies. See
9746 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009747
9748 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9749
Davor Ocelice9ed2812017-12-25 17:49:28 +01009750 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009751 called "TCP content inspection". During this stage, ACL-based rules are
9752 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009753 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9754 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009755
9756 Most often, these decisions will consider a protocol recognition or validity.
9757
9758 Content-based rules are evaluated in their exact declaration order. If no
9759 rule matches or if there is no rule, the default action is to accept the
9760 contents. There is no specific limit to the number of rules which may be
9761 inserted.
9762
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009763 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009764 - accept :
9765 accepts the response if the condition is true (when used with "if")
9766 or false (when used with "unless"). The first such rule executed ends
9767 the rules evaluation.
9768
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009769 - close :
9770 immediately closes the connection with the server if the condition is
9771 true (when used with "if"), or false (when used with "unless"). The
9772 first such rule executed ends the rules evaluation. The main purpose of
9773 this action is to force a connection to be finished between a client
9774 and a server after an exchange when the application protocol expects
9775 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009776 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009777 protocols.
9778
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009779 - reject :
9780 rejects the response if the condition is true (when used with "if")
9781 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009782 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009783
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009784 - set-var(<var-name>) <expr>
9785 Sets a variable.
9786
Christopher Faulet85d79c92016-11-09 16:54:56 +01009787 - unset-var(<var-name>)
9788 Unsets a variable.
9789
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009790 - sc-inc-gpc0(<sc-id>):
9791 This action increments the GPC0 counter according to the sticky
9792 counter designated by <sc-id>. If an error occurs, this action fails
9793 silently and the actions evaluation continues.
9794
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009795 - sc-inc-gpc1(<sc-id>):
9796 This action increments the GPC1 counter according to the sticky
9797 counter designated by <sc-id>. If an error occurs, this action fails
9798 silently and the actions evaluation continues.
9799
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009800 - sc-set-gpt0(<sc-id>) <int> :
9801 This action sets the GPT0 tag according to the sticky counter designated
9802 by <sc-id> and the value of <int>. The expected result is a boolean. If
9803 an error occurs, this action silently fails and the actions evaluation
9804 continues.
9805
Willy Tarreau2d392c22015-08-24 01:43:45 +02009806 - "silent-drop" :
9807 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009808 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009809 to prevent the client from being notified. The effect it then that the
9810 client still sees an established connection while there's none on
9811 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9812 except that it doesn't use any local resource at all on the machine
9813 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009814 slow down stronger attackers. It is important to understand the impact
9815 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009816 client and HAProxy (firewalls, proxies, load balancers) will also keep
9817 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009818 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009819 TCP_REPAIR socket option is used to block the emission of a TCP
9820 reset. On other systems, the socket's TTL is reduced to 1 so that the
9821 TCP reset doesn't pass the first router, though it's still delivered to
9822 local networks. Do not use it unless you fully understand how it works.
9823
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009824 - send-spoe-group <engine-name> <group-name>
9825 Send a group of SPOE messages.
9826
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009827 Note that the "if/unless" condition is optional. If no condition is set on
9828 the action, it is simply performed unconditionally. That can be useful for
9829 for changing the default action to a reject.
9830
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009831 It is perfectly possible to match layer 7 contents with "tcp-response
9832 content" rules, but then it is important to ensure that a full response has
9833 been buffered, otherwise no contents will match. In order to achieve this,
9834 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009835 period.
9836
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009837 The "set-var" is used to set the content of a variable. The variable is
9838 declared inline.
9839
Daniel Schneller0b547052016-03-21 20:46:57 +01009840 <var-name> The name of the variable starts with an indication about
9841 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009842 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009843 "sess" : the variable is shared with the whole session
9844 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009845 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009846 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009847 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009848 "res" : the variable is shared only during response
9849 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009850 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009851 The name may only contain characters 'a-z', 'A-Z', '0-9',
9852 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009853
9854 <expr> Is a standard HAProxy expression formed by a sample-fetch
9855 followed by some converters.
9856
9857 Example:
9858
9859 tcp-request content set-var(sess.my_var) src
9860
Christopher Faulet85d79c92016-11-09 16:54:56 +01009861 The "unset-var" is used to unset a variable. See above for details about
9862 <var-name>.
9863
9864 Example:
9865
9866 tcp-request content unset-var(sess.my_var)
9867
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009868 The "send-spoe-group" is used to trigger sending of a group of SPOE
9869 messages. To do so, the SPOE engine used to send messages must be defined, as
9870 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9871 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9872 the SPOE agent name must be used.
9873
9874 <engine-name> The SPOE engine name.
9875
9876 <group-name> The SPOE group name as specified in the engine configuration.
9877
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009878 See section 7 about ACL usage.
9879
9880 See also : "tcp-request content", "tcp-response inspect-delay"
9881
9882
Willy Tarreau4f614292016-10-21 17:49:36 +02009883tcp-request session <action> [{if | unless} <condition>]
9884 Perform an action on a validated session depending on a layer 5 condition
9885 May be used in sections : defaults | frontend | listen | backend
9886 no | yes | yes | no
9887 Arguments :
9888 <action> defines the action to perform if the condition applies. See
9889 below.
9890
9891 <condition> is a standard layer5-only ACL-based condition (see section 7).
9892
Davor Ocelice9ed2812017-12-25 17:49:28 +01009893 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +02009894 it is possible to evaluate some conditions to decide whether this session
9895 must be accepted or dropped or have its counters tracked. Those conditions
9896 cannot make use of any data contents because no buffers are allocated yet and
9897 the processing cannot wait at this stage. The main use case it to copy some
9898 early information into variables (since variables are accessible in the
9899 session), or to keep track of some information collected after the handshake,
9900 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +01009901 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +02009902 extracted information can thus be copied to a variable or tracked using
9903 "track-sc" rules. Of course it is also possible to decide to accept/reject as
9904 with other rulesets. Most operations performed here could also be performed
9905 in "tcp-request content" rules, except that in HTTP these rules are evaluated
9906 for each new request, and that might not always be acceptable. For example a
9907 rule might increment a counter on each evaluation. It would also be possible
9908 that a country is resolved by geolocation from the source IP address,
9909 assigned to a session-wide variable, then the source address rewritten from
9910 an HTTP header for all requests. If some contents need to be inspected in
9911 order to take the decision, the "tcp-request content" statements must be used
9912 instead.
9913
9914 The "tcp-request session" rules are evaluated in their exact declaration
9915 order. If no rule matches or if there is no rule, the default action is to
9916 accept the incoming session. There is no specific limit to the number of
9917 rules which may be inserted.
9918
9919 Several types of actions are supported :
9920 - accept : the request is accepted
9921 - reject : the request is rejected and the connection is closed
9922 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
9923 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009924 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009925 - sc-set-gpt0(<sc-id>) <int>
9926 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009927 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009928 - silent-drop
9929
9930 These actions have the same meaning as their respective counter-parts in
9931 "tcp-request connection" and "tcp-request content", so please refer to these
9932 sections for a complete description.
9933
9934 Note that the "if/unless" condition is optional. If no condition is set on
9935 the action, it is simply performed unconditionally. That can be useful for
9936 "track-sc*" actions as well as for changing the default action to a reject.
9937
9938 Example: track the original source address by default, or the one advertised
9939 in the PROXY protocol header for connection coming from the local
9940 proxies. The first connection-level rule enables receipt of the
9941 PROXY protocol for these ones, the second rule tracks whatever
9942 address we decide to keep after optional decoding.
9943
9944 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9945 tcp-request session track-sc0 src
9946
9947 Example: accept all sessions from white-listed hosts, reject too fast
9948 sessions without counting them, and track accepted sessions.
9949 This results in session rate being capped from abusive sources.
9950
9951 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9952 tcp-request session reject if { src_sess_rate gt 10 }
9953 tcp-request session track-sc0 src
9954
9955 Example: accept all sessions from white-listed hosts, count all other
9956 sessions and reject too fast ones. This results in abusive ones
9957 being blocked as long as they don't slow down.
9958
9959 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9960 tcp-request session track-sc0 src
9961 tcp-request session reject if { sc0_sess_rate gt 10 }
9962
9963 See section 7 about ACL usage.
9964
9965 See also : "tcp-request connection", "tcp-request content", "stick-table"
9966
9967
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009968tcp-response inspect-delay <timeout>
9969 Set the maximum allowed time to wait for a response during content inspection
9970 May be used in sections : defaults | frontend | listen | backend
9971 no | no | yes | yes
9972 Arguments :
9973 <timeout> is the timeout value specified in milliseconds by default, but
9974 can be in any other unit if the number is suffixed by the unit,
9975 as explained at the top of this document.
9976
9977 See also : "tcp-response content", "tcp-request inspect-delay".
9978
9979
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009980timeout check <timeout>
9981 Set additional check timeout, but only after a connection has been already
9982 established.
9983
9984 May be used in sections: defaults | frontend | listen | backend
9985 yes | no | yes | yes
9986 Arguments:
9987 <timeout> is the timeout value specified in milliseconds by default, but
9988 can be in any other unit if the number is suffixed by the unit,
9989 as explained at the top of this document.
9990
9991 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
9992 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009993 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009994 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01009995 (Please also note that there is no valid reason to have such long connect
9996 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
9997 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009998
9999 If "timeout check" is not set haproxy uses "inter" for complete check
10000 timeout (connect + read) exactly like all <1.3.15 version.
10001
10002 In most cases check request is much simpler and faster to handle than normal
10003 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010004 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010005
10006 This parameter is specific to backends, but can be specified once for all in
10007 "defaults" sections. This is in fact one of the easiest solutions not to
10008 forget about it.
10009
Willy Tarreau41a340d2008-01-22 12:25:31 +010010010 See also: "timeout connect", "timeout queue", "timeout server",
10011 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010012
10013
Willy Tarreau0ba27502007-12-24 16:55:16 +010010014timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010015 Set the maximum inactivity time on the client side.
10016 May be used in sections : defaults | frontend | listen | backend
10017 yes | yes | yes | no
10018 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010019 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010020 can be in any other unit if the number is suffixed by the unit,
10021 as explained at the top of this document.
10022
10023 The inactivity timeout applies when the client is expected to acknowledge or
10024 send data. In HTTP mode, this timeout is particularly important to consider
10025 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010026 response while it is reading data sent by the server. That said, for the
10027 first phase, it is preferable to set the "timeout http-request" to better
10028 protect HAProxy from Slowloris like attacks. The value is specified in
10029 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010030 suffixed by the unit, as specified at the top of this document. In TCP mode
10031 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10032 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010033 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010034 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010035 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10036 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010037 which overrides "timeout client" and "timeout server" for tunnels, as well as
10038 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010039
10040 This parameter is specific to frontends, but can be specified once for all in
10041 "defaults" sections. This is in fact one of the easiest solutions not to
10042 forget about it. An unspecified timeout results in an infinite timeout, which
10043 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010044 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010045 the system if the system's timeouts are not configured either.
10046
Willy Tarreau95c4e142017-11-26 12:18:55 +010010047 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010048
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010049 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010050
Willy Tarreau0ba27502007-12-24 16:55:16 +010010051
Willy Tarreau05cdd962014-05-10 14:30:07 +020010052timeout client-fin <timeout>
10053 Set the inactivity timeout on the client side for half-closed connections.
10054 May be used in sections : defaults | frontend | listen | backend
10055 yes | yes | yes | no
10056 Arguments :
10057 <timeout> is the timeout value specified in milliseconds by default, but
10058 can be in any other unit if the number is suffixed by the unit,
10059 as explained at the top of this document.
10060
10061 The inactivity timeout applies when the client is expected to acknowledge or
10062 send data while one direction is already shut down. This timeout is different
10063 from "timeout client" in that it only applies to connections which are closed
10064 in one direction. This is particularly useful to avoid keeping connections in
10065 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10066 problem is particularly common long connections such as RDP or WebSocket.
10067 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010068 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10069 frame was sent, often indicating an expectation that the connection quickly
10070 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010071
10072 This parameter is specific to frontends, but can be specified once for all in
10073 "defaults" sections. By default it is not set, so half-closed connections
10074 will use the other timeouts (timeout.client or timeout.tunnel).
10075
10076 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10077
10078
Willy Tarreau0ba27502007-12-24 16:55:16 +010010079timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010080 Set the maximum time to wait for a connection attempt to a server to succeed.
10081 May be used in sections : defaults | frontend | listen | backend
10082 yes | no | yes | yes
10083 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010084 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010085 can be in any other unit if the number is suffixed by the unit,
10086 as explained at the top of this document.
10087
10088 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010089 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010090 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010091 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010092 connect timeout also presets both queue and tarpit timeouts to the same value
10093 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010094
10095 This parameter is specific to backends, but can be specified once for all in
10096 "defaults" sections. This is in fact one of the easiest solutions not to
10097 forget about it. An unspecified timeout results in an infinite timeout, which
10098 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010099 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010100 the system if the system's timeouts are not configured either.
10101
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010102 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010103
Willy Tarreau0ba27502007-12-24 16:55:16 +010010104
Willy Tarreaub16a5742010-01-10 14:46:16 +010010105timeout http-keep-alive <timeout>
10106 Set the maximum allowed time to wait for a new HTTP request to appear
10107 May be used in sections : defaults | frontend | listen | backend
10108 yes | yes | yes | yes
10109 Arguments :
10110 <timeout> is the timeout value specified in milliseconds by default, but
10111 can be in any other unit if the number is suffixed by the unit,
10112 as explained at the top of this document.
10113
10114 By default, the time to wait for a new request in case of keep-alive is set
10115 by "timeout http-request". However this is not always convenient because some
10116 people want very short keep-alive timeouts in order to release connections
10117 faster, and others prefer to have larger ones but still have short timeouts
10118 once the request has started to present itself.
10119
10120 The "http-keep-alive" timeout covers these needs. It will define how long to
10121 wait for a new HTTP request to start coming after a response was sent. Once
10122 the first byte of request has been seen, the "http-request" timeout is used
10123 to wait for the complete request to come. Note that empty lines prior to a
10124 new request do not refresh the timeout and are not counted as a new request.
10125
10126 There is also another difference between the two timeouts : when a connection
10127 expires during timeout http-keep-alive, no error is returned, the connection
10128 just closes. If the connection expires in "http-request" while waiting for a
10129 connection to complete, a HTTP 408 error is returned.
10130
10131 In general it is optimal to set this value to a few tens to hundreds of
10132 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010133 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010134 1 millisecond) it will probably only accept pipelined requests but not the
10135 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010136 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010137
10138 If this parameter is not set, the "http-request" timeout applies, and if both
10139 are not set, "timeout client" still applies at the lower level. It should be
10140 set in the frontend to take effect, unless the frontend is in TCP mode, in
10141 which case the HTTP backend's timeout will be used.
10142
Willy Tarreau95c4e142017-11-26 12:18:55 +010010143 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10144 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010145 (where we only have one connection per client and a connection setup).
10146
Willy Tarreaub16a5742010-01-10 14:46:16 +010010147 See also : "timeout http-request", "timeout client".
10148
10149
Willy Tarreau036fae02008-01-06 13:24:40 +010010150timeout http-request <timeout>
10151 Set the maximum allowed time to wait for a complete HTTP request
10152 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010153 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010154 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010155 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010156 can be in any other unit if the number is suffixed by the unit,
10157 as explained at the top of this document.
10158
10159 In order to offer DoS protection, it may be required to lower the maximum
10160 accepted time to receive a complete HTTP request without affecting the client
10161 timeout. This helps protecting against established connections on which
10162 nothing is sent. The client timeout cannot offer a good protection against
10163 this abuse because it is an inactivity timeout, which means that if the
10164 attacker sends one character every now and then, the timeout will not
10165 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010166 types, the request will be aborted if it does not complete in time. When the
10167 timeout expires, an HTTP 408 response is sent to the client to inform it
10168 about the problem, and the connection is closed. The logs will report
10169 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010170 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010171 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10172 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010173
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010174 By default, this timeout only applies to the header part of the request,
10175 and not to any data. As soon as the empty line is received, this timeout is
10176 not used anymore. When combined with "option http-buffer-request", this
10177 timeout also applies to the body of the request..
10178 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010179 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010180
10181 Generally it is enough to set it to a few seconds, as most clients send the
10182 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010183 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010184 generally work on local networks as long as there are no packet losses. This
10185 will prevent people from sending bare HTTP requests using telnet.
10186
10187 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010188 chunk of the incoming request. It should be set in the frontend to take
10189 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10190 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010191
Willy Tarreau0f228a02015-05-01 15:37:53 +020010192 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010193 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010194
Willy Tarreau844e3c52008-01-11 16:28:18 +010010195
10196timeout queue <timeout>
10197 Set the maximum time to wait in the queue for a connection slot to be free
10198 May be used in sections : defaults | frontend | listen | backend
10199 yes | no | yes | yes
10200 Arguments :
10201 <timeout> is the timeout value specified in milliseconds by default, but
10202 can be in any other unit if the number is suffixed by the unit,
10203 as explained at the top of this document.
10204
10205 When a server's maxconn is reached, connections are left pending in a queue
10206 which may be server-specific or global to the backend. In order not to wait
10207 indefinitely, a timeout is applied to requests pending in the queue. If the
10208 timeout is reached, it is considered that the request will almost never be
10209 served, so it is dropped and a 503 error is returned to the client.
10210
10211 The "timeout queue" statement allows to fix the maximum time for a request to
10212 be left pending in a queue. If unspecified, the same value as the backend's
10213 connection timeout ("timeout connect") is used, for backwards compatibility
10214 with older versions with no "timeout queue" parameter.
10215
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010216 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010217
10218
10219timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010010220 Set the maximum inactivity time on the server side.
10221 May be used in sections : defaults | frontend | listen | backend
10222 yes | no | yes | yes
10223 Arguments :
10224 <timeout> is the timeout value specified in milliseconds by default, but
10225 can be in any other unit if the number is suffixed by the unit,
10226 as explained at the top of this document.
10227
10228 The inactivity timeout applies when the server is expected to acknowledge or
10229 send data. In HTTP mode, this timeout is particularly important to consider
10230 during the first phase of the server's response, when it has to send the
10231 headers, as it directly represents the server's processing time for the
10232 request. To find out what value to put there, it's often good to start with
10233 what would be considered as unacceptable response times, then check the logs
10234 to observe the response time distribution, and adjust the value accordingly.
10235
10236 The value is specified in milliseconds by default, but can be in any other
10237 unit if the number is suffixed by the unit, as specified at the top of this
10238 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10239 recommended that the client timeout remains equal to the server timeout in
10240 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010241 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010242 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010243 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10244 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010245 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10246 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010247
10248 This parameter is specific to backends, but can be specified once for all in
10249 "defaults" sections. This is in fact one of the easiest solutions not to
10250 forget about it. An unspecified timeout results in an infinite timeout, which
10251 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010252 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010010253 the system if the system's timeouts are not configured either.
10254
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010255 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010256
Willy Tarreau05cdd962014-05-10 14:30:07 +020010257
10258timeout server-fin <timeout>
10259 Set the inactivity timeout on the server side for half-closed connections.
10260 May be used in sections : defaults | frontend | listen | backend
10261 yes | no | yes | yes
10262 Arguments :
10263 <timeout> is the timeout value specified in milliseconds by default, but
10264 can be in any other unit if the number is suffixed by the unit,
10265 as explained at the top of this document.
10266
10267 The inactivity timeout applies when the server is expected to acknowledge or
10268 send data while one direction is already shut down. This timeout is different
10269 from "timeout server" in that it only applies to connections which are closed
10270 in one direction. This is particularly useful to avoid keeping connections in
10271 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10272 This problem is particularly common long connections such as RDP or WebSocket.
10273 Note that this timeout can override "timeout tunnel" when a connection shuts
10274 down in one direction. This setting was provided for completeness, but in most
10275 situations, it should not be needed.
10276
10277 This parameter is specific to backends, but can be specified once for all in
10278 "defaults" sections. By default it is not set, so half-closed connections
10279 will use the other timeouts (timeout.server or timeout.tunnel).
10280
10281 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10282
Willy Tarreau844e3c52008-01-11 16:28:18 +010010283
10284timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010285 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010286 May be used in sections : defaults | frontend | listen | backend
10287 yes | yes | yes | yes
10288 Arguments :
10289 <timeout> is the tarpit duration specified in milliseconds by default, but
10290 can be in any other unit if the number is suffixed by the unit,
10291 as explained at the top of this document.
10292
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020010293 When a connection is tarpitted using "http-request tarpit", it is maintained
10294 open with no activity for a certain amount of time, then closed. "timeout
10295 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010296
10297 The value is specified in milliseconds by default, but can be in any other
10298 unit if the number is suffixed by the unit, as specified at the top of this
10299 document. If unspecified, the same value as the backend's connection timeout
10300 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010301 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010302
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010303 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010304
10305
Willy Tarreauce887fd2012-05-12 12:50:00 +020010306timeout tunnel <timeout>
10307 Set the maximum inactivity time on the client and server side for tunnels.
10308 May be used in sections : defaults | frontend | listen | backend
10309 yes | no | yes | yes
10310 Arguments :
10311 <timeout> is the timeout value specified in milliseconds by default, but
10312 can be in any other unit if the number is suffixed by the unit,
10313 as explained at the top of this document.
10314
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010315 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010316 between a client and a server, and the connection remains inactive in both
10317 directions. This timeout supersedes both the client and server timeouts once
10318 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010319 analyzer remains attached to either connection (e.g. tcp content rules are
10320 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010321 when switching to the WebSocket protocol, or forwarding a CONNECT request
10322 to a proxy), or after the first response when no keepalive/close option is
10323 specified.
10324
Willy Tarreau05cdd962014-05-10 14:30:07 +020010325 Since this timeout is usually used in conjunction with long-lived connections,
10326 it usually is a good idea to also set "timeout client-fin" to handle the
10327 situation where a client suddenly disappears from the net and does not
10328 acknowledge a close, or sends a shutdown and does not acknowledge pending
10329 data anymore. This can happen in lossy networks where firewalls are present,
10330 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10331 state.
10332
Willy Tarreauce887fd2012-05-12 12:50:00 +020010333 The value is specified in milliseconds by default, but can be in any other
10334 unit if the number is suffixed by the unit, as specified at the top of this
10335 document. Whatever the expected normal idle time, it is a good practice to
10336 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010337 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010338
10339 This parameter is specific to backends, but can be specified once for all in
10340 "defaults" sections. This is in fact one of the easiest solutions not to
10341 forget about it.
10342
10343 Example :
10344 defaults http
10345 option http-server-close
10346 timeout connect 5s
10347 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010348 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010349 timeout server 30s
10350 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10351
Willy Tarreau05cdd962014-05-10 14:30:07 +020010352 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010353
10354
Willy Tarreau844e3c52008-01-11 16:28:18 +010010355transparent (deprecated)
10356 Enable client-side transparent proxying
10357 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010358 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010359 Arguments : none
10360
10361 This keyword was introduced in order to provide layer 7 persistence to layer
10362 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10363 connection for a remote address to a local process (here HAProxy), and let
10364 this process know what address was initially requested. When this option is
10365 used, sessions without cookies will be forwarded to the original destination
10366 IP address of the incoming request (which should match that of another
10367 equipment), while requests with cookies will still be forwarded to the
10368 appropriate server.
10369
10370 The "transparent" keyword is deprecated, use "option transparent" instead.
10371
10372 Note that contrary to a common belief, this option does NOT make HAProxy
10373 present the client's IP to the server when establishing the connection.
10374
Willy Tarreau844e3c52008-01-11 16:28:18 +010010375 See also: "option transparent"
10376
William Lallemanda73203e2012-03-12 12:48:57 +010010377unique-id-format <string>
10378 Generate a unique ID for each request.
10379 May be used in sections : defaults | frontend | listen | backend
10380 yes | yes | yes | no
10381 Arguments :
10382 <string> is a log-format string.
10383
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010384 This keyword creates a ID for each request using the custom log format. A
10385 unique ID is useful to trace a request passing through many components of
10386 a complex infrastructure. The newly created ID may also be logged using the
10387 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010388
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010389 The format should be composed from elements that are guaranteed to be
10390 unique when combined together. For instance, if multiple haproxy instances
10391 are involved, it might be important to include the node name. It is often
10392 needed to log the incoming connection's source and destination addresses
10393 and ports. Note that since multiple requests may be performed over the same
10394 connection, including a request counter may help differentiate them.
10395 Similarly, a timestamp may protect against a rollover of the counter.
10396 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010397
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010398 It is recommended to use hexadecimal notation for many fields since it
10399 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010400
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010401 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010402
Julien Vehentf21be322014-03-07 08:27:34 -050010403 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010404
10405 will generate:
10406
10407 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10408
10409 See also: "unique-id-header"
10410
10411unique-id-header <name>
10412 Add a unique ID header in the HTTP request.
10413 May be used in sections : defaults | frontend | listen | backend
10414 yes | yes | yes | no
10415 Arguments :
10416 <name> is the name of the header.
10417
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010418 Add a unique-id header in the HTTP request sent to the server, using the
10419 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010420
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010421 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010422
Julien Vehentf21be322014-03-07 08:27:34 -050010423 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010424 unique-id-header X-Unique-ID
10425
10426 will generate:
10427
10428 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10429
10430 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010431
Willy Tarreauf51658d2014-04-23 01:21:56 +020010432use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010433 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010434 May be used in sections : defaults | frontend | listen | backend
10435 no | yes | yes | no
10436 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010437 <backend> is the name of a valid backend or "listen" section, or a
10438 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010439
Willy Tarreauf51658d2014-04-23 01:21:56 +020010440 <condition> is a condition composed of ACLs, as described in section 7. If
10441 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010442
10443 When doing content-switching, connections arrive on a frontend and are then
10444 dispatched to various backends depending on a number of conditions. The
10445 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010446 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010447 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010448 source address validation) or combined with a "tcp-request" rule to wait for
10449 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010450
10451 There may be as many "use_backend" rules as desired. All of these rules are
10452 evaluated in their declaration order, and the first one which matches will
10453 assign the backend.
10454
10455 In the first form, the backend will be used if the condition is met. In the
10456 second form, the backend will be used if the condition is not met. If no
10457 condition is valid, the backend defined with "default_backend" will be used.
10458 If no default backend is defined, either the servers in the same section are
10459 used (in case of a "listen" section) or, in case of a frontend, no server is
10460 used and a 503 service unavailable response is returned.
10461
Willy Tarreau51aecc72009-07-12 09:47:04 +020010462 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010463 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010464 and backend processing will immediately follow, or the backend will wait for
10465 a complete HTTP request to get in. This feature is useful when a frontend
10466 must decode several protocols on a unique port, one of them being HTTP.
10467
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010468 When <backend> is a simple name, it is resolved at configuration time, and an
10469 error is reported if the specified backend does not exist. If <backend> is
10470 a log-format string instead, no check may be done at configuration time, so
10471 the backend name is resolved dynamically at run time. If the resulting
10472 backend name does not correspond to any valid backend, no other rule is
10473 evaluated, and the default_backend directive is applied instead. Note that
10474 when using dynamic backend names, it is highly recommended to use a prefix
10475 that no other backend uses in order to ensure that an unauthorized backend
10476 cannot be forced from the request.
10477
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010478 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010479 used to detect the association between frontends and backends to compute the
10480 backend's "fullconn" setting. This cannot be done for dynamic names.
10481
10482 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10483 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010484
Christopher Fauletb30b3102019-09-12 23:03:09 +020010485use-fcgi-app <name>
10486 Defines the FastCGI application to use for the backend.
10487 May be used in sections : defaults | frontend | listen | backend
10488 no | no | yes | yes
10489 Arguments :
10490 <name> is the name of the FastCGI application to use.
10491
10492 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010010493
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010494use-server <server> if <condition>
10495use-server <server> unless <condition>
10496 Only use a specific server if/unless an ACL-based condition is matched.
10497 May be used in sections : defaults | frontend | listen | backend
10498 no | no | yes | yes
10499 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010500 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010501
10502 <condition> is a condition composed of ACLs, as described in section 7.
10503
10504 By default, connections which arrive to a backend are load-balanced across
10505 the available servers according to the configured algorithm, unless a
10506 persistence mechanism such as a cookie is used and found in the request.
10507
10508 Sometimes it is desirable to forward a particular request to a specific
10509 server without having to declare a dedicated backend for this server. This
10510 can be achieved using the "use-server" rules. These rules are evaluated after
10511 the "redirect" rules and before evaluating cookies, and they have precedence
10512 on them. There may be as many "use-server" rules as desired. All of these
10513 rules are evaluated in their declaration order, and the first one which
10514 matches will assign the server.
10515
10516 If a rule designates a server which is down, and "option persist" is not used
10517 and no force-persist rule was validated, it is ignored and evaluation goes on
10518 with the next rules until one matches.
10519
10520 In the first form, the server will be used if the condition is met. In the
10521 second form, the server will be used if the condition is not met. If no
10522 condition is valid, the processing continues and the server will be assigned
10523 according to other persistence mechanisms.
10524
10525 Note that even if a rule is matched, cookie processing is still performed but
10526 does not assign the server. This allows prefixed cookies to have their prefix
10527 stripped.
10528
10529 The "use-server" statement works both in HTTP and TCP mode. This makes it
10530 suitable for use with content-based inspection. For instance, a server could
10531 be selected in a farm according to the TLS SNI field. And if these servers
10532 have their weight set to zero, they will not be used for other traffic.
10533
10534 Example :
10535 # intercept incoming TLS requests based on the SNI field
10536 use-server www if { req_ssl_sni -i www.example.com }
10537 server www 192.168.0.1:443 weight 0
10538 use-server mail if { req_ssl_sni -i mail.example.com }
10539 server mail 192.168.0.1:587 weight 0
10540 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010541 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010542 # all the rest is forwarded to this server
10543 server default 192.168.0.2:443 check
10544
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010545 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010546
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010547
Davor Ocelice9ed2812017-12-25 17:49:28 +0100105485. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010549--------------------------
10550
10551The "bind", "server" and "default-server" keywords support a number of settings
10552depending on some build options and on the system HAProxy was built on. These
10553settings generally each consist in one word sometimes followed by a value,
10554written on the same line as the "bind" or "server" line. All these options are
10555described in this section.
10556
10557
105585.1. Bind options
10559-----------------
10560
10561The "bind" keyword supports a certain number of settings which are all passed
10562as arguments on the same line. The order in which those arguments appear makes
10563no importance, provided that they appear after the bind address. All of these
10564parameters are optional. Some of them consist in a single words (booleans),
10565while other ones expect a value after them. In this case, the value must be
10566provided immediately after the setting name.
10567
10568The currently supported settings are the following ones.
10569
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010570accept-netscaler-cip <magic number>
10571 Enforces the use of the NetScaler Client IP insertion protocol over any
10572 connection accepted by any of the TCP sockets declared on the same line. The
10573 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10574 the incoming connection to be used everywhere an address is used, with the
10575 only exception of "tcp-request connection" rules which will only see the
10576 real connection address. Logs will reflect the addresses indicated in the
10577 protocol, unless it is violated, in which case the real address will still
10578 be used. This keyword combined with support from external components can be
10579 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010580 mechanism which is not always reliable and not even always usable. See also
10581 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10582 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010583
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010584accept-proxy
10585 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010586 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10587 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010588 3/4 addresses of the incoming connection to be used everywhere an address is
10589 used, with the only exception of "tcp-request connection" rules which will
10590 only see the real connection address. Logs will reflect the addresses
10591 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010592 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010593 components can be used as an efficient and reliable alternative to the
10594 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010595 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10596 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010597
Olivier Houchardc2aae742017-09-22 18:26:28 +020010598allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010599 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010010600 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050010601 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010010602 that are idempotent. You can use the "wait-for-handshake" action for any
10603 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020010604
Willy Tarreauab861d32013-04-02 02:30:41 +020010605alpn <protocols>
10606 This enables the TLS ALPN extension and advertises the specified protocol
10607 list as supported on top of ALPN. The protocol list consists in a comma-
10608 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050010609 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020010610 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010611 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10612 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10613 now obsolete NPN extension. At the time of writing this, most browsers still
10614 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10615 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10616 are expected to be supported, both versions can be advertised, in order of
10617 preference, like below :
10618
10619 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010620
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010621backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010010622 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010623 backlog is used instead, which generally defaults to the maxconn value.
10624
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010625curves <curves>
10626 This setting is only available when support for OpenSSL was built in. It sets
10627 the string describing the list of elliptic curves algorithms ("curve suite")
10628 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10629 string is a colon-delimited list of curve name.
10630 Example: "X25519:P-256" (without quote)
10631 When "curves" is set, "ecdhe" parameter is ignored.
10632
Emeric Brun7fb34422012-09-28 15:26:15 +020010633ecdhe <named curve>
10634 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010635 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10636 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010637
Emeric Brunfd33a262012-10-11 16:28:27 +020010638ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010639 This setting is only available when support for OpenSSL was built in. It
10640 designates a PEM file from which to load CA certificates used to verify
10641 client's certificate.
10642
Emeric Brunb6dc9342012-09-28 17:55:37 +020010643ca-ignore-err [all|<errorID>,...]
10644 This setting is only available when support for OpenSSL was built in.
10645 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10646 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10647 error is ignored.
10648
Christopher Faulet31af49d2015-06-09 17:29:50 +020010649ca-sign-file <cafile>
10650 This setting is only available when support for OpenSSL was built in. It
10651 designates a PEM file containing both the CA certificate and the CA private
10652 key used to create and sign server's certificates. This is a mandatory
10653 setting when the dynamic generation of certificates is enabled. See
10654 'generate-certificates' for details.
10655
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010656ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010657 This setting is only available when support for OpenSSL was built in. It is
10658 the CA private key passphrase. This setting is optional and used only when
10659 the dynamic generation of certificates is enabled. See
10660 'generate-certificates' for details.
10661
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010662ciphers <ciphers>
10663 This setting is only available when support for OpenSSL was built in. It sets
10664 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000010665 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010666 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010667 information and recommendations see e.g.
10668 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10669 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10670 cipher configuration, please check the "ciphersuites" keyword.
10671
10672ciphersuites <ciphersuites>
10673 This setting is only available when support for OpenSSL was built in and
10674 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10675 the list of cipher algorithms ("cipher suite") that are negotiated during the
10676 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010677 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
10678 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010679
Emeric Brunfd33a262012-10-11 16:28:27 +020010680crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010681 This setting is only available when support for OpenSSL was built in. It
10682 designates a PEM file from which to load certificate revocation list used
10683 to verify client's certificate.
10684
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010685crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010686 This setting is only available when support for OpenSSL was built in. It
10687 designates a PEM file containing both the required certificates and any
10688 associated private keys. This file can be built by concatenating multiple
10689 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10690 requires an intermediate certificate, this can also be concatenated into this
10691 file.
10692
10693 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10694 are loaded.
10695
10696 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010697 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010698 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10699 specified multiple times in order to load certificates from multiple files or
10700 directories. The certificates will be presented to clients who provide a
10701 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010702 subjects. Wildcards are supported, where a wildcard character '*' is used
10703 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010704 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010705
10706 If no SNI is provided by the client or if the SSL library does not support
10707 TLS extensions, or if the client provides an SNI hostname which does not
10708 match any certificate, then the first loaded certificate will be presented.
10709 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010710 recommended to load the default one first as a file or to ensure that it will
10711 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010712
Emeric Brune032bfa2012-09-28 13:01:45 +020010713 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010714
Davor Ocelice9ed2812017-12-25 17:49:28 +010010715 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010716 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010717 choose a web server that the CA believes requires an intermediate CA (for
10718 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010719 others, e.g. nginx, result in a wrong bundle that will not work for some
10720 clients).
10721
Emeric Brun4147b2e2014-06-16 18:36:30 +020010722 For each PEM file, haproxy checks for the presence of file at the same path
10723 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10724 Status Request extension (also known as "OCSP stapling") is automatically
10725 enabled. The content of this file is optional. If not empty, it must contain
10726 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10727 must comply with the following rules: it has to indicate a good status,
10728 it has to be a single response for the certificate of the PEM file, and it
10729 has to be valid at the moment of addition. If these rules are not respected
10730 the OCSP Response is ignored and a warning is emitted. In order to identify
10731 which certificate an OCSP Response applies to, the issuer's certificate is
10732 necessary. If the issuer's certificate is not found in the PEM file, it will
10733 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10734 if it exists otherwise it will fail with an error.
10735
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010736 For each PEM file, haproxy also checks for the presence of file at the same
10737 path suffixed by ".sctl". If such file is found, support for Certificate
10738 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10739 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10740 to check basic syntax, but no signatures are verified.
10741
yanbzhu6c25e9e2016-01-05 12:52:02 -050010742 There are cases where it is desirable to support multiple key types, e.g. RSA
10743 and ECDSA in the cipher suites offered to the clients. This allows clients
10744 that support EC certificates to be able to use EC ciphers, while
10745 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010746
10747 In order to provide this functionality, multiple PEM files, each with a
10748 different key type, are required. To associate these PEM files into a
10749 "cert bundle" that is recognized by haproxy, they must be named in the
10750 following way: All PEM files that are to be bundled must have the same base
10751 name, with a suffix indicating the key type. Currently, three suffixes are
10752 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10753 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10754 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10755 suffix matters. To load this bundle into haproxy, specify the base name only:
10756
10757 Example : bind :8443 ssl crt example.pem
10758
yanbzhu6c25e9e2016-01-05 12:52:02 -050010759 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010760 a cert bundle.
10761
Davor Ocelice9ed2812017-12-25 17:49:28 +010010762 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010763 support multiple key types. PEM files are combined based on Common Name
10764 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10765 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10766 entries in the certificates in that bundle, haproxy will not be able to
10767 provide multi-cert support.
10768
10769 Assuming bundle in the example above contained the following:
10770
10771 Filename | CN | SAN
10772 -------------------+-----------------+-------------------
10773 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010774 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010775 example.pem.ecdsa | www.example.com | ecdsa.example.com
10776 -------------------+-----------------+-------------------
10777
10778 Users connecting with an SNI of "www.example.com" will be able
10779 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10780 "rsa.example.com" will only be able to use RSA cipher suites, and users
10781 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010782 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10783 no need to bundle certificates. ECDSA certificate will be preferred if client
10784 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010785
10786 If a directory name is given as the <cert> argument, haproxy will
10787 automatically search and load bundled files in that directory.
10788
10789 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10790 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10791 time, sctl is not supported in multi-certificate bundling.
10792
Emeric Brunb6dc9342012-09-28 17:55:37 +020010793crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010794 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010795 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010796 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010797 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010798
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010799crt-list <file>
10800 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010801 designates a list of PEM file with an optional ssl configuration and a SNI
10802 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010803
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010804 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10805
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010806 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10807 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010808 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010809 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010810
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010811 Wildcards are supported in the SNI filter. Negative filter are also supported,
10812 only useful in combination with a wildcard filter to exclude a particular SNI.
10813 The certificates will be presented to clients who provide a valid TLS Server
10814 Name Indication field matching one of the SNI filters. If no SNI filter is
10815 specified, the CN and alt subjects are used. This directive may be specified
10816 multiple times. See the "crt" option for more information. The default
10817 certificate is still needed to meet OpenSSL expectations. If it is not used,
10818 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010819
yanbzhu6c25e9e2016-01-05 12:52:02 -050010820 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010821 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010822 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10823 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10824 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010825
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010826 crt-list file example:
10827 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010828 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010829 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010830 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010831
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010832defer-accept
10833 Is an optional keyword which is supported only on certain Linux kernels. It
10834 states that a connection will only be accepted once some data arrive on it,
10835 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010010836 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010837 performance by ensuring that most of the request is already available when
10838 the connection is accepted. On the other hand, it will not be able to detect
10839 connections which don't talk. It is important to note that this option is
10840 broken in all kernels up to 2.6.31, as the connection is never accepted until
10841 the client talks. This can cause issues with front firewalls which would see
10842 an established connection while the proxy will only see it in SYN_RECV. This
10843 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
10844
William Lallemandf6975e92017-05-26 17:42:10 +020010845expose-fd listeners
10846 This option is only usable with the stats socket. It gives your stats socket
10847 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020010848 During a reload with the master-worker mode, the process is automatically
10849 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010010850 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020010851
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010852force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010853 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010854 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010855 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010856 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010857
10858force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010859 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010860 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010861 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010862
10863force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010864 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010865 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010866 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010867
10868force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010869 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010870 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010871 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010872
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010873force-tlsv13
10874 This option enforces use of TLSv1.3 only on SSL connections instantiated from
10875 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010876 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010877
Christopher Faulet31af49d2015-06-09 17:29:50 +020010878generate-certificates
10879 This setting is only available when support for OpenSSL was built in. It
10880 enables the dynamic SSL certificates generation. A CA certificate and its
10881 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
10882 a transparent forward proxy, SSL requests generate errors because of a common
10883 name mismatch on the certificate presented to the client. With this option
10884 enabled, HAProxy will try to forge a certificate using the SNI hostname
10885 indicated by the client. This is done only if no certificate matches the SNI
10886 hostname (see 'crt-list'). If an error occurs, the default certificate is
10887 used, else the 'strict-sni' option is set.
10888 It can also be used when HAProxy is configured as a reverse proxy to ease the
10889 deployment of an architecture with many backends.
10890
10891 Creating a SSL certificate is an expensive operation, so a LRU cache is used
10892 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010010893 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020010894 certificate is used many times.
10895
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010896gid <gid>
10897 Sets the group of the UNIX sockets to the designated system gid. It can also
10898 be set by default in the global section's "unix-bind" statement. Note that
10899 some platforms simply ignore this. This setting is equivalent to the "group"
10900 setting except that the group ID is used instead of its name. This setting is
10901 ignored by non UNIX sockets.
10902
10903group <group>
10904 Sets the group of the UNIX sockets to the designated system group. It can
10905 also be set by default in the global section's "unix-bind" statement. Note
10906 that some platforms simply ignore this. This setting is equivalent to the
10907 "gid" setting except that the group name is used instead of its gid. This
10908 setting is ignored by non UNIX sockets.
10909
10910id <id>
10911 Fixes the socket ID. By default, socket IDs are automatically assigned, but
10912 sometimes it is more convenient to fix them to ease monitoring. This value
10913 must be strictly positive and unique within the listener/frontend. This
10914 option can only be used when defining only a single socket.
10915
10916interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010010917 Restricts the socket to a specific interface. When specified, only packets
10918 received from that particular interface are processed by the socket. This is
10919 currently only supported on Linux. The interface must be a primary system
10920 interface, not an aliased interface. It is also possible to bind multiple
10921 frontends to the same address if they are bound to different interfaces. Note
10922 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010010923 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
10924 uses the same interface as inbound traffic, and its associated routing table,
10925 even if there are explicit routes through different interfaces configured.
10926 This can prove useful to address asymmetric routing issues when the same
10927 client IP addresses need to be able to reach frontends hosted on different
10928 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010929
Willy Tarreauabb175f2012-09-24 12:43:26 +020010930level <level>
10931 This setting is used with the stats sockets only to restrict the nature of
10932 the commands that can be issued on the socket. It is ignored by other
10933 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010010934 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020010935 read, and no change is allowed. It would make sense on systems where it
10936 is not easy to restrict access to the socket.
10937 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010938 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020010939 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010010940 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020010941 all counters).
10942
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020010943severity-output <format>
10944 This setting is used with the stats sockets only to configure severity
10945 level output prepended to informational feedback messages. Severity
10946 level of messages can range between 0 and 7, conforming to syslog
10947 rfc5424. Valid and successful socket commands requesting data
10948 (i.e. "show map", "get acl foo" etc.) will never have a severity level
10949 prepended. It is ignored by other sockets. <format> can be one of :
10950 - "none" (default) no severity level is prepended to feedback messages.
10951 - "number" severity level is prepended as a number.
10952 - "string" severity level is prepended as a string following the
10953 rfc5424 convention.
10954
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010955maxconn <maxconn>
10956 Limits the sockets to this number of concurrent connections. Extraneous
10957 connections will remain in the system's backlog until a connection is
10958 released. If unspecified, the limit will be the same as the frontend's
10959 maxconn. Note that in case of port ranges or multiple addresses, the same
10960 value will be applied to each socket. This setting enables different
10961 limitations on expensive sockets, for instance SSL entries which may easily
10962 eat all memory.
10963
10964mode <mode>
10965 Sets the octal mode used to define access permissions on the UNIX socket. It
10966 can also be set by default in the global section's "unix-bind" statement.
10967 Note that some platforms simply ignore this. This setting is ignored by non
10968 UNIX sockets.
10969
10970mss <maxseg>
10971 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
10972 connections. This can be used to force a lower MSS for certain specific
10973 ports, for instance for connections passing through a VPN. Note that this
10974 relies on a kernel feature which is theoretically supported under Linux but
10975 was buggy in all versions prior to 2.6.28. It may or may not work on other
10976 operating systems. It may also not change the advertised value but change the
10977 effective size of outgoing segments. The commonly advertised value for TCPv4
10978 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
10979 positive, it will be used as the advertised MSS. If it is negative, it will
10980 indicate by how much to reduce the incoming connection's advertised MSS for
10981 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
10982
10983name <name>
10984 Sets an optional name for these sockets, which will be reported on the stats
10985 page.
10986
Willy Tarreaud72f0f32015-10-13 14:50:22 +020010987namespace <name>
10988 On Linux, it is possible to specify which network namespace a socket will
10989 belong to. This directive makes it possible to explicitly bind a listener to
10990 a namespace different from the default one. Please refer to your operating
10991 system's documentation to find more details about network namespaces.
10992
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010993nice <nice>
10994 Sets the 'niceness' of connections initiated from the socket. Value must be
10995 in the range -1024..1024 inclusive, and defaults to zero. Positive values
10996 means that such connections are more friendly to others and easily offer
10997 their place in the scheduler. On the opposite, negative values mean that
10998 connections want to run with a higher priority than others. The difference
10999 only happens under high loads when the system is close to saturation.
11000 Negative values are appropriate for low-latency or administration services,
11001 and high values are generally recommended for CPU intensive tasks such as SSL
11002 processing or bulk transfers which are less sensible to latency. For example,
11003 it may make sense to use a positive value for an SMTP socket and a negative
11004 one for an RDP socket.
11005
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011006no-ca-names
11007 This setting is only available when support for OpenSSL was built in. It
11008 prevents from send CA names in server hello message when ca-file is used.
11009
Emeric Brun9b3009b2012-10-05 11:55:06 +020011010no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011011 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011012 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011013 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011014 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011015 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11016 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011017
Emeric Brun90ad8722012-10-02 14:00:59 +020011018no-tls-tickets
11019 This setting is only available when support for OpenSSL was built in. It
11020 disables the stateless session resumption (RFC 5077 TLS Ticket
11021 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011022 session resumption is more expensive in CPU usage. This option is also
11023 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011024
Emeric Brun9b3009b2012-10-05 11:55:06 +020011025no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011026 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011027 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011028 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011029 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011030 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11031 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011032
Emeric Brun9b3009b2012-10-05 11:55:06 +020011033no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011034 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011035 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011036 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011037 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011038 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11039 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011040
Emeric Brun9b3009b2012-10-05 11:55:06 +020011041no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011042 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011043 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011044 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011045 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011046 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11047 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011048
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011049no-tlsv13
11050 This setting is only available when support for OpenSSL was built in. It
11051 disables support for TLSv1.3 on any sockets instantiated from the listener
11052 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11053 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011054 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11055 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011056
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011057npn <protocols>
11058 This enables the NPN TLS extension and advertises the specified protocol list
11059 as supported on top of NPN. The protocol list consists in a comma-delimited
11060 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050011061 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011062 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011063 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11064 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11065 version of OpenSSL, NPN might still be used as most clients still support it
11066 at the time of writing this. It is possible to enable both NPN and ALPN
11067 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011068
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011069prefer-client-ciphers
11070 Use the client's preference when selecting the cipher suite, by default
11071 the server's preference is enforced. This option is also available on
11072 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011073 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11074 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11075 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011076
Christopher Fauletc644fa92017-11-23 22:44:11 +010011077process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010011078 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010011079 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011080 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011081 between the two is applied. If in the end the listener is not allowed to run
11082 on any remaining process, a warning is emitted, and the listener will either
11083 run on the first process of the listener if a single process was specified,
11084 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011085 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010011086 connections for this listener, for the the process set. If multiple processes
11087 and threads are configured, a warning is emitted, as it either results from a
11088 configuration error or a misunderstanding of these models. For the unlikely
11089 case where several ranges are needed, this directive may be repeated.
11090 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010011091
11092 all | odd | even | number[-[number]]
11093
11094 Ranges can be partially defined. The higher bound can be omitted. In such
11095 case, it is replaced by the corresponding maximum value. The main purpose of
11096 this directive is to be used with the stats sockets and have one different
11097 socket per process. The second purpose is to have multiple bind lines sharing
11098 the same IP:port but not the same process in a listener, so that the system
11099 can distribute the incoming connections into multiple queues and allow a
11100 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11101 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011102
Christopher Fauleta717b992018-04-10 14:43:00 +020011103proto <name>
11104 Forces the multiplexer's protocol to use for the incoming connections. It
11105 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11106 be usable on the frontend side. The list of available protocols is reported
11107 in haproxy -vv.
11108 Idea behind this optipon is to bypass the selection of the best multiplexer's
11109 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011110 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011111 h2" on the bind line.
11112
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011113ssl
11114 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011115 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011116 certificate is necessary (see "crt" above). All contents in the buffers will
11117 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011118 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11119 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011120
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011121ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11122 This option enforces use of <version> or lower on SSL connections instantiated
11123 from this listener. This option is also available on global statement
11124 "ssl-default-bind-options". See also "ssl-min-ver".
11125
11126ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11127 This option enforces use of <version> or upper on SSL connections instantiated
11128 from this listener. This option is also available on global statement
11129 "ssl-default-bind-options". See also "ssl-max-ver".
11130
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011131strict-sni
11132 This setting is only available when support for OpenSSL was built in. The
11133 SSL/TLS negotiation is allow only if the client provided an SNI which match
11134 a certificate. The default certificate is not used.
11135 See the "crt" option for more information.
11136
Willy Tarreau2af207a2015-02-04 00:45:58 +010011137tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011138 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011139 listening socket. This option is available on Linux since version 2.6.37. It
11140 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011141 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011142 useful on long-lived connections experiencing long idle periods such as
11143 remote terminals or database connection pools, where the client and server
11144 timeouts must remain high to allow a long period of idle, but where it is
11145 important to detect that the client has disappeared in order to release all
11146 resources associated with its connection (and the server's session). The
11147 argument is a delay expressed in milliseconds by default. This only works
11148 for regular TCP connections, and is ignored for other protocols.
11149
Willy Tarreau1c862c52012-10-05 16:21:00 +020011150tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011151 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011152 enables TCP Fast Open on the listening socket, which means that clients which
11153 support this feature will be able to send a request and receive a response
11154 during the 3-way handshake starting from second connection, thus saving one
11155 round-trip after the first connection. This only makes sense with protocols
11156 that use high connection rates and where each round trip matters. This can
11157 possibly cause issues with many firewalls which do not accept data on SYN
11158 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011159 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11160 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11161 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011162
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011163tls-ticket-keys <keyfile>
11164 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010011165 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
11166 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
11167 The first key determines the key length used for next keys: you can't mix
11168 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
11169 build option (default 3) and at least as many keys need to be present in
11170 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
11171 penultimate one for encryption. This enables easy key rotation by just
11172 appending new key to the file and reloading the process. Keys must be
11173 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
11174 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011175 storage such as hard drives (hint: use tmpfs and don't swap those files).
11176 Lifetime hint can be changed using tune.ssl.timeout.
11177
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011178transparent
11179 Is an optional keyword which is supported only on certain Linux kernels. It
11180 indicates that the addresses will be bound even if they do not belong to the
11181 local machine, and that packets targeting any of these addresses will be
11182 intercepted just as if the addresses were locally configured. This normally
11183 requires that IP forwarding is enabled. Caution! do not use this with the
11184 default address '*', as it would redirect any traffic for the specified port.
11185 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11186 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11187 kernel version. Some distribution kernels include backports of the feature,
11188 so check for support with your vendor.
11189
Willy Tarreau77e3af92012-11-24 15:07:23 +010011190v4v6
11191 Is an optional keyword which is supported only on most recent systems
11192 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11193 and IPv6 when it uses the default address. Doing so is sometimes necessary
11194 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011195 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011196
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011197v6only
11198 Is an optional keyword which is supported only on most recent systems
11199 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11200 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011201 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11202 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011203
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011204uid <uid>
11205 Sets the owner of the UNIX sockets to the designated system uid. It can also
11206 be set by default in the global section's "unix-bind" statement. Note that
11207 some platforms simply ignore this. This setting is equivalent to the "user"
11208 setting except that the user numeric ID is used instead of its name. This
11209 setting is ignored by non UNIX sockets.
11210
11211user <user>
11212 Sets the owner of the UNIX sockets to the designated system user. It can also
11213 be set by default in the global section's "unix-bind" statement. Note that
11214 some platforms simply ignore this. This setting is equivalent to the "uid"
11215 setting except that the user name is used instead of its uid. This setting is
11216 ignored by non UNIX sockets.
11217
Emeric Brun1a073b42012-09-28 17:07:34 +020011218verify [none|optional|required]
11219 This setting is only available when support for OpenSSL was built in. If set
11220 to 'none', client certificate is not requested. This is the default. In other
11221 cases, a client certificate is requested. If the client does not provide a
11222 certificate after the request and if 'verify' is set to 'required', then the
11223 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011224 certificate provided by the client is always verified using CAs from
11225 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11226 is aborted, regardless of the 'verify' option, unless the error code exactly
11227 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011228
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200112295.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011230------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011231
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011232The "server" and "default-server" keywords support a certain number of settings
11233which are all passed as arguments on the server line. The order in which those
11234arguments appear does not count, and they are all optional. Some of those
11235settings are single words (booleans) while others expect one or several values
11236after them. In this case, the values must immediately follow the setting name.
11237Except default-server, all those settings must be specified after the server's
11238address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011239
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011240 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011241 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011242
Frédéric Lécailled2376272017-03-21 18:52:12 +010011243Note that all these settings are supported both by "server" and "default-server"
11244keywords, except "id" which is only supported by "server".
11245
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011246The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011247
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011248addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011249 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011250 to send health-checks or to probe the agent-check. On some servers, it may be
11251 desirable to dedicate an IP address to specific component able to perform
11252 complex tests which are more suitable to health-checks than the application.
11253 This parameter is ignored if the "check" parameter is not set. See also the
11254 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011255
Simon Hormand60d6912013-11-25 10:46:36 +090011256agent-check
11257 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011258 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010011259 to the port set by the "agent-port" parameter and reading an ASCII string
11260 terminated by the first '\r' or '\n' met. The string is made of a series of
11261 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011262
Willy Tarreau81f5d942013-12-09 20:51:51 +010011263 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011264 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011265 weight of a server as configured when haproxy starts. Note that a zero
11266 weight is reported on the stats page as "DRAIN" since it has the same
11267 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011268
Davor Ocelice9ed2812017-12-25 17:49:28 +010011269 - The string "maxconn:" followed by an integer (no space between). Values
11270 in this format will set the maxconn of a server. The maximum number of
11271 connections advertised needs to be multiplied by the number of load
11272 balancers and different backends that use this health check to get the
11273 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011274
Willy Tarreau81f5d942013-12-09 20:51:51 +010011275 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011276 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011277
Willy Tarreau81f5d942013-12-09 20:51:51 +010011278 - The word "drain". This will turn the server's administrative state to the
11279 DRAIN mode, thus it will not accept any new connections other than those
11280 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011281
Willy Tarreau81f5d942013-12-09 20:51:51 +010011282 - The word "maint". This will turn the server's administrative state to the
11283 MAINT mode, thus it will not accept any new connections at all, and health
11284 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011285
Willy Tarreau81f5d942013-12-09 20:51:51 +010011286 - The words "down", "failed", or "stopped", optionally followed by a
11287 description string after a sharp ('#'). All of these mark the server's
11288 operating state as DOWN, but since the word itself is reported on the stats
11289 page, the difference allows an administrator to know if the situation was
11290 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011291 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011292 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011293
Willy Tarreau81f5d942013-12-09 20:51:51 +010011294 - The word "up" sets back the server's operating state as UP if health checks
11295 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011296
Willy Tarreau81f5d942013-12-09 20:51:51 +010011297 Parameters which are not advertised by the agent are not changed. For
11298 example, an agent might be designed to monitor CPU usage and only report a
11299 relative weight and never interact with the operating status. Similarly, an
11300 agent could be designed as an end-user interface with 3 radio buttons
11301 allowing an administrator to change only the administrative state. However,
11302 it is important to consider that only the agent may revert its own actions,
11303 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11304 agent must implement the other equivalent actions to bring the service into
11305 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011306
Simon Horman2f1f9552013-11-25 10:46:37 +090011307 Failure to connect to the agent is not considered an error as connectivity
11308 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011309 parameter. Warning though, it is not a good idea to stop an agent after it
11310 reports "down", since only an agent reporting "up" will be able to turn the
11311 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011312 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011313
Willy Tarreau81f5d942013-12-09 20:51:51 +010011314 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011315 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011316
James Brown55f9ff12015-10-21 18:19:05 -070011317agent-send <string>
11318 If this option is specified, haproxy will send the given string (verbatim)
11319 to the agent server upon connection. You could, for example, encode
11320 the backend name into this string, which would enable your agent to send
11321 different responses based on the backend. Make sure to include a '\n' if
11322 you want to terminate your request with a newline.
11323
Simon Hormand60d6912013-11-25 10:46:36 +090011324agent-inter <delay>
11325 The "agent-inter" parameter sets the interval between two agent checks
11326 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11327
11328 Just as with every other time-based parameter, it may be entered in any
11329 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11330 parameter also serves as a timeout for agent checks "timeout check" is
11331 not set. In order to reduce "resonance" effects when multiple servers are
11332 hosted on the same hardware, the agent and health checks of all servers
11333 are started with a small time offset between them. It is also possible to
11334 add some random noise in the agent and health checks interval using the
11335 global "spread-checks" keyword. This makes sense for instance when a lot
11336 of backends use the same servers.
11337
11338 See also the "agent-check" and "agent-port" parameters.
11339
Misiek768d8602017-01-09 09:52:43 +010011340agent-addr <addr>
11341 The "agent-addr" parameter sets address for agent check.
11342
11343 You can offload agent-check to another target, so you can make single place
11344 managing status and weights of servers defined in haproxy in case you can't
11345 make self-aware and self-managing services. You can specify both IP or
11346 hostname, it will be resolved.
11347
Simon Hormand60d6912013-11-25 10:46:36 +090011348agent-port <port>
11349 The "agent-port" parameter sets the TCP port used for agent checks.
11350
11351 See also the "agent-check" and "agent-inter" parameters.
11352
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020011353allow-0rtt
11354 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020011355 Note that early data will be sent only if the client used early data, or
11356 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020011357
Olivier Houchardc7566002018-11-20 23:33:50 +010011358alpn <protocols>
11359 This enables the TLS ALPN extension and advertises the specified protocol
11360 list as supported on top of ALPN. The protocol list consists in a comma-
11361 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050011362 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010011363 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
11364 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
11365 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11366 now obsolete NPN extension.
11367 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
11368 be advertised, in order of preference, like below :
11369
11370 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
11371
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011372backup
11373 When "backup" is present on a server line, the server is only used in load
11374 balancing when all other non-backup servers are unavailable. Requests coming
11375 with a persistence cookie referencing the server will always be served
11376 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011377 the "allbackups" option is set in the backend. See also the "no-backup" and
11378 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011379
Emeric Brunef42d922012-10-11 16:11:36 +020011380ca-file <cafile>
11381 This setting is only available when support for OpenSSL was built in. It
11382 designates a PEM file from which to load CA certificates used to verify
11383 server's certificate.
11384
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011385check
11386 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011387 always considered available. If "check" is set, the server is available when
11388 accepting periodic TCP connections, to ensure that it is really able to serve
11389 requests. The default address and port to send the tests to are those of the
11390 server, and the default source is the same as the one defined in the
11391 backend. It is possible to change the address using the "addr" parameter, the
11392 port using the "port" parameter, the source address using the "source"
11393 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011394 parameters. The request method is define in the backend using the "httpchk",
11395 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011396 refer to those options and parameters for more information. See also
11397 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011398
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011399check-send-proxy
11400 This option forces emission of a PROXY protocol line with outgoing health
11401 checks, regardless of whether the server uses send-proxy or not for the
11402 normal traffic. By default, the PROXY protocol is enabled for health checks
11403 if it is already enabled for normal traffic and if no "port" nor "addr"
11404 directive is present. However, if such a directive is present, the
11405 "check-send-proxy" option needs to be used to force the use of the
11406 protocol. See also the "send-proxy" option for more information.
11407
Olivier Houchard92150142018-12-21 19:47:01 +010011408check-alpn <protocols>
11409 Defines which protocols to advertise with ALPN. The protocol list consists in
11410 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
11411 (without quotes). If it is not set, the server ALPN is used.
11412
Jérôme Magninae9bb762018-12-09 16:08:26 +010011413check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020011414 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010011415 over SSL. It is only possible to use a string to set <sni>. If you want to
11416 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020011417
Willy Tarreau763a95b2012-10-04 23:15:39 +020011418check-ssl
11419 This option forces encryption of all health checks over SSL, regardless of
11420 whether the server uses SSL or not for the normal traffic. This is generally
11421 used when an explicit "port" or "addr" directive is specified and SSL health
11422 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011423 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011424 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11425 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011426 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011427 See the "ssl" option for more information and "no-check-ssl" to disable
11428 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011429
Alexander Liu2a54bb72019-05-22 19:44:48 +080011430check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050011431 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080011432 default, the health checks won't go through socks tunnel even it was enabled
11433 for normal traffic.
11434
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011435ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011436 This setting is only available when support for OpenSSL was built in. This
11437 option sets the string describing the list of cipher algorithms that is
11438 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011439 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
11440 information and recommendations see e.g.
11441 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
11442 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
11443 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011444
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011445ciphersuites <ciphersuites>
11446 This setting is only available when support for OpenSSL was built in and
11447 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11448 describing the list of cipher algorithms that is negotiated during the TLS
11449 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011450 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
11451 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
11452 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011453
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011454cookie <value>
11455 The "cookie" parameter sets the cookie value assigned to the server to
11456 <value>. This value will be checked in incoming requests, and the first
11457 operational server possessing the same value will be selected. In return, in
11458 cookie insertion or rewrite modes, this value will be assigned to the cookie
11459 sent to the client. There is nothing wrong in having several servers sharing
11460 the same cookie value, and it is in fact somewhat common between normal and
11461 backup servers. See also the "cookie" keyword in backend section.
11462
Emeric Brunef42d922012-10-11 16:11:36 +020011463crl-file <crlfile>
11464 This setting is only available when support for OpenSSL was built in. It
11465 designates a PEM file from which to load certificate revocation list used
11466 to verify server's certificate.
11467
Emeric Bruna7aa3092012-10-26 12:58:00 +020011468crt <cert>
11469 This setting is only available when support for OpenSSL was built in.
11470 It designates a PEM file from which to load both a certificate and the
11471 associated private key. This file can be built by concatenating both PEM
11472 files into one. This certificate will be sent if the server send a client
11473 certificate request.
11474
Willy Tarreau96839092010-03-29 10:02:24 +020011475disabled
11476 The "disabled" keyword starts the server in the "disabled" state. That means
11477 that it is marked down in maintenance mode, and no connection other than the
11478 ones allowed by persist mode will reach it. It is very well suited to setup
11479 new servers, because normal traffic will never reach them, while it is still
11480 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011481 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011482
Frédéric Lécailled2376272017-03-21 18:52:12 +010011483enabled
11484 This option may be used as 'server' setting to reset any 'disabled'
11485 setting which would have been inherited from 'default-server' directive as
11486 default value.
11487 It may also be used as 'default-server' setting to reset any previous
11488 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011489
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011490error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011491 If health observing is enabled, the "error-limit" parameter specifies the
11492 number of consecutive errors that triggers event selected by the "on-error"
11493 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011494
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011495 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011496
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011497fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011498 The "fall" parameter states that a server will be considered as dead after
11499 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11500 unspecified. See also the "check", "inter" and "rise" parameters.
11501
Emeric Brun8694b9a2012-10-05 14:39:07 +020011502force-sslv3
11503 This option enforces use of SSLv3 only when SSL is used to communicate with
11504 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011505 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011506 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011507
11508force-tlsv10
11509 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011510 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011511 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011512
11513force-tlsv11
11514 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011515 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011516 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011517
11518force-tlsv12
11519 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011520 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011521 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011522
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011523force-tlsv13
11524 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11525 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011526 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011527
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011528id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011529 Set a persistent ID for the server. This ID must be positive and unique for
11530 the proxy. An unused ID will automatically be assigned if unset. The first
11531 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011532
Willy Tarreau6a031d12016-11-07 19:42:35 +010011533init-addr {last | libc | none | <ip>},[...]*
11534 Indicate in what order the server's address should be resolved upon startup
11535 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011536 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011537 method which succeeds is used. If the end of the list is reached without
11538 finding a working method, an error is thrown. Method "last" suggests to pick
11539 the address which appears in the state file (see "server-state-file"). Method
11540 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11541 depending on the operating system and build options). Method "none"
11542 specifically indicates that the server should start without any valid IP
11543 address in a down state. It can be useful to ignore some DNS issues upon
11544 startup, waiting for the situation to get fixed later. Finally, an IP address
11545 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011546 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011547 server used to catch old sessions and present them with a decent error
11548 message for example. When the "first" load balancing algorithm is used, this
11549 IP address could point to a fake server used to trigger the creation of new
11550 instances on the fly. This option defaults to "last,libc" indicating that the
11551 previous address found in the state file (if any) is used first, otherwise
11552 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011553 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011554
11555 Example:
11556 defaults
11557 # never fail on address resolution
11558 default-server init-addr last,libc,none
11559
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011560inter <delay>
11561fastinter <delay>
11562downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011563 The "inter" parameter sets the interval between two consecutive health checks
11564 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11565 It is also possible to use "fastinter" and "downinter" to optimize delays
11566 between checks depending on the server state :
11567
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011568 Server state | Interval used
11569 ----------------------------------------+----------------------------------
11570 UP 100% (non-transitional) | "inter"
11571 ----------------------------------------+----------------------------------
11572 Transitionally UP (going down "fall"), | "fastinter" if set,
11573 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11574 or yet unchecked. |
11575 ----------------------------------------+----------------------------------
11576 DOWN 100% (non-transitional) | "downinter" if set,
11577 | "inter" otherwise.
11578 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011579
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011580 Just as with every other time-based parameter, they can be entered in any
11581 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11582 serves as a timeout for health checks sent to servers if "timeout check" is
11583 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011584 hosted on the same hardware, the agent and health checks of all servers
11585 are started with a small time offset between them. It is also possible to
11586 add some random noise in the agent and health checks interval using the
11587 global "spread-checks" keyword. This makes sense for instance when a lot
11588 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011589
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011590maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011591 The "maxconn" parameter specifies the maximal number of concurrent
11592 connections that will be sent to this server. If the number of incoming
11593 concurrent requests goes higher than this value, they will be queued, waiting
11594 for a connection to be released. This parameter is very important as it can
11595 save fragile servers from going down under extreme loads. If a "minconn"
11596 parameter is specified, the limit becomes dynamic. The default value is "0"
11597 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11598 the backend's "fullconn" keyword.
11599
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011600maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011601 The "maxqueue" parameter specifies the maximal number of connections which
11602 will wait in the queue for this server. If this limit is reached, next
11603 requests will be redispatched to other servers instead of indefinitely
11604 waiting to be served. This will break persistence but may allow people to
11605 quickly re-log in when the server they try to connect to is dying. The
11606 default value is "0" which means the queue is unlimited. See also the
11607 "maxconn" and "minconn" parameters.
11608
Willy Tarreau9c538e02019-01-23 10:21:49 +010011609max-reuse <count>
11610 The "max-reuse" argument indicates the HTTP connection processors that they
11611 should not reuse a server connection more than this number of times to send
11612 new requests. Permitted values are -1 (the default), which disables this
11613 limit, or any positive value. Value zero will effectively disable keep-alive.
11614 This is only used to work around certain server bugs which cause them to leak
11615 resources over time. The argument is not necessarily respected by the lower
11616 layers as there might be technical limitations making it impossible to
11617 enforce. At least HTTP/2 connections to servers will respect it.
11618
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011619minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011620 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11621 limit following the backend's load. The server will always accept at least
11622 <minconn> connections, never more than <maxconn>, and the limit will be on
11623 the ramp between both values when the backend has less than <fullconn>
11624 concurrent connections. This makes it possible to limit the load on the
11625 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011626 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011627 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011628
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011629namespace <name>
11630 On Linux, it is possible to specify which network namespace a socket will
11631 belong to. This directive makes it possible to explicitly bind a server to
11632 a namespace different from the default one. Please refer to your operating
11633 system's documentation to find more details about network namespaces.
11634
Frédéric Lécailled2376272017-03-21 18:52:12 +010011635no-agent-check
11636 This option may be used as "server" setting to reset any "agent-check"
11637 setting which would have been inherited from "default-server" directive as
11638 default value.
11639 It may also be used as "default-server" setting to reset any previous
11640 "default-server" "agent-check" setting.
11641
11642no-backup
11643 This option may be used as "server" setting to reset any "backup"
11644 setting which would have been inherited from "default-server" directive as
11645 default value.
11646 It may also be used as "default-server" setting to reset any previous
11647 "default-server" "backup" setting.
11648
11649no-check
11650 This option may be used as "server" setting to reset any "check"
11651 setting which would have been inherited from "default-server" directive as
11652 default value.
11653 It may also be used as "default-server" setting to reset any previous
11654 "default-server" "check" setting.
11655
11656no-check-ssl
11657 This option may be used as "server" setting to reset any "check-ssl"
11658 setting which would have been inherited from "default-server" directive as
11659 default value.
11660 It may also be used as "default-server" setting to reset any previous
11661 "default-server" "check-ssl" setting.
11662
Frédéric Lécailled2376272017-03-21 18:52:12 +010011663no-send-proxy
11664 This option may be used as "server" setting to reset any "send-proxy"
11665 setting which would have been inherited from "default-server" directive as
11666 default value.
11667 It may also be used as "default-server" setting to reset any previous
11668 "default-server" "send-proxy" setting.
11669
11670no-send-proxy-v2
11671 This option may be used as "server" setting to reset any "send-proxy-v2"
11672 setting which would have been inherited from "default-server" directive as
11673 default value.
11674 It may also be used as "default-server" setting to reset any previous
11675 "default-server" "send-proxy-v2" setting.
11676
11677no-send-proxy-v2-ssl
11678 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11679 setting which would have been inherited from "default-server" directive as
11680 default value.
11681 It may also be used as "default-server" setting to reset any previous
11682 "default-server" "send-proxy-v2-ssl" setting.
11683
11684no-send-proxy-v2-ssl-cn
11685 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11686 setting which would have been inherited from "default-server" directive as
11687 default value.
11688 It may also be used as "default-server" setting to reset any previous
11689 "default-server" "send-proxy-v2-ssl-cn" setting.
11690
11691no-ssl
11692 This option may be used as "server" setting to reset any "ssl"
11693 setting which would have been inherited from "default-server" directive as
11694 default value.
11695 It may also be used as "default-server" setting to reset any previous
11696 "default-server" "ssl" setting.
11697
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011698no-ssl-reuse
11699 This option disables SSL session reuse when SSL is used to communicate with
11700 the server. It will force the server to perform a full handshake for every
11701 new connection. It's probably only useful for benchmarking, troubleshooting,
11702 and for paranoid users.
11703
Emeric Brun9b3009b2012-10-05 11:55:06 +020011704no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011705 This option disables support for SSLv3 when SSL is used to communicate with
11706 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011707 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011708
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011709 Supported in default-server: No
11710
Emeric Brunf9c5c472012-10-11 15:28:34 +020011711no-tls-tickets
11712 This setting is only available when support for OpenSSL was built in. It
11713 disables the stateless session resumption (RFC 5077 TLS Ticket
11714 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011715 session resumption is more expensive in CPU usage for servers. This option
11716 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011717 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011718
Emeric Brun9b3009b2012-10-05 11:55:06 +020011719no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011720 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011721 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11722 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011723 often makes sense to disable it when communicating with local servers. This
11724 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011725 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011726
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011727 Supported in default-server: No
11728
Emeric Brun9b3009b2012-10-05 11:55:06 +020011729no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011730 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011731 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11732 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011733 often makes sense to disable it when communicating with local servers. This
11734 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011735 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011736
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011737 Supported in default-server: No
11738
Emeric Brun9b3009b2012-10-05 11:55:06 +020011739no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011740 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011741 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11742 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011743 often makes sense to disable it when communicating with local servers. This
11744 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011745 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011746
11747 Supported in default-server: No
11748
11749no-tlsv13
11750 This option disables support for TLSv1.3 when SSL is used to communicate with
11751 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11752 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11753 often makes sense to disable it when communicating with local servers. This
11754 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011755 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011756
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011757 Supported in default-server: No
11758
Frédéric Lécailled2376272017-03-21 18:52:12 +010011759no-verifyhost
11760 This option may be used as "server" setting to reset any "verifyhost"
11761 setting which would have been inherited from "default-server" directive as
11762 default value.
11763 It may also be used as "default-server" setting to reset any previous
11764 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011765
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020011766no-tfo
11767 This option may be used as "server" setting to reset any "tfo"
11768 setting which would have been inherited from "default-server" directive as
11769 default value.
11770 It may also be used as "default-server" setting to reset any previous
11771 "default-server" "tfo" setting.
11772
Simon Hormanfa461682011-06-25 09:39:49 +090011773non-stick
11774 Never add connections allocated to this sever to a stick-table.
11775 This may be used in conjunction with backup to ensure that
11776 stick-table persistence is disabled for backup servers.
11777
Olivier Houchardc7566002018-11-20 23:33:50 +010011778npn <protocols>
11779 This enables the NPN TLS extension and advertises the specified protocol list
11780 as supported on top of NPN. The protocol list consists in a comma-delimited
11781 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050011782 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010011783 enabled (check with haproxy -vv). Note that the NPN extension has been
11784 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11785 only available starting with OpenSSL 1.0.2.
11786
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011787observe <mode>
11788 This option enables health adjusting based on observing communication with
11789 the server. By default this functionality is disabled and enabling it also
11790 requires to enable health checks. There are two supported modes: "layer4" and
11791 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11792 significant. In layer7, which is only allowed for http proxies, responses
11793 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011794 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011795
11796 See also the "check", "on-error" and "error-limit".
11797
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011798on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011799 Select what should happen when enough consecutive errors are detected.
11800 Currently, four modes are available:
11801 - fastinter: force fastinter
11802 - fail-check: simulate a failed check, also forces fastinter (default)
11803 - sudden-death: simulate a pre-fatal failed health check, one more failed
11804 check will mark a server down, forces fastinter
11805 - mark-down: mark the server immediately down and force fastinter
11806
11807 See also the "check", "observe" and "error-limit".
11808
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011809on-marked-down <action>
11810 Modify what occurs when a server is marked down.
11811 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011812 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11813 all connections to the server are immediately terminated when the server
11814 goes down. It might be used if the health check detects more complex cases
11815 than a simple connection status, and long timeouts would cause the service
11816 to remain unresponsive for too long a time. For instance, a health check
11817 might detect that a database is stuck and that there's no chance to reuse
11818 existing connections anymore. Connections killed this way are logged with
11819 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011820
11821 Actions are disabled by default
11822
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011823on-marked-up <action>
11824 Modify what occurs when a server is marked up.
11825 Currently one action is available:
11826 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11827 done only if the server is not in backup state and if it is not disabled
11828 (it must have an effective weight > 0). This can be used sometimes to force
11829 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011830 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11831 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011832 with extreme care. Sessions killed because a server comes up are logged
11833 with an 'U' termination code (for "Up").
11834
11835 Actions are disabled by default
11836
Olivier Houchard006e3102018-12-10 18:30:32 +010011837pool-max-conn <max>
11838 Set the maximum number of idling connections for a server. -1 means unlimited
11839 connections, 0 means no idle connections. The default is -1. When idle
11840 connections are enabled, orphaned idle connections which do not belong to any
11841 client session anymore are moved to a dedicated pool so that they remain
11842 usable by future clients. This only applies to connections that can be shared
11843 according to the same principles as those applying to "http-reuse".
11844
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010011845pool-purge-delay <delay>
11846 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010011847 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020011848 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010011849
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011850port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011851 Using the "port" parameter, it becomes possible to use a different port to
11852 send health-checks. On some servers, it may be desirable to dedicate a port
11853 to a specific component able to perform complex tests which are more suitable
11854 to health-checks than the application. It is common to run a simple script in
11855 inetd for instance. This parameter is ignored if the "check" parameter is not
11856 set. See also the "addr" parameter.
11857
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020011858proto <name>
11859
11860 Forces the multiplexer's protocol to use for the outgoing connections to this
11861 server. It must be compatible with the mode of the backend (TCP or HTTP). It
11862 must also be usable on the backend side. The list of available protocols is
11863 reported in haproxy -vv.
11864 Idea behind this optipon is to bypass the selection of the best multiplexer's
11865 protocol for all connections established to this server.
11866
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011867redir <prefix>
11868 The "redir" parameter enables the redirection mode for all GET and HEAD
11869 requests addressing this server. This means that instead of having HAProxy
11870 forward the request to the server, it will send an "HTTP 302" response with
11871 the "Location" header composed of this prefix immediately followed by the
11872 requested URI beginning at the leading '/' of the path component. That means
11873 that no trailing slash should be used after <prefix>. All invalid requests
11874 will be rejected, and all non-GET or HEAD requests will be normally served by
11875 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011876 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011877 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011878 users to a remote location in case of local disaster. Main use consists in
11879 increasing bandwidth for static servers by having the clients directly
11880 connect to them. Note: never use a relative location here, it would cause a
11881 loop between the client and HAProxy!
11882
11883 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
11884
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011885rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011886 The "rise" parameter states that a server will be considered as operational
11887 after <count> consecutive successful health checks. This value defaults to 2
11888 if unspecified. See also the "check", "inter" and "fall" parameters.
11889
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020011890resolve-opts <option>,<option>,...
11891 Comma separated list of options to apply to DNS resolution linked to this
11892 server.
11893
11894 Available options:
11895
11896 * allow-dup-ip
11897 By default, HAProxy prevents IP address duplication in a backend when DNS
11898 resolution at runtime is in operation.
11899 That said, for some cases, it makes sense that two servers (in the same
11900 backend, being resolved by the same FQDN) have the same IP address.
11901 For such case, simply enable this option.
11902 This is the opposite of prevent-dup-ip.
11903
11904 * prevent-dup-ip
11905 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
11906 an IP address already set to a server in the same backend and sharing the
11907 same fqdn.
11908 This is the opposite of allow-dup-ip.
11909
11910 Example:
11911 backend b_myapp
11912 default-server init-addr none resolvers dns
11913 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
11914 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
11915
11916 With the option allow-dup-ip set:
11917 * if the nameserver returns a single IP address, then both servers will use
11918 it
11919 * If the nameserver returns 2 IP addresses, then each server will pick up a
11920 different address
11921
11922 Default value: not set
11923
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011924resolve-prefer <family>
11925 When DNS resolution is enabled for a server and multiple IP addresses from
11926 different families are returned, HAProxy will prefer using an IP address
11927 from the family mentioned in the "resolve-prefer" parameter.
11928 Available families: "ipv4" and "ipv6"
11929
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020011930 Default value: ipv6
11931
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011932 Example:
11933
11934 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011935
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011936resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050011937 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011938 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010011939 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010011940 different datacenters. The latency between datacenter is not negligible, so
11941 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011942 configured network, another address is selected.
11943
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011944 Example:
11945
11946 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011947
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011948resolvers <id>
11949 Points to an existing "resolvers" section to resolve current server's
11950 hostname.
11951
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011952 Example:
11953
11954 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011955
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011956 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011957
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011958send-proxy
11959 The "send-proxy" parameter enforces use of the PROXY protocol over any
11960 connection established to this server. The PROXY protocol informs the other
11961 end about the layer 3/4 addresses of the incoming connection, so that it can
11962 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011963 upper layer protocol. For connections accepted by an "accept-proxy" or
11964 "accept-netscaler-cip" listener, the advertised address will be used. Only
11965 TCPv4 and TCPv6 address families are supported. Other families such as
11966 Unix sockets, will report an UNKNOWN family. Servers using this option can
11967 fully be chained to another instance of haproxy listening with an
11968 "accept-proxy" setting. This setting must not be used if the server isn't
11969 aware of the protocol. When health checks are sent to the server, the PROXY
11970 protocol is automatically used when this option is set, unless there is an
11971 explicit "port" or "addr" directive, in which case an explicit
11972 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011973 See also the "no-send-proxy" option of this section and "accept-proxy" and
11974 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011975
David Safb76832014-05-08 23:42:08 -040011976send-proxy-v2
11977 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
11978 over any connection established to this server. The PROXY protocol informs
11979 the other end about the layer 3/4 addresses of the incoming connection, so
11980 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020011981 whatever the upper layer protocol. It also send ALPN information if an alpn
11982 have been negotiated. This setting must not be used if the server isn't aware
11983 of this version of the protocol. See also the "no-send-proxy-v2" option of
11984 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011985
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010011986proxy-v2-options <option>[,<option>]*
11987 The "proxy-v2-options" parameter add option to send in PROXY protocol version
11988 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010011989 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
11990 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010011991 certificate, "cert-key": key algorithm of the used certificate), "authority":
11992 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010011993 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010011994
David Safb76832014-05-08 23:42:08 -040011995send-proxy-v2-ssl
11996 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11997 2 over any connection established to this server. The PROXY protocol informs
11998 the other end about the layer 3/4 addresses of the incoming connection, so
11999 that it can know the client's address or the public address it accessed to,
12000 whatever the upper layer protocol. In addition, the SSL information extension
12001 of the PROXY protocol is added to the PROXY protocol header. This setting
12002 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012003 See also the "no-send-proxy-v2-ssl" option of this section and the
12004 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012005
12006send-proxy-v2-ssl-cn
12007 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12008 2 over any connection established to this server. The PROXY protocol informs
12009 the other end about the layer 3/4 addresses of the incoming connection, so
12010 that it can know the client's address or the public address it accessed to,
12011 whatever the upper layer protocol. In addition, the SSL information extension
12012 of the PROXY protocol, along along with the Common Name from the subject of
12013 the client certificate (if any), is added to the PROXY protocol header. This
12014 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012015 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12016 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012017
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012018slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012019 The "slowstart" parameter for a server accepts a value in milliseconds which
12020 indicates after how long a server which has just come back up will run at
12021 full speed. Just as with every other time-based parameter, it can be entered
12022 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12023 linearly from 0 to 100% during this time. The limitation applies to two
12024 parameters :
12025
12026 - maxconn: the number of connections accepted by the server will grow from 1
12027 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12028
12029 - weight: when the backend uses a dynamic weighted algorithm, the weight
12030 grows linearly from 1 to 100%. In this case, the weight is updated at every
12031 health-check. For this reason, it is important that the "inter" parameter
12032 is smaller than the "slowstart", in order to maximize the number of steps.
12033
12034 The slowstart never applies when haproxy starts, otherwise it would cause
12035 trouble to running servers. It only applies when a server has been previously
12036 seen as failed.
12037
Willy Tarreau732eac42015-07-09 11:40:25 +020012038sni <expression>
12039 The "sni" parameter evaluates the sample fetch expression, converts it to a
12040 string and uses the result as the host name sent in the SNI TLS extension to
12041 the server. A typical use case is to send the SNI received from the client in
12042 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012043 expression, though alternatives such as req.hdr(host) can also make sense. If
12044 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012045 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012046 "verify" directive for more details. If you want to set a SNI for health
12047 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012048
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012049source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012050source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012051source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012052 The "source" parameter sets the source address which will be used when
12053 connecting to the server. It follows the exact same parameters and principle
12054 as the backend "source" keyword, except that it only applies to the server
12055 referencing it. Please consult the "source" keyword for details.
12056
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012057 Additionally, the "source" statement on a server line allows one to specify a
12058 source port range by indicating the lower and higher bounds delimited by a
12059 dash ('-'). Some operating systems might require a valid IP address when a
12060 source port range is specified. It is permitted to have the same IP/range for
12061 several servers. Doing so makes it possible to bypass the maximum of 64k
12062 total concurrent connections. The limit will then reach 64k connections per
12063 server.
12064
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012065 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12066 specifying the source address without port(s).
12067
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012068ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012069 This option enables SSL ciphering on outgoing connections to the server. It
12070 is critical to verify server certificates using "verify" when using SSL to
12071 connect to servers, otherwise the communication is prone to trivial man in
12072 the-middle attacks rendering SSL useless. When this option is used, health
12073 checks are automatically sent in SSL too unless there is a "port" or an
12074 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012075 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12076 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012077
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012078ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12079 This option enforces use of <version> or lower when SSL is used to communicate
12080 with the server. This option is also available on global statement
12081 "ssl-default-server-options". See also "ssl-min-ver".
12082
12083ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12084 This option enforces use of <version> or upper when SSL is used to communicate
12085 with the server. This option is also available on global statement
12086 "ssl-default-server-options". See also "ssl-max-ver".
12087
Frédéric Lécailled2376272017-03-21 18:52:12 +010012088ssl-reuse
12089 This option may be used as "server" setting to reset any "no-ssl-reuse"
12090 setting which would have been inherited from "default-server" directive as
12091 default value.
12092 It may also be used as "default-server" setting to reset any previous
12093 "default-server" "no-ssl-reuse" setting.
12094
12095stick
12096 This option may be used as "server" setting to reset any "non-stick"
12097 setting which would have been inherited from "default-server" directive as
12098 default value.
12099 It may also be used as "default-server" setting to reset any previous
12100 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012101
Alexander Liu2a54bb72019-05-22 19:44:48 +080012102socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050012103 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080012104 server. Using this option won't force the health check to go via socks4 by
12105 default. You will have to use the keyword "check-via-socks4" to enable it.
12106
Willy Tarreau163d4622015-10-13 16:16:41 +020012107tcp-ut <delay>
12108 Sets the TCP User Timeout for all outgoing connections to this server. This
12109 option is available on Linux since version 2.6.37. It allows haproxy to
12110 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012111 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012112 long-lived connections experiencing long idle periods such as remote
12113 terminals or database connection pools, where the client and server timeouts
12114 must remain high to allow a long period of idle, but where it is important to
12115 detect that the server has disappeared in order to release all resources
12116 associated with its connection (and the client's session). One typical use
12117 case is also to force dead server connections to die when health checks are
12118 too slow or during a soft reload since health checks are then disabled. The
12119 argument is a delay expressed in milliseconds by default. This only works for
12120 regular TCP connections, and is ignored for other protocols.
12121
Willy Tarreau034c88c2017-01-23 23:36:45 +010012122tfo
12123 This option enables using TCP fast open when connecting to servers, on
12124 systems that support it (currently only the Linux kernel >= 4.11).
12125 See the "tfo" bind option for more information about TCP fast open.
12126 Please note that when using tfo, you should also use the "conn-failure",
12127 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020012128 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010012129
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012130track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012131 This option enables ability to set the current state of the server by tracking
12132 another one. It is possible to track a server which itself tracks another
12133 server, provided that at the end of the chain, a server has health checks
12134 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012135 used, it has to be enabled on both proxies.
12136
Frédéric Lécailled2376272017-03-21 18:52:12 +010012137tls-tickets
12138 This option may be used as "server" setting to reset any "no-tls-tickets"
12139 setting which would have been inherited from "default-server" directive as
12140 default value.
12141 It may also be used as "default-server" setting to reset any previous
12142 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012143
Emeric Brunef42d922012-10-11 16:11:36 +020012144verify [none|required]
12145 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012146 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012147 certificate provided by the server is verified using CAs from 'ca-file' and
12148 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012149 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012150 the name passed using the "sni" directive, or if not provided, the static
12151 host name passed using the "verifyhost" directive. When no name is found, the
12152 certificate's names are ignored. For this reason, without SNI it's important
12153 to use "verifyhost". On verification failure the handshake is aborted. It is
12154 critically important to verify server certificates when using SSL to connect
12155 to servers, otherwise the communication is prone to trivial man-in-the-middle
12156 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12157 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012158
Evan Broderbe554312013-06-27 00:05:25 -070012159verifyhost <hostname>
12160 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012161 only takes effect if 'verify required' is also specified. This directive sets
12162 a default static hostname to check the server's certificate against when no
12163 SNI was used to connect to the server. If SNI is not used, this is the only
12164 way to enable hostname verification. This static hostname, when set, will
12165 also be used for health checks (which cannot provide an SNI value). If none
12166 of the hostnames in the certificate match the specified hostname, the
12167 handshake is aborted. The hostnames in the server-provided certificate may
12168 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012169
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012170weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012171 The "weight" parameter is used to adjust the server's weight relative to
12172 other servers. All servers will receive a load proportional to their weight
12173 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012174 load. The default weight is 1, and the maximal value is 256. A value of 0
12175 means the server will not participate in load-balancing but will still accept
12176 persistent connections. If this parameter is used to distribute the load
12177 according to server's capacity, it is recommended to start with values which
12178 can both grow and shrink, for instance between 10 and 100 to leave enough
12179 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012180
12181
Cyril Bonté46175dd2015-07-02 22:45:32 +0200121825.3. Server IP address resolution using DNS
12183-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012184
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012185HAProxy allows using a host name on the server line to retrieve its IP address
12186using name servers. By default, HAProxy resolves the name when parsing the
12187configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012188This is not sufficient in some cases, such as in Amazon where a server's IP
12189can change after a reboot or an ELB Virtual IP can change based on current
12190workload.
12191This chapter describes how HAProxy can be configured to process server's name
12192resolution at run time.
12193Whether run time server name resolution has been enable or not, HAProxy will
12194carry on doing the first resolution when parsing the configuration.
12195
12196
Cyril Bonté46175dd2015-07-02 22:45:32 +0200121975.3.1. Global overview
12198----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012199
12200As we've seen in introduction, name resolution in HAProxy occurs at two
12201different steps of the process life:
12202
12203 1. when starting up, HAProxy parses the server line definition and matches a
12204 host name. It uses libc functions to get the host name resolved. This
12205 resolution relies on /etc/resolv.conf file.
12206
Christopher Faulet67957bd2017-09-27 11:00:59 +020012207 2. at run time, HAProxy performs periodically name resolutions for servers
12208 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012209
12210A few other events can trigger a name resolution at run time:
12211 - when a server's health check ends up in a connection timeout: this may be
12212 because the server has a new IP address. So we need to trigger a name
12213 resolution to know this new IP.
12214
Christopher Faulet67957bd2017-09-27 11:00:59 +020012215When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012216HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012217SRV label is specified, then the corresponding SRV records will be retrieved
12218from the DNS server, and the provided hostnames will be used. The SRV label
12219will be checked periodically, and if any server are added or removed, haproxy
12220will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012221
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012222A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050012223 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012224 first valid response.
12225
12226 - a resolution is considered as invalid (NX, timeout, refused), when all the
12227 servers return an error.
12228
12229
Cyril Bonté46175dd2015-07-02 22:45:32 +0200122305.3.2. The resolvers section
12231----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012232
12233This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012234HAProxy. There can be as many as resolvers section as needed. Each section can
12235contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012236
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012237When multiple name servers are configured in a resolvers section, then HAProxy
12238uses the first valid response. In case of invalid responses, only the last one
12239is treated. Purpose is to give the chance to a slow server to deliver a valid
12240answer after a fast faulty or outdated server.
12241
12242When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012243used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012244
Christopher Faulet67957bd2017-09-27 11:00:59 +020012245 1. HAProxy retries the same DNS query with a new query type. The A queries are
12246 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12247 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012248
Christopher Faulet67957bd2017-09-27 11:00:59 +020012249 2. When the fallback on the query type was done (or not applicable), HAProxy
12250 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012251
Christopher Faulet67957bd2017-09-27 11:00:59 +020012252 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12253 response is received after that, it stops the DNS resolution and reports
12254 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012255
Christopher Faulet67957bd2017-09-27 11:00:59 +020012256For example, with 2 name servers configured in a resolvers section, the
12257following scenarios are possible:
12258
12259 - First response is valid and is applied directly, second response is
12260 ignored
12261
12262 - First response is invalid and second one is valid, then second response is
12263 applied
12264
12265 - First response is a NX domain and second one a truncated response, then
12266 HAProxy retries the query with a new type
12267
12268 - First response is a NX domain and second one is a timeout, then HAProxy
12269 retries the query with a new type
12270
12271 - Query timed out for both name servers, then HAProxy retries it with the
12272 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012273
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012274As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12275a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012276<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012277
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012278
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012279resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012280 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012281
12282A resolvers section accept the following parameters:
12283
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012284accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012285 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012286 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012287 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12288 by RFC 6891)
12289
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012290 Note: the maximum allowed value is 8192.
12291
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012292nameserver <id> <ip>:<port>
12293 DNS server description:
12294 <id> : label of the server, should be unique
12295 <ip> : IP address of the server
12296 <port> : port where the DNS service actually runs
12297
Ben Draut44e609b2018-05-29 15:40:08 -060012298parse-resolv-conf
12299 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12300 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12301 placed in the resolvers section in place of this directive.
12302
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012303hold <status> <period>
12304 Defines <period> during which the last name resolution should be kept based
12305 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012306 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012307 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012308 <period> : interval between two successive name resolution when the last
12309 answer was in <status>. It follows the HAProxy time format.
12310 <period> is in milliseconds by default.
12311
Baptiste Assmann686408b2017-08-18 10:15:42 +020012312 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012313
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012314resolve_retries <nb>
12315 Defines the number <nb> of queries to send to resolve a server name before
12316 giving up.
12317 Default value: 3
12318
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012319 A retry occurs on name server timeout or when the full sequence of DNS query
12320 type failover is over and we need to start up from the default ANY query
12321 type.
12322
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012323timeout <event> <time>
12324 Defines timeouts related to name resolution
12325 <event> : the event on which the <time> timeout period applies to.
12326 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012327 - resolve : default time to trigger name resolutions when no
12328 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012329 Default value: 1s
12330 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012331 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012332 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012333 <time> : time related to the event. It follows the HAProxy time format.
12334 <time> is expressed in milliseconds.
12335
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012336 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012337
12338 resolvers mydns
12339 nameserver dns1 10.0.0.1:53
12340 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012341 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012342 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012343 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012344 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012345 hold other 30s
12346 hold refused 30s
12347 hold nx 30s
12348 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012349 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012350 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012351
12352
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200123536. HTTP header manipulation
12354---------------------------
12355
12356In HTTP mode, it is possible to rewrite, add or delete some of the request and
12357response headers based on regular expressions. It is also possible to block a
12358request or a response if a particular header matches a regular expression,
12359which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010012360against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012361
Willy Tarreau70dffda2014-01-30 03:07:23 +010012362If HAProxy encounters an "Informational Response" (status code 1xx), it is able
12363to process all rsp* rules which can allow, deny, rewrite or delete a header,
12364but it will refuse to add a header to any such messages as this is not
12365HTTP-compliant. The reason for still processing headers in such responses is to
12366stop and/or fix any possible information leak which may happen, for instance
12367because another downstream equipment would unconditionally add a header, or if
12368a server name appears there. When such messages are seen, normal processing
12369still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020012370
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012371This section covers common usage of the following keywords, described in detail
12372in section 4.2 :
12373
12374 - reqadd <string>
12375 - reqallow <search>
12376 - reqiallow <search>
12377 - reqdel <search>
12378 - reqidel <search>
12379 - reqdeny <search>
12380 - reqideny <search>
12381 - reqpass <search>
12382 - reqipass <search>
12383 - reqrep <search> <replace>
12384 - reqirep <search> <replace>
12385 - reqtarpit <search>
12386 - reqitarpit <search>
12387 - rspadd <string>
12388 - rspdel <search>
12389 - rspidel <search>
12390 - rspdeny <search>
12391 - rspideny <search>
12392 - rsprep <search> <replace>
12393 - rspirep <search> <replace>
12394
12395With all these keywords, the same conventions are used. The <search> parameter
12396is a POSIX extended regular expression (regex) which supports grouping through
12397parenthesis (without the backslash). Spaces and other delimiters must be
12398prefixed with a backslash ('\') to avoid confusion with a field delimiter.
12399Other characters may be prefixed with a backslash to change their meaning :
12400
12401 \t for a tab
12402 \r for a carriage return (CR)
12403 \n for a new line (LF)
12404 \ to mark a space and differentiate it from a delimiter
12405 \# to mark a sharp and differentiate it from a comment
12406 \\ to use a backslash in a regex
12407 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
12408 \xXX to write the ASCII hex code XX as in the C language
12409
12410The <replace> parameter contains the string to be used to replace the largest
12411portion of text matching the regex. It can make use of the special characters
12412above, and can reference a substring which is delimited by parenthesis in the
12413regex, by writing a backslash ('\') immediately followed by one digit from 0 to
124149 indicating the group position (0 designating the entire line). This practice
12415is very common to users of the "sed" program.
12416
12417The <string> parameter represents the string which will systematically be added
12418after the last header line. It can also use special character sequences above.
12419
12420Notes related to these keywords :
12421---------------------------------
12422 - these keywords are not always convenient to allow/deny based on header
12423 contents. It is strongly recommended to use ACLs with the "block" keyword
12424 instead, resulting in far more flexible and manageable rules.
12425
12426 - lines are always considered as a whole. It is not possible to reference
12427 a header name only or a value only. This is important because of the way
12428 headers are written (notably the number of spaces after the colon).
12429
12430 - the first line is always considered as a header, which makes it possible to
12431 rewrite or filter HTTP requests URIs or response codes, but in turn makes
12432 it harder to distinguish between headers and request line. The regex prefix
12433 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
12434 ^[^ \t:]*: matches any header name followed by a colon.
12435
12436 - for performances reasons, the number of characters added to a request or to
12437 a response is limited at build time to values between 1 and 4 kB. This
12438 should normally be far more than enough for most usages. If it is too short
12439 on occasional usages, it is possible to gain some space by removing some
12440 useless headers before adding new ones.
12441
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012442 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012443 without the 'i' letter except that they ignore case when matching patterns.
12444
12445 - when a request passes through a frontend then a backend, all req* rules
12446 from the frontend will be evaluated, then all req* rules from the backend
12447 will be evaluated. The reverse path is applied to responses.
12448
12449 - req* statements are applied after "block" statements, so that "block" is
12450 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010012451 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012452
12453
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200124546. Cache
12455---------
12456
12457HAProxy provides a cache, which was designed to perform cache on small objects
12458(favicon, css...). This is a minimalist low-maintenance cache which runs in
12459RAM.
12460
12461The cache is based on a memory which is shared between processes and threads,
12462this memory is split in blocks of 1k.
12463
12464If an object is not used anymore, it can be deleted to store a new object
12465independently of its expiration date. The oldest objects are deleted first
12466when we try to allocate a new one.
12467
12468The cache uses a hash of the host header and the URI as the key.
12469
12470It's possible to view the status of a cache using the Unix socket command
12471"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
12472for more details.
12473
12474When an object is delivered from the cache, the server name in the log is
12475replaced by "<CACHE>".
12476
12477
124786.1. Limitation
12479----------------
12480
12481The cache won't store and won't deliver objects in these cases:
12482
12483- If the response is not a 200
12484- If the response contains a Vary header
12485- If the Content-Length + the headers size is greater than "max-object-size"
12486- If the response is not cacheable
12487
12488- If the request is not a GET
12489- If the HTTP version of the request is smaller than 1.1
12490- If the request contains an Authorization header
12491
12492
124936.2. Setup
12494-----------
12495
12496To setup a cache, you must define a cache section and use it in a proxy with
12497the corresponding http-request and response actions.
12498
12499
125006.2.1. Cache section
12501---------------------
12502
12503cache <name>
12504 Declare a cache section, allocate a shared cache memory named <name>, the
12505 size of cache is mandatory.
12506
12507total-max-size <megabytes>
12508 Define the size in RAM of the cache in megabytes. This size is split in
12509 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
12510
12511max-object-size <bytes>
12512 Define the maximum size of the objects to be cached. Must not be greater than
12513 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
12514 All objects with sizes larger than "max-object-size" will not be cached.
12515
12516max-age <seconds>
12517 Define the maximum expiration duration. The expiration is set has the lowest
12518 value between the s-maxage or max-age (in this order) directive in the
12519 Cache-Control response header and this value. The default value is 60
12520 seconds, which means that you can't cache an object more than 60 seconds by
12521 default.
12522
12523
125246.2.2. Proxy section
12525---------------------
12526
12527http-request cache-use <name> [ { if | unless } <condition> ]
12528 Try to deliver a cached object from the cache <name>. This directive is also
12529 mandatory to store the cache as it calculates the cache hash. If you want to
12530 use a condition for both storage and delivering that's a good idea to put it
12531 after this one.
12532
12533http-response cache-store <name> [ { if | unless } <condition> ]
12534 Store an http-response within the cache. The storage of the response headers
12535 is done at this step, which means you can use others http-response actions
12536 to modify headers before or after the storage of the response. This action
12537 is responsible for the setup of the cache storage filter.
12538
12539
12540Example:
12541
12542 backend bck1
12543 mode http
12544
12545 http-request cache-use foobar
12546 http-response cache-store foobar
12547 server srv1 127.0.0.1:80
12548
12549 cache foobar
12550 total-max-size 4
12551 max-age 240
12552
12553
Willy Tarreau74ca5042013-06-11 23:12:07 +0200125547. Using ACLs and fetching samples
12555----------------------------------
12556
Davor Ocelice9ed2812017-12-25 17:49:28 +010012557HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012558client or server information, from tables, environmental information etc...
12559The action of extracting such data is called fetching a sample. Once retrieved,
12560these samples may be used for various purposes such as a key to a stick-table,
12561but most common usages consist in matching them against predefined constant
12562data called patterns.
12563
12564
125657.1. ACL basics
12566---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012567
12568The use of Access Control Lists (ACL) provides a flexible solution to perform
12569content switching and generally to take decisions based on content extracted
12570from the request, the response or any environmental status. The principle is
12571simple :
12572
Willy Tarreau74ca5042013-06-11 23:12:07 +020012573 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012574 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012575 - apply one or multiple pattern matching methods on this sample
12576 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012577
Willy Tarreau74ca5042013-06-11 23:12:07 +020012578The actions generally consist in blocking a request, selecting a backend, or
12579adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012580
12581In order to define a test, the "acl" keyword is used. The syntax is :
12582
Willy Tarreau74ca5042013-06-11 23:12:07 +020012583 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012584
12585This creates a new ACL <aclname> or completes an existing one with new tests.
12586Those tests apply to the portion of request/response specified in <criterion>
12587and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012588an operator which may be specified before the set of values. Optionally some
12589conversion operators may be applied to the sample, and they will be specified
12590as a comma-delimited list of keywords just after the first keyword. The values
12591are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012592
12593ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12594'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12595which means that "my_acl" and "My_Acl" are two different ACLs.
12596
12597There is no enforced limit to the number of ACLs. The unused ones do not affect
12598performance, they just consume a small amount of memory.
12599
Willy Tarreau74ca5042013-06-11 23:12:07 +020012600The criterion generally is the name of a sample fetch method, or one of its ACL
12601specific declinations. The default test method is implied by the output type of
12602this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012603methods of a same sample fetch method. The sample fetch methods are the only
12604ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012605
12606Sample fetch methods return data which can be of the following types :
12607 - boolean
12608 - integer (signed or unsigned)
12609 - IPv4 or IPv6 address
12610 - string
12611 - data block
12612
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012613Converters transform any of these data into any of these. For example, some
12614converters might convert a string to a lower-case string while other ones
12615would turn a string to an IPv4 address, or apply a netmask to an IP address.
12616The resulting sample is of the type of the last converter applied to the list,
12617which defaults to the type of the sample fetch method.
12618
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012619Each sample or converter returns data of a specific type, specified with its
12620keyword in this documentation. When an ACL is declared using a standard sample
12621fetch method, certain types automatically involved a default matching method
12622which are summarized in the table below :
12623
12624 +---------------------+-----------------+
12625 | Sample or converter | Default |
12626 | output type | matching method |
12627 +---------------------+-----------------+
12628 | boolean | bool |
12629 +---------------------+-----------------+
12630 | integer | int |
12631 +---------------------+-----------------+
12632 | ip | ip |
12633 +---------------------+-----------------+
12634 | string | str |
12635 +---------------------+-----------------+
12636 | binary | none, use "-m" |
12637 +---------------------+-----------------+
12638
12639Note that in order to match a binary samples, it is mandatory to specify a
12640matching method, see below.
12641
Willy Tarreau74ca5042013-06-11 23:12:07 +020012642The ACL engine can match these types against patterns of the following types :
12643 - boolean
12644 - integer or integer range
12645 - IP address / network
12646 - string (exact, substring, suffix, prefix, subdir, domain)
12647 - regular expression
12648 - hex block
12649
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012650The following ACL flags are currently supported :
12651
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012652 -i : ignore case during matching of all subsequent patterns.
12653 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012654 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012655 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012656 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012657 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012658 -- : force end of flags. Useful when a string looks like one of the flags.
12659
Willy Tarreau74ca5042013-06-11 23:12:07 +020012660The "-f" flag is followed by the name of a file from which all lines will be
12661read as individual values. It is even possible to pass multiple "-f" arguments
12662if the patterns are to be loaded from multiple files. Empty lines as well as
12663lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12664will be stripped. If it is absolutely necessary to insert a valid pattern
12665beginning with a sharp, just prefix it with a space so that it is not taken for
12666a comment. Depending on the data type and match method, haproxy may load the
12667lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12668exact string matching. In this case, duplicates will automatically be removed.
12669
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012670The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12671parsed as two column file. The first column contains the patterns used by the
12672ACL, and the second column contain the samples. The sample can be used later by
12673a map. This can be useful in some rare cases where an ACL would just be used to
12674check for the existence of a pattern in a map before a mapping is applied.
12675
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012676The "-u" flag forces the unique id of the ACL. This unique id is used with the
12677socket interface to identify ACL and dynamically change its values. Note that a
12678file is always identified by its name even if an id is set.
12679
Willy Tarreau74ca5042013-06-11 23:12:07 +020012680Also, note that the "-i" flag applies to subsequent entries and not to entries
12681loaded from files preceding it. For instance :
12682
12683 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12684
12685In this example, each line of "exact-ua.lst" will be exactly matched against
12686the "user-agent" header of the request. Then each line of "generic-ua" will be
12687case-insensitively matched. Then the word "test" will be insensitively matched
12688as well.
12689
12690The "-m" flag is used to select a specific pattern matching method on the input
12691sample. All ACL-specific criteria imply a pattern matching method and generally
12692do not need this flag. However, this flag is useful with generic sample fetch
12693methods to describe how they're going to be matched against the patterns. This
12694is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012695obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012696followed by a pattern matching method name, this method is used instead of the
12697default one for the criterion. This makes it possible to match contents in ways
12698that were not initially planned, or with sample fetch methods which return a
12699string. The matching method also affects the way the patterns are parsed.
12700
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012701The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12702By default, if the parser cannot parse ip address it considers that the parsed
12703string is maybe a domain name and try dns resolution. The flag "-n" disable this
12704resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12705server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050012706waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012707flag "-n" disable this behavior. Note also that during the runtime, this
12708function is disabled for the dynamic acl modifications.
12709
Willy Tarreau74ca5042013-06-11 23:12:07 +020012710There are some restrictions however. Not all methods can be used with all
12711sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12712be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012713
12714 - "found" : only check if the requested sample could be found in the stream,
12715 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012716 to pass any pattern to avoid confusion. This matching method is
12717 particularly useful to detect presence of certain contents such
12718 as headers, cookies, etc... even if they are empty and without
12719 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012720
12721 - "bool" : check the value as a boolean. It can only be applied to fetches
12722 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012723 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012724
12725 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012726 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012727
12728 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012729 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012730
Davor Ocelice9ed2812017-12-25 17:49:28 +010012731 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012732 binary sequence. This may be used with binary or string samples.
12733
12734 - "len" : match the sample's length as an integer. This may be used with
12735 binary or string samples.
12736
Willy Tarreau74ca5042013-06-11 23:12:07 +020012737 - "str" : exact match : match the contents against a string. This may be
12738 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012739
Willy Tarreau74ca5042013-06-11 23:12:07 +020012740 - "sub" : substring match : check that the contents contain at least one of
12741 the provided string patterns. This may be used with binary or
12742 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012743
Willy Tarreau74ca5042013-06-11 23:12:07 +020012744 - "reg" : regex match : match the contents against a list of regular
12745 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012746
Willy Tarreau74ca5042013-06-11 23:12:07 +020012747 - "beg" : prefix match : check that the contents begin like the provided
12748 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012749
Willy Tarreau74ca5042013-06-11 23:12:07 +020012750 - "end" : suffix match : check that the contents end like the provided
12751 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012752
Willy Tarreau74ca5042013-06-11 23:12:07 +020012753 - "dir" : subdir match : check that a slash-delimited portion of the
12754 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012755 This may be used with binary or string samples.
12756
Willy Tarreau74ca5042013-06-11 23:12:07 +020012757 - "dom" : domain match : check that a dot-delimited portion of the contents
12758 exactly match one of the provided string patterns. This may be
12759 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012760
12761For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12762request, it is possible to do :
12763
12764 acl jsess_present cook(JSESSIONID) -m found
12765
12766In order to apply a regular expression on the 500 first bytes of data in the
12767buffer, one would use the following acl :
12768
12769 acl script_tag payload(0,500) -m reg -i <script>
12770
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012771On systems where the regex library is much slower when using "-i", it is
12772possible to convert the sample to lowercase before matching, like this :
12773
12774 acl script_tag payload(0,500),lower -m reg <script>
12775
Willy Tarreau74ca5042013-06-11 23:12:07 +020012776All ACL-specific criteria imply a default matching method. Most often, these
12777criteria are composed by concatenating the name of the original sample fetch
12778method and the matching method. For example, "hdr_beg" applies the "beg" match
12779to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12780criteria rely on a sample fetch method, it is always possible instead to use
12781the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012782
Willy Tarreau74ca5042013-06-11 23:12:07 +020012783If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012784the matching method is simply applied to the underlying sample fetch method.
12785For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012786
Willy Tarreau74ca5042013-06-11 23:12:07 +020012787 acl short_form hdr_beg(host) www.
12788 acl alternate1 hdr_beg(host) -m beg www.
12789 acl alternate2 hdr_dom(host) -m beg www.
12790 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012791
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012792
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012793The table below summarizes the compatibility matrix between sample or converter
12794types and the pattern types to fetch against. It indicates for each compatible
12795combination the name of the matching method to be used, surrounded with angle
12796brackets ">" and "<" when the method is the default one and will work by
12797default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012798
Willy Tarreau74ca5042013-06-11 23:12:07 +020012799 +-------------------------------------------------+
12800 | Input sample type |
12801 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012802 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012803 +----------------------+---------+---------+---------+---------+---------+
12804 | none (presence only) | found | found | found | found | found |
12805 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012806 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012807 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012808 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012809 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012810 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012811 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012812 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012813 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012814 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012815 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012816 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012817 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012818 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012819 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012820 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012821 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012822 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012823 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012824 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012825 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012826 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012827 +----------------------+---------+---------+---------+---------+---------+
12828 | hex block | | | | bin | bin |
12829 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012830
12831
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128327.1.1. Matching booleans
12833------------------------
12834
12835In order to match a boolean, no value is needed and all values are ignored.
12836Boolean matching is used by default for all fetch methods of type "boolean".
12837When boolean matching is used, the fetched value is returned as-is, which means
12838that a boolean "true" will always match and a boolean "false" will never match.
12839
12840Boolean matching may also be enforced using "-m bool" on fetch methods which
12841return an integer value. Then, integer value 0 is converted to the boolean
12842"false" and all other values are converted to "true".
12843
Willy Tarreau6a06a402007-07-15 20:15:28 +020012844
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128457.1.2. Matching integers
12846------------------------
12847
12848Integer matching applies by default to integer fetch methods. It can also be
12849enforced on boolean fetches using "-m int". In this case, "false" is converted
12850to the integer 0, and "true" is converted to the integer 1.
12851
12852Integer matching also supports integer ranges and operators. Note that integer
12853matching only applies to positive values. A range is a value expressed with a
12854lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012855
12856For instance, "1024:65535" is a valid range to represent a range of
12857unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12858representation of privileged ports, and ":1023" would also work.
12859
Willy Tarreau62644772008-07-16 18:36:06 +020012860As a special case, some ACL functions support decimal numbers which are in fact
12861two integers separated by a dot. This is used with some version checks for
12862instance. All integer properties apply to those decimal numbers, including
12863ranges and operators.
12864
Willy Tarreau6a06a402007-07-15 20:15:28 +020012865For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012866operators with ranges does not make much sense and is strongly discouraged.
12867Similarly, it does not make much sense to perform order comparisons with a set
12868of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012869
Willy Tarreau0ba27502007-12-24 16:55:16 +010012870Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012871
12872 eq : true if the tested value equals at least one value
12873 ge : true if the tested value is greater than or equal to at least one value
12874 gt : true if the tested value is greater than at least one value
12875 le : true if the tested value is less than or equal to at least one value
12876 lt : true if the tested value is less than at least one value
12877
Willy Tarreau0ba27502007-12-24 16:55:16 +010012878For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012879
12880 acl negative-length hdr_val(content-length) lt 0
12881
Willy Tarreau62644772008-07-16 18:36:06 +020012882This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12883
12884 acl sslv3 req_ssl_ver 3:3.1
12885
Willy Tarreau6a06a402007-07-15 20:15:28 +020012886
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128877.1.3. Matching strings
12888-----------------------
12889
12890String matching applies to string or binary fetch methods, and exists in 6
12891different forms :
12892
12893 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012894 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012895
12896 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012897 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012898
12899 - prefix match (-m beg) : the patterns are compared with the beginning of
12900 the extracted string, and the ACL matches if any of them matches.
12901
12902 - suffix match (-m end) : the patterns are compared with the end of the
12903 extracted string, and the ACL matches if any of them matches.
12904
Baptiste Assmann33db6002016-03-06 23:32:10 +010012905 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012906 string, delimited with slashes ("/"), and the ACL matches if any of them
12907 matches.
12908
12909 - domain match (-m dom) : the patterns are looked up inside the extracted
12910 string, delimited with dots ("."), and the ACL matches if any of them
12911 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012912
12913String matching applies to verbatim strings as they are passed, with the
12914exception of the backslash ("\") which makes it possible to escape some
12915characters such as the space. If the "-i" flag is passed before the first
12916string, then the matching will be performed ignoring the case. In order
12917to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012918before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012919
12920
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129217.1.4. Matching regular expressions (regexes)
12922---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012923
12924Just like with string matching, regex matching applies to verbatim strings as
12925they are passed, with the exception of the backslash ("\") which makes it
12926possible to escape some characters such as the space. If the "-i" flag is
12927passed before the first regex, then the matching will be performed ignoring
12928the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012929the "--" flag before the first string. Same principle applies of course to
12930match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012931
12932
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129337.1.5. Matching arbitrary data blocks
12934-------------------------------------
12935
12936It is possible to match some extracted samples against a binary block which may
12937not safely be represented as a string. For this, the patterns must be passed as
12938a series of hexadecimal digits in an even number, when the match method is set
12939to binary. Each sequence of two digits will represent a byte. The hexadecimal
12940digits may be used upper or lower case.
12941
12942Example :
12943 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12944 acl hello payload(0,6) -m bin 48656c6c6f0a
12945
12946
129477.1.6. Matching IPv4 and IPv6 addresses
12948---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012949
12950IPv4 addresses values can be specified either as plain addresses or with a
12951netmask appended, in which case the IPv4 address matches whenever it is
12952within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012953host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012954difficult to read and debug configurations. If hostnames are used, you should
12955at least ensure that they are present in /etc/hosts so that the configuration
12956does not depend on any random DNS match at the moment the configuration is
12957parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012958
Daniel Schnellereba56342016-04-13 00:26:52 +020012959The dotted IPv4 address notation is supported in both regular as well as the
12960abbreviated form with all-0-octets omitted:
12961
12962 +------------------+------------------+------------------+
12963 | Example 1 | Example 2 | Example 3 |
12964 +------------------+------------------+------------------+
12965 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12966 | 192.168.1 | 10.12 | 127.1 |
12967 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
12968 | 192.168.1/22 | 10.12/8 | 127.1/8 |
12969 +------------------+------------------+------------------+
12970
12971Notice that this is different from RFC 4632 CIDR address notation in which
12972192.168.42/24 would be equivalent to 192.168.42.0/24.
12973
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012974IPv6 may be entered in their usual form, with or without a netmask appended.
12975Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
12976trouble with randomly resolved IP addresses, host names are never allowed in
12977IPv6 patterns.
12978
12979HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
12980following situations :
12981 - tested address is IPv4, pattern address is IPv4, the match applies
12982 in IPv4 using the supplied mask if any.
12983 - tested address is IPv6, pattern address is IPv6, the match applies
12984 in IPv6 using the supplied mask if any.
12985 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
12986 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
12987 ::IPV4 or ::ffff:IPV4, otherwise it fails.
12988 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
12989 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
12990 applied in IPv6 using the supplied IPv6 mask.
12991
Willy Tarreau74ca5042013-06-11 23:12:07 +020012992
129937.2. Using ACLs to form conditions
12994----------------------------------
12995
12996Some actions are only performed upon a valid condition. A condition is a
12997combination of ACLs with operators. 3 operators are supported :
12998
12999 - AND (implicit)
13000 - OR (explicit with the "or" keyword or the "||" operator)
13001 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020013002
Willy Tarreau74ca5042013-06-11 23:12:07 +020013003A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013004
Willy Tarreau74ca5042013-06-11 23:12:07 +020013005 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020013006
Willy Tarreau74ca5042013-06-11 23:12:07 +020013007Such conditions are generally used after an "if" or "unless" statement,
13008indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020013009
Willy Tarreau74ca5042013-06-11 23:12:07 +020013010For instance, to block HTTP requests to the "*" URL with methods other than
13011"OPTIONS", as well as POST requests without content-length, and GET or HEAD
13012requests with a content-length greater than 0, and finally every request which
13013is not either GET/HEAD/POST/OPTIONS !
13014
13015 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013016 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
13017 http-request deny if METH_GET HTTP_CONTENT
13018 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020013019
13020To select a different backend for requests to static contents on the "www" site
13021and to every request on the "img", "video", "download" and "ftp" hosts :
13022
13023 acl url_static path_beg /static /images /img /css
13024 acl url_static path_end .gif .png .jpg .css .js
13025 acl host_www hdr_beg(host) -i www
13026 acl host_static hdr_beg(host) -i img. video. download. ftp.
13027
Davor Ocelice9ed2812017-12-25 17:49:28 +010013028 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020013029 # of host "www". Use backend "www" for the rest.
13030 use_backend static if host_static or host_www url_static
13031 use_backend www if host_www
13032
13033It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13034expressions that are built on the fly without needing to be declared. They must
13035be enclosed between braces, with a space before and after each brace (because
13036the braces must be seen as independent words). Example :
13037
13038 The following rule :
13039
13040 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013041 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013042
13043 Can also be written that way :
13044
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013045 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013046
13047It is generally not recommended to use this construct because it's a lot easier
13048to leave errors in the configuration when written that way. However, for very
13049simple rules matching only one source IP address for instance, it can make more
13050sense to use them than to declare ACLs with random names. Another example of
13051good use is the following :
13052
13053 With named ACLs :
13054
13055 acl site_dead nbsrv(dynamic) lt 2
13056 acl site_dead nbsrv(static) lt 2
13057 monitor fail if site_dead
13058
13059 With anonymous ACLs :
13060
13061 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13062
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013063See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13064keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013065
13066
130677.3. Fetching samples
13068---------------------
13069
13070Historically, sample fetch methods were only used to retrieve data to match
13071against patterns using ACLs. With the arrival of stick-tables, a new class of
13072sample fetch methods was created, most often sharing the same syntax as their
13073ACL counterpart. These sample fetch methods are also known as "fetches". As
13074of now, ACLs and fetches have converged. All ACL fetch methods have been made
13075available as fetch methods, and ACLs may use any sample fetch method as well.
13076
13077This section details all available sample fetch methods and their output type.
13078Some sample fetch methods have deprecated aliases that are used to maintain
13079compatibility with existing configurations. They are then explicitly marked as
13080deprecated and should not be used in new setups.
13081
13082The ACL derivatives are also indicated when available, with their respective
13083matching methods. These ones all have a well defined default pattern matching
13084method, so it is never necessary (though allowed) to pass the "-m" option to
13085indicate how the sample will be matched using ACLs.
13086
13087As indicated in the sample type versus matching compatibility matrix above,
13088when using a generic sample fetch method in an ACL, the "-m" option is
13089mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13090the same keyword exists as an ACL keyword and as a standard fetch method, the
13091ACL engine will automatically pick the ACL-only one by default.
13092
13093Some of these keywords support one or multiple mandatory arguments, and one or
13094multiple optional arguments. These arguments are strongly typed and are checked
13095when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013096incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13097are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013098is optional, it will be indicated below between square brackets ('[ ]'). When
13099all arguments are optional, the parenthesis may be omitted.
13100
13101Thus, the syntax of a standard sample fetch method is one of the following :
13102 - name
13103 - name(arg1)
13104 - name(arg1,arg2)
13105
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013106
131077.3.1. Converters
13108-----------------
13109
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013110Sample fetch methods may be combined with transformations to be applied on top
13111of the fetched sample (also called "converters"). These combinations form what
13112is called "sample expressions" and the result is a "sample". Initially this
13113was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013114has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013115unique-id-format, add-header, ...).
13116
13117These transformations are enumerated as a series of specific keywords after the
13118sample fetch method. These keywords may equally be appended immediately after
13119the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013120support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013121
Willy Tarreau97707872015-01-27 15:12:13 +010013122A certain category of converters are bitwise and arithmetic operators which
13123support performing basic operations on integers. Some bitwise operations are
13124supported (and, or, xor, cpl) and some arithmetic operations are supported
13125(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13126bool) which make it possible to report a match without having to write an ACL.
13127
Willy Tarreau74ca5042013-06-11 23:12:07 +020013128The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013129
Ben Shillitof25e8e52016-12-02 14:25:37 +00001313051d.single(<prop>[,<prop>*])
13131 Returns values for the properties requested as a string, where values are
13132 separated by the delimiter specified with "51degrees-property-separator".
13133 The device is identified using the User-Agent header passed to the
13134 converter. The function can be passed up to five property names, and if a
13135 property name can't be found, the value "NoData" is returned.
13136
13137 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013138 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13139 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013140 # User-Agent passed to the converter.
13141 frontend http-in
13142 bind *:8081
13143 default_backend servers
13144 http-request set-header X-51D-DeviceTypeMobileTablet \
13145 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13146
Willy Tarreau97707872015-01-27 15:12:13 +010013147add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013148 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013149 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013150 name. The name of the variable starts with an indication about its scope. The
13151 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013152 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013153 "sess" : the variable is shared with the whole session
13154 "txn" : the variable is shared with the transaction (request and response)
13155 "req" : the variable is shared only during request processing
13156 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013157 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013158 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013159
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010013160aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
13161 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
13162 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
13163 need to be base64 encoded and the returned result is in raw byte format.
13164 If the <aead_tag> validation fails, the converter doesn't return any data.
13165 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
13166 converter requires at least OpenSSL 1.0.1.
13167
13168 Example:
13169 http-response set-header X-Decrypted-Text %[var(txn.enc),\
13170 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
13171
Willy Tarreau97707872015-01-27 15:12:13 +010013172and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013173 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013174 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013175 numeric value or a variable name. The name of the variable starts with an
13176 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013177 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013178 "sess" : the variable is shared with the whole session
13179 "txn" : the variable is shared with the transaction (request and response)
13180 "req" : the variable is shared only during request processing
13181 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013182 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013183 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013184
Holger Just1bfc24b2017-05-06 00:56:53 +020013185b64dec
13186 Converts (decodes) a base64 encoded input string to its binary
13187 representation. It performs the inverse operation of base64().
13188
Emeric Brun53d1a982014-04-30 18:21:37 +020013189base64
13190 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013191 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013192 an SSL ID can be copied in a header).
13193
Willy Tarreau97707872015-01-27 15:12:13 +010013194bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013195 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013196 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013197 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013198 presence of a flag).
13199
Emeric Brun54c4ac82014-11-03 15:32:43 +010013200bytes(<offset>[,<length>])
13201 Extracts some bytes from an input binary sample. The result is a binary
13202 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013203 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013204
Willy Tarreau280f42b2018-02-19 15:34:12 +010013205concat([<start>],[<var>],[<end>])
13206 Concatenates up to 3 fields after the current sample which is then turned to
13207 a string. The first one, <start>, is a constant string, that will be appended
13208 immediately after the existing sample. It may be omitted if not used. The
13209 second one, <var>, is a variable name. The variable will be looked up, its
13210 contents converted to a string, and it will be appended immediately after the
13211 <first> part. If the variable is not found, nothing is appended. It may be
13212 omitted as well. The third field, <end> is a constant string that will be
13213 appended after the variable. It may also be omitted. Together, these elements
13214 allow to concatenate variables with delimiters to an existing set of
13215 variables. This can be used to build new variables made of a succession of
John Roeslerfb2fce12019-07-10 15:45:51 -050013216 other variables, such as colon-delimited values. Note that due to the config
Willy Tarreau280f42b2018-02-19 15:34:12 +010013217 parser, it is not possible to use a comma nor a closing parenthesis as
John Roeslerfb2fce12019-07-10 15:45:51 -050013218 delimiters.
Willy Tarreau280f42b2018-02-19 15:34:12 +010013219
13220 Example:
13221 tcp-request session set-var(sess.src) src
13222 tcp-request session set-var(sess.dn) ssl_c_s_dn
13223 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13224 http-request set-header x-hap-sig %[var(txn.sig)]
13225
Willy Tarreau97707872015-01-27 15:12:13 +010013226cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013227 Takes the input value of type signed integer, applies a ones-complement
13228 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013229
Willy Tarreau80599772015-01-20 19:35:24 +010013230crc32([<avalanche>])
13231 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13232 hash function. Optionally, it is possible to apply a full avalanche hash
13233 function to the output if the optional <avalanche> argument equals 1. This
13234 converter uses the same functions as used by the various hash-based load
13235 balancing algorithms, so it will provide exactly the same results. It is
13236 provided for compatibility with other software which want a CRC32 to be
13237 computed on some input keys, so it follows the most common implementation as
13238 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13239 but may provide a better or at least less predictable distribution. It must
13240 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013241 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13242
13243crc32c([<avalanche>])
13244 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13245 hash function. Optionally, it is possible to apply a full avalanche hash
13246 function to the output if the optional <avalanche> argument equals 1. This
13247 converter uses the same functions as described in RFC4960, Appendix B [8].
13248 It is provided for compatibility with other software which want a CRC32C to be
13249 computed on some input keys. It is slower than the other algorithms and it must
13250 not be used for security purposes as a 32-bit hash is trivial to break. See
13251 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013252
David Carlier29b3ca32015-09-25 14:09:21 +010013253da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013254 Asks the DeviceAtlas converter to identify the User Agent string passed on
13255 input, and to emit a string made of the concatenation of the properties
13256 enumerated in argument, delimited by the separator defined by the global
13257 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013258 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013259 configuration language.
13260
13261 Example:
13262 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013263 bind *:8881
13264 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013265 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 +020013266
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013267debug
13268 This converter is used as debug tool. It dumps on screen the content and the
13269 type of the input sample. The sample is returned as is on its output. This
13270 converter only exists when haproxy was built with debugging enabled.
13271
Willy Tarreau97707872015-01-27 15:12:13 +010013272div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013273 Divides the input value of type signed integer by <value>, and returns the
13274 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013275 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013276 variable name. The name of the variable starts with an indication about its
13277 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013278 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013279 "sess" : the variable is shared with the whole session
13280 "txn" : the variable is shared with the transaction (request and response)
13281 "req" : the variable is shared only during request processing
13282 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013283 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013284 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013285
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013286djb2([<avalanche>])
13287 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13288 hash function. Optionally, it is possible to apply a full avalanche hash
13289 function to the output if the optional <avalanche> argument equals 1. This
13290 converter uses the same functions as used by the various hash-based load
13291 balancing algorithms, so it will provide exactly the same results. It is
13292 mostly intended for debugging, but can be used as a stick-table entry to
13293 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013294 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13295 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013296
Willy Tarreau97707872015-01-27 15:12:13 +010013297even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013298 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013299 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13300
Marcin Deranek9631a282018-04-16 14:30:46 +020013301field(<index>,<delimiters>[,<count>])
13302 Extracts the substring at the given index counting from the beginning
13303 (positive index) or from the end (negative index) considering given delimiters
13304 from an input string. Indexes start at 1 or -1 and delimiters are a string
13305 formatted list of chars. Optionally you can specify <count> of fields to
13306 extract (default: 1). Value of 0 indicates extraction of all remaining
13307 fields.
13308
13309 Example :
13310 str(f1_f2_f3__f5),field(5,_) # f5
13311 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13312 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13313 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13314 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013315
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013316hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013317 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013318 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013319 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 +020013320 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013321
Dragan Dosen3f957b22017-10-24 09:27:34 +020013322hex2i
13323 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050013324 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020013325
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013326http_date([<offset>])
13327 Converts an integer supposed to contain a date since epoch to a string
13328 representing this date in a format suitable for use in HTTP header fields. If
13329 an offset value is specified, then it is a number of seconds that is added to
13330 the date before the conversion is operated. This is particularly useful to
13331 emit Date header fields, Expires values in responses when combined with a
13332 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013333
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013334in_table(<table>)
13335 Uses the string representation of the input sample to perform a look up in
13336 the specified table. If the key is not found in the table, a boolean false
13337 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013338 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013339 or not a source IP address or an Authorization header was already seen).
13340
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013341ipmask(<mask4>, [<mask6>])
13342 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013343 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013344 table entries and as such use the same server. The mask4 can be passed in
13345 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13346 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13347 If no mask6 is given IPv6 addresses will fail to convert for backwards
13348 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013349
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013350json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013351 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013352 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013353 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013354 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13355 of errors:
13356 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13357 bytes, ...)
13358 - invalid range (the decoded value is within a UTF-8 prohibited range),
13359 - code overlong (the value is encoded with more bytes than necessary).
13360
13361 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13362 character is greater than 0xffff because the JSON string escape specification
13363 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13364 in 4 variants designated by a combination of two suffix letters : "p" for
13365 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13366 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013367 - "ascii" : never fails;
13368 - "utf8" : fails on any detected errors;
13369 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013370 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013371 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013372 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13373 characters corresponding to the other errors.
13374
13375 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013376 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013377
13378 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013379 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013380 capture request header user-agent len 150
13381 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013382
13383 Input request from client 127.0.0.1:
13384 GET / HTTP/1.0
13385 User-Agent: Very "Ugly" UA 1/2
13386
13387 Output log:
13388 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13389
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013390language(<value>[,<default>])
13391 Returns the value with the highest q-factor from a list as extracted from the
13392 "accept-language" header using "req.fhdr". Values with no q-factor have a
13393 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13394 belong to the list of semi-colon delimited <values> will be considered. The
13395 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13396 given list and a default value is provided, it is returned. Note that language
13397 names may have a variant after a dash ('-'). If this variant is present in the
13398 list, it will be matched, but if it is not, only the base language is checked.
13399 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013400 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013401 ordering of the values in the request counts, as the first value among
13402 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013403
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013404 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013405
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013406 # this configuration switches to the backend matching a
13407 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013408
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013409 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13410 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13411 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13412 use_backend spanish if es
13413 use_backend french if fr
13414 use_backend english if en
13415 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013416
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013417length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013418 Get the length of the string. This can only be placed after a string
13419 sample fetch function or after a transformation keyword returning a string
13420 type. The result is of type integer.
13421
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013422lower
13423 Convert a string sample to lower case. This can only be placed after a string
13424 sample fetch function or after a transformation keyword returning a string
13425 type. The result is of type string.
13426
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013427ltime(<format>[,<offset>])
13428 Converts an integer supposed to contain a date since epoch to a string
13429 representing this date in local time using a format defined by the <format>
13430 string using strftime(3). The purpose is to allow any date format to be used
13431 in logs. An optional <offset> in seconds may be applied to the input date
13432 (positive or negative). See the strftime() man page for the format supported
13433 by your operating system. See also the utime converter.
13434
13435 Example :
13436
13437 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013438 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013439 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13440
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013441map(<map_file>[,<default_value>])
13442map_<match_type>(<map_file>[,<default_value>])
13443map_<match_type>_<output_type>(<map_file>[,<default_value>])
13444 Search the input value from <map_file> using the <match_type> matching method,
13445 and return the associated value converted to the type <output_type>. If the
13446 input value cannot be found in the <map_file>, the converter returns the
13447 <default_value>. If the <default_value> is not set, the converter fails and
13448 acts as if no input value could be fetched. If the <match_type> is not set, it
13449 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13450 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13451 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013452
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013453 It is important to avoid overlapping between the keys : IP addresses and
13454 strings are stored in trees, so the first of the finest match will be used.
13455 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013456
Tim Düsterhus4896c442016-11-29 02:15:19 +010013457 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013458 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013459
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013460 input type | match method | output type str | output type int | output type ip
13461 -----------+--------------+-----------------+-----------------+---------------
13462 str | str | map_str | map_str_int | map_str_ip
13463 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013464 str | beg | map_beg | map_beg_int | map_end_ip
13465 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013466 str | sub | map_sub | map_sub_int | map_sub_ip
13467 -----------+--------------+-----------------+-----------------+---------------
13468 str | dir | map_dir | map_dir_int | map_dir_ip
13469 -----------+--------------+-----------------+-----------------+---------------
13470 str | dom | map_dom | map_dom_int | map_dom_ip
13471 -----------+--------------+-----------------+-----------------+---------------
13472 str | end | map_end | map_end_int | map_end_ip
13473 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013474 str | reg | map_reg | map_reg_int | map_reg_ip
13475 -----------+--------------+-----------------+-----------------+---------------
13476 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013477 -----------+--------------+-----------------+-----------------+---------------
13478 int | int | map_int | map_int_int | map_int_ip
13479 -----------+--------------+-----------------+-----------------+---------------
13480 ip | ip | map_ip | map_ip_int | map_ip_ip
13481 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013482
Thierry Fournier8feaa662016-02-10 22:55:20 +010013483 The special map called "map_regm" expect matching zone in the regular
13484 expression and modify the output replacing back reference (like "\1") by
13485 the corresponding match text.
13486
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013487 The file contains one key + value per line. Lines which start with '#' are
13488 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13489 is then the first "word" (series of non-space/tabs characters), and the value
13490 is what follows this series of space/tab till the end of the line excluding
13491 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013492
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013493 Example :
13494
13495 # this is a comment and is ignored
13496 2.22.246.0/23 United Kingdom \n
13497 <-><-----------><--><------------><---->
13498 | | | | `- trailing spaces ignored
13499 | | | `---------- value
13500 | | `-------------------- middle spaces ignored
13501 | `---------------------------- key
13502 `------------------------------------ leading spaces ignored
13503
Willy Tarreau97707872015-01-27 15:12:13 +010013504mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013505 Divides the input value of type signed integer by <value>, and returns the
13506 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013507 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013508 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013509 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013510 "sess" : the variable is shared with the whole session
13511 "txn" : the variable is shared with the transaction (request and response)
13512 "req" : the variable is shared only during request processing
13513 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013514 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013515 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013516
13517mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013518 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013519 the product as an signed integer. In case of overflow, the largest possible
13520 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013521 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013522 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013523 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013524 "sess" : the variable is shared with the whole session
13525 "txn" : the variable is shared with the transaction (request and response)
13526 "req" : the variable is shared only during request processing
13527 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013528 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013529 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013530
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013531nbsrv
13532 Takes an input value of type string, interprets it as a backend name and
13533 returns the number of usable servers in that backend. Can be used in places
13534 where we want to look up a backend from a dynamic name, like a result of a
13535 map lookup.
13536
Willy Tarreau97707872015-01-27 15:12:13 +010013537neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013538 Takes the input value of type signed integer, computes the opposite value,
13539 and returns the remainder as an signed integer. 0 is identity. This operator
13540 is provided for reversed subtracts : in order to subtract the input from a
13541 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013542
13543not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013544 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013545 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013546 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013547 absence of a flag).
13548
13549odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013550 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013551 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13552
13553or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013554 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013555 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013556 numeric value or a variable name. The name of the variable starts with an
13557 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013558 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013559 "sess" : the variable is shared with the whole session
13560 "txn" : the variable is shared with the transaction (request and response)
13561 "req" : the variable is shared only during request processing
13562 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013563 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013564 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013565
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013566protobuf(<field_number>,[<field_type>])
13567 This extracts the protocol buffers message field in raw mode of an input binary
13568 sample representation of a protocol buffer message with <field_number> as field
13569 number (dotted notation) if <field_type> is not present, or as an integer sample
13570 if this field is present (see also "ungrpc" below).
13571 The list of the authorized types is the following one: "int32", "int64", "uint32",
13572 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13573 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13574 "float" for the wire type 5. Note that "string" is considered as a length-delimited
13575 type, so it does not require any <field_type> argument to be extracted.
13576 More information may be found here about the protocol buffers message field types:
13577 https://developers.google.com/protocol-buffers/docs/encoding
13578
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013579regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013580 Applies a regex-based substitution to the input string. It does the same
13581 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13582 default it will replace in the input string the first occurrence of the
13583 largest part matching the regular expression <regex> with the substitution
13584 string <subst>. It is possible to replace all occurrences instead by adding
13585 the flag "g" in the third argument <flags>. It is also possible to make the
13586 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13587 string, it is made up from the concatenation of all desired flags. Thus if
13588 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13589 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013590 configuration parser, some characters such as closing parenthesis, closing
13591 square brackets or comma are not possible to use in the arguments. The first
13592 use of this converter is to replace certain characters or sequence of
13593 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013594
13595 Example :
13596
13597 # de-duplicate "/" in header "x-path".
13598 # input: x-path: /////a///b/c/xzxyz/
13599 # output: x-path: /a/b/c/xzxyz/
13600 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13601
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013602capture-req(<id>)
13603 Capture the string entry in the request slot <id> and returns the entry as
13604 is. If the slot doesn't exist, the capture fails silently.
13605
13606 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013607 "http-response capture", "capture.req.hdr" and
13608 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013609
13610capture-res(<id>)
13611 Capture the string entry in the response slot <id> and returns the entry as
13612 is. If the slot doesn't exist, the capture fails silently.
13613
13614 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013615 "http-response capture", "capture.req.hdr" and
13616 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013617
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013618sdbm([<avalanche>])
13619 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13620 hash function. Optionally, it is possible to apply a full avalanche hash
13621 function to the output if the optional <avalanche> argument equals 1. This
13622 converter uses the same functions as used by the various hash-based load
13623 balancing algorithms, so it will provide exactly the same results. It is
13624 mostly intended for debugging, but can be used as a stick-table entry to
13625 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013626 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13627 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013628
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013629set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013630 Sets a variable with the input content and returns the content on the output
13631 as-is. The variable keeps the value and the associated input type. The name of
13632 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013633 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013634 "sess" : the variable is shared with the whole session
13635 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013636 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013637 "req" : the variable is shared only during request processing,
13638 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013639 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013640 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013641
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013642sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020013643 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013644 sample with length of 20 bytes.
13645
Tim Duesterhusd4376302019-06-17 12:41:44 +020013646sha2([<bits>])
13647 Converts a binary input sample to a digest in the SHA-2 family. The result
13648 is a binary sample with length of <bits>/8 bytes.
13649
13650 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
13651 SHA-<bits>. The default value is 256.
13652
13653 Please note that this converter is only available when haproxy has been
13654 compiled with USE_OPENSSL.
13655
Nenad Merdanovic177adc92019-08-27 01:58:13 +020013656srv_queue
13657 Takes an input value of type string, either a server name or <backend>/<server>
13658 format and returns the number of queued sessions on that server. Can be used
13659 in places where we want to look up queued sessions from a dynamic name, like a
13660 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
13661 persistence or direct a request elsewhere.
13662
Tim Duesterhusca097c12018-04-27 21:18:45 +020013663strcmp(<var>)
13664 Compares the contents of <var> with the input value of type string. Returns
13665 the result as a signed integer compatible with strcmp(3): 0 if both strings
13666 are identical. A value less than 0 if the left string is lexicographically
13667 smaller than the right string or if the left string is shorter. A value greater
13668 than 0 otherwise (right string greater than left string or the right string is
13669 shorter).
13670
13671 Example :
13672
13673 http-request set-var(txn.host) hdr(host)
13674 # Check whether the client is attempting domain fronting.
13675 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13676
13677
Willy Tarreau97707872015-01-27 15:12:13 +010013678sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013679 Subtracts <value> from the input value of type signed integer, and returns
13680 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013681 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013682 or a variable name. The name of the variable starts with an indication about
13683 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013684 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013685 "sess" : the variable is shared with the whole session
13686 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013687 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013688 "req" : the variable is shared only during request processing,
13689 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013690 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013691 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013692
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013693table_bytes_in_rate(<table>)
13694 Uses the string representation of the input sample to perform a look up in
13695 the specified table. If the key is not found in the table, integer value zero
13696 is returned. Otherwise the converter returns the average client-to-server
13697 bytes rate associated with the input sample in the designated table, measured
13698 in amount of bytes over the period configured in the table. See also the
13699 sc_bytes_in_rate sample fetch keyword.
13700
13701
13702table_bytes_out_rate(<table>)
13703 Uses the string representation of the input sample to perform a look up in
13704 the specified table. If the key is not found in the table, integer value zero
13705 is returned. Otherwise the converter returns the average server-to-client
13706 bytes rate associated with the input sample in the designated table, measured
13707 in amount of bytes over the period configured in the table. See also the
13708 sc_bytes_out_rate sample fetch keyword.
13709
13710table_conn_cnt(<table>)
13711 Uses the string representation of the input sample to perform a look up in
13712 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013713 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013714 connections associated with the input sample in the designated table. See
13715 also the sc_conn_cnt sample fetch keyword.
13716
13717table_conn_cur(<table>)
13718 Uses the string representation of the input sample to perform a look up in
13719 the specified table. If the key is not found in the table, integer value zero
13720 is returned. Otherwise the converter returns the current amount of concurrent
13721 tracked connections associated with the input sample in the designated table.
13722 See also the sc_conn_cur sample fetch keyword.
13723
13724table_conn_rate(<table>)
13725 Uses the string representation of the input sample to perform a look up in
13726 the specified table. If the key is not found in the table, integer value zero
13727 is returned. Otherwise the converter returns the average incoming connection
13728 rate associated with the input sample in the designated table. See also the
13729 sc_conn_rate sample fetch keyword.
13730
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013731table_gpt0(<table>)
13732 Uses the string representation of the input sample to perform a look up in
13733 the specified table. If the key is not found in the table, boolean value zero
13734 is returned. Otherwise the converter returns the current value of the first
13735 general purpose tag associated with the input sample in the designated table.
13736 See also the sc_get_gpt0 sample fetch keyword.
13737
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013738table_gpc0(<table>)
13739 Uses the string representation of the input sample to perform a look up in
13740 the specified table. If the key is not found in the table, integer value zero
13741 is returned. Otherwise the converter returns the current value of the first
13742 general purpose counter associated with the input sample in the designated
13743 table. See also the sc_get_gpc0 sample fetch keyword.
13744
13745table_gpc0_rate(<table>)
13746 Uses the string representation of the input sample to perform a look up in
13747 the specified table. If the key is not found in the table, integer value zero
13748 is returned. Otherwise the converter returns the frequency which the gpc0
13749 counter was incremented over the configured period in the table, associated
13750 with the input sample in the designated table. See also the sc_get_gpc0_rate
13751 sample fetch keyword.
13752
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013753table_gpc1(<table>)
13754 Uses the string representation of the input sample to perform a look up in
13755 the specified table. If the key is not found in the table, integer value zero
13756 is returned. Otherwise the converter returns the current value of the second
13757 general purpose counter associated with the input sample in the designated
13758 table. See also the sc_get_gpc1 sample fetch keyword.
13759
13760table_gpc1_rate(<table>)
13761 Uses the string representation of the input sample to perform a look up in
13762 the specified table. If the key is not found in the table, integer value zero
13763 is returned. Otherwise the converter returns the frequency which the gpc1
13764 counter was incremented over the configured period in the table, associated
13765 with the input sample in the designated table. See also the sc_get_gpc1_rate
13766 sample fetch keyword.
13767
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013768table_http_err_cnt(<table>)
13769 Uses the string representation of the input sample to perform a look up in
13770 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013771 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013772 errors associated with the input sample in the designated table. See also the
13773 sc_http_err_cnt sample fetch keyword.
13774
13775table_http_err_rate(<table>)
13776 Uses the string representation of the input sample to perform a look up in
13777 the specified table. If the key is not found in the table, integer value zero
13778 is returned. Otherwise the average rate of HTTP errors associated with the
13779 input sample in the designated table, measured in amount of errors over the
13780 period configured in the table. See also the sc_http_err_rate sample fetch
13781 keyword.
13782
13783table_http_req_cnt(<table>)
13784 Uses the string representation of the input sample to perform a look up in
13785 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013786 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013787 requests associated with the input sample in the designated table. See also
13788 the sc_http_req_cnt sample fetch keyword.
13789
13790table_http_req_rate(<table>)
13791 Uses the string representation of the input sample to perform a look up in
13792 the specified table. If the key is not found in the table, integer value zero
13793 is returned. Otherwise the average rate of HTTP requests associated with the
13794 input sample in the designated table, measured in amount of requests over the
13795 period configured in the table. See also the sc_http_req_rate sample fetch
13796 keyword.
13797
13798table_kbytes_in(<table>)
13799 Uses the string representation of the input sample to perform a look up in
13800 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013801 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013802 to-server data associated with the input sample in the designated table,
13803 measured in kilobytes. The test is currently performed on 32-bit integers,
13804 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13805 keyword.
13806
13807table_kbytes_out(<table>)
13808 Uses the string representation of the input sample to perform a look up in
13809 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013810 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013811 to-client data associated with the input sample in the designated table,
13812 measured in kilobytes. The test is currently performed on 32-bit integers,
13813 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13814 keyword.
13815
13816table_server_id(<table>)
13817 Uses the string representation of the input sample to perform a look up in
13818 the specified table. If the key is not found in the table, integer value zero
13819 is returned. Otherwise the converter returns the server ID associated with
13820 the input sample in the designated table. A server ID is associated to a
13821 sample by a "stick" rule when a connection to a server succeeds. A server ID
13822 zero means that no server is associated with this key.
13823
13824table_sess_cnt(<table>)
13825 Uses the string representation of the input sample to perform a look up in
13826 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013827 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013828 sessions associated with the input sample in the designated table. Note that
13829 a session here refers to an incoming connection being accepted by the
13830 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13831 keyword.
13832
13833table_sess_rate(<table>)
13834 Uses the string representation of the input sample to perform a look up in
13835 the specified table. If the key is not found in the table, integer value zero
13836 is returned. Otherwise the converter returns the average incoming session
13837 rate associated with the input sample in the designated table. Note that a
13838 session here refers to an incoming connection being accepted by the
13839 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13840 keyword.
13841
13842table_trackers(<table>)
13843 Uses the string representation of the input sample to perform a look up in
13844 the specified table. If the key is not found in the table, integer value zero
13845 is returned. Otherwise the converter returns the current amount of concurrent
13846 connections tracking the same key as the input sample in the designated
13847 table. It differs from table_conn_cur in that it does not rely on any stored
13848 information but on the table's reference count (the "use" value which is
13849 returned by "show table" on the CLI). This may sometimes be more suited for
13850 layer7 tracking. It can be used to tell a server how many concurrent
13851 connections there are from a given address for example. See also the
13852 sc_trackers sample fetch keyword.
13853
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013854upper
13855 Convert a string sample to upper case. This can only be placed after a string
13856 sample fetch function or after a transformation keyword returning a string
13857 type. The result is of type string.
13858
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013859url_dec
13860 Takes an url-encoded string provided as input and returns the decoded
13861 version as output. The input and the output are of type string.
13862
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013863ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013864 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013865 sample representation of a gRPC message with <field_number> as field number
13866 (dotted notation) if <field_type> is not present, or as an integer sample if this
13867 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013868 The list of the authorized types is the following one: "int32", "int64", "uint32",
13869 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13870 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13871 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013872 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013873 More information may be found here about the protocol buffers message field types:
13874 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013875
13876 Example:
13877 // with such a protocol buffer .proto file content adapted from
13878 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
13879
13880 message Point {
13881 int32 latitude = 1;
13882 int32 longitude = 2;
13883 }
13884
13885 message PPoint {
13886 Point point = 59;
13887 }
13888
13889 message Rectangle {
13890 // One corner of the rectangle.
13891 PPoint lo = 48;
13892 // The other corner of the rectangle.
13893 PPoint hi = 49;
13894 }
13895
13896 let's say a body request is made of a "Rectangle" object value (two PPoint
13897 protocol buffers messages), the four protocol buffers fields could be
13898 extracted with these "ungrpc" directives:
13899
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013900 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
13901 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050013902 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013903 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
13904
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013905 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013906
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013907 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013908
John Roeslerfb2fce12019-07-10 15:45:51 -050013909 As a gRPC message is always made of a gRPC header followed by protocol buffers
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013910 messages, in the previous example the "latitude" of "lo" first PPoint
13911 could be extracted with these equivalent directives:
13912
13913 req.body,ungrpc(48.59),protobuf(1,int32)
13914 req.body,ungrpc(48),protobuf(59.1,int32)
13915 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
13916
13917 Note that the first convert must be "ungrpc", the remaining ones must be
13918 "protobuf" and only the last one may have or not a second argument to
13919 interpret the previous binary sample.
13920
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013921
Christopher Faulet85d79c92016-11-09 16:54:56 +010013922unset-var(<var name>)
13923 Unsets a variable if the input content is defined. The name of the variable
13924 starts with an indication about its scope. The scopes allowed are:
13925 "proc" : the variable is shared with the whole process
13926 "sess" : the variable is shared with the whole session
13927 "txn" : the variable is shared with the transaction (request and
13928 response),
13929 "req" : the variable is shared only during request processing,
13930 "res" : the variable is shared only during response processing.
13931 This prefix is followed by a name. The separator is a '.'. The name may only
13932 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13933
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013934utime(<format>[,<offset>])
13935 Converts an integer supposed to contain a date since epoch to a string
13936 representing this date in UTC time using a format defined by the <format>
13937 string using strftime(3). The purpose is to allow any date format to be used
13938 in logs. An optional <offset> in seconds may be applied to the input date
13939 (positive or negative). See the strftime() man page for the format supported
13940 by your operating system. See also the ltime converter.
13941
13942 Example :
13943
13944 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013945 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013946 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13947
Marcin Deranek9631a282018-04-16 14:30:46 +020013948word(<index>,<delimiters>[,<count>])
13949 Extracts the nth word counting from the beginning (positive index) or from
13950 the end (negative index) considering given delimiters from an input string.
13951 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
13952 Optionally you can specify <count> of words to extract (default: 1).
13953 Value of 0 indicates extraction of all remaining words.
13954
13955 Example :
13956 str(f1_f2_f3__f5),word(4,_) # f5
13957 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
13958 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
13959 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
13960 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013961
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013962wt6([<avalanche>])
13963 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13964 hash function. Optionally, it is possible to apply a full avalanche hash
13965 function to the output if the optional <avalanche> argument equals 1. This
13966 converter uses the same functions as used by the various hash-based load
13967 balancing algorithms, so it will provide exactly the same results. It is
13968 mostly intended for debugging, but can be used as a stick-table entry to
13969 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013970 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
13971 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013972
Willy Tarreau97707872015-01-27 15:12:13 +010013973xor(<value>)
13974 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013975 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013976 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013977 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013978 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013979 "sess" : the variable is shared with the whole session
13980 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013981 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013982 "req" : the variable is shared only during request processing,
13983 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013984 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013985 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013986
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013987xxh32([<seed>])
13988 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13989 variant of the XXHash hash function. This hash supports a seed which defaults
13990 to zero but a different value maybe passed as the <seed> argument. This hash
13991 is known to be very good and very fast so it can be used to hash URLs and/or
13992 URL parameters for use as stick-table keys to collect statistics with a low
13993 collision rate, though care must be taken as the algorithm is not considered
13994 as cryptographically secure.
13995
13996xxh64([<seed>])
13997 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13998 variant of the XXHash hash function. This hash supports a seed which defaults
13999 to zero but a different value maybe passed as the <seed> argument. This hash
14000 is known to be very good and very fast so it can be used to hash URLs and/or
14001 URL parameters for use as stick-table keys to collect statistics with a low
14002 collision rate, though care must be taken as the algorithm is not considered
14003 as cryptographically secure.
14004
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014005
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200140067.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020014007--------------------------------------------
14008
14009A first set of sample fetch methods applies to internal information which does
14010not even relate to any client information. These ones are sometimes used with
14011"monitor-fail" directives to report an internal status to external watchers.
14012The sample fetch methods described in this section are usable anywhere.
14013
14014always_false : boolean
14015 Always returns the boolean "false" value. It may be used with ACLs as a
14016 temporary replacement for another one when adjusting configurations.
14017
14018always_true : boolean
14019 Always returns the boolean "true" value. It may be used with ACLs as a
14020 temporary replacement for another one when adjusting configurations.
14021
14022avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014023 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020014024 divided by the number of active servers. The current backend is used if no
14025 backend is specified. This is very similar to "queue" except that the size of
14026 the farm is considered, in order to give a more accurate measurement of the
14027 time it may take for a new connection to be processed. The main usage is with
14028 ACL to return a sorry page to new users when it becomes certain they will get
14029 a degraded service, or to pass to the backend servers in a header so that
14030 they decide to work in degraded mode or to disable some functions to speed up
14031 the processing a bit. Note that in the event there would not be any active
14032 server anymore, twice the number of queued connections would be considered as
14033 the measured value. This is a fair estimate, as we expect one server to get
14034 back soon anyway, but we still prefer to send new traffic to another backend
14035 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
14036 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010014037
Willy Tarreau74ca5042013-06-11 23:12:07 +020014038be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020014039 Applies to the number of currently established connections on the backend,
14040 possibly including the connection being evaluated. If no backend name is
14041 specified, the current one is used. But it is also possible to check another
14042 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014043 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
14044
14045be_conn_free([<backend>]) : integer
14046 Returns an integer value corresponding to the number of available connections
14047 across available servers in the backend. Queue slots are not included. Backup
14048 servers are also not included, unless all other servers are down. If no
14049 backend name is specified, the current one is used. But it is also possible
14050 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014051 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
14052 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014053
14054 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
14055 (meaning unlimited), then this fetch clearly does not make sense, in which
14056 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014057
Willy Tarreau74ca5042013-06-11 23:12:07 +020014058be_sess_rate([<backend>]) : integer
14059 Returns an integer value corresponding to the sessions creation rate on the
14060 backend, in number of new sessions per second. This is used with ACLs to
14061 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010014062 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014063 online dictionary). It can also be useful to add this element to logs using a
14064 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014065
14066 Example :
14067 # Redirect to an error page if the dictionary is requested too often
14068 backend dynamic
14069 mode http
14070 acl being_scanned be_sess_rate gt 100
14071 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010014072
Davor Ocelice9ed2812017-12-25 17:49:28 +010014073bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014074 Returns a binary chain. The input is the hexadecimal representation
14075 of the string.
14076
14077bool(<bool>) : bool
14078 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14079 'false' and '0' are the same. 'true' and '1' are the same.
14080
Willy Tarreau74ca5042013-06-11 23:12:07 +020014081connslots([<backend>]) : integer
14082 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014083 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014084 connections on all servers and the maximum queue size. This is probably only
14085 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014086
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014087 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014088 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014089 usage; see "use_backend" keyword) can be redirected to a different backend.
14090
Willy Tarreau55165fe2009-05-10 12:02:55 +020014091 'connslots' = number of available server connection slots, + number of
14092 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014093
Willy Tarreaua36af912009-10-10 12:02:45 +020014094 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014095 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014096 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014097 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014098 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014099 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014100 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014101
Willy Tarreau55165fe2009-05-10 12:02:55 +020014102 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14103 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014104 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014105 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014106
Willy Tarreau70fe9442018-11-22 16:07:39 +010014107cpu_calls : integer
14108 Returns the number of calls to the task processing the stream or current
14109 request since it was allocated. This number is reset for each new request on
14110 the same connections in case of HTTP keep-alive. This value should usually be
14111 low and stable (around 2 calls for a typically simple request) but may become
14112 high if some processing (compression, caching or analysis) is performed. This
14113 is purely for performance monitoring purposes.
14114
14115cpu_ns_avg : integer
14116 Returns the average number of nanoseconds spent in each call to the task
14117 processing the stream or current request. This number is reset for each new
14118 request on the same connections in case of HTTP keep-alive. This value
14119 indicates the overall cost of processing the request or the connection for
14120 each call. There is no good nor bad value but the time spent in a call
14121 automatically causes latency for other processing (see lat_ns_avg below),
14122 and may affect other connection's apparent response time. Certain operations
14123 like compression, complex regex matching or heavy Lua operations may directly
14124 affect this value, and having it in the logs will make it easier to spot the
14125 faulty processing that needs to be fixed to recover decent performance.
14126 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14127
14128cpu_ns_tot : integer
14129 Returns the total number of nanoseconds spent in each call to the task
14130 processing the stream or current request. This number is reset for each new
14131 request on the same connections in case of HTTP keep-alive. This value
14132 indicates the overall cost of processing the request or the connection for
14133 each call. There is no good nor bad value but the time spent in a call
14134 automatically causes latency for other processing (see lat_ns_avg below),
14135 induces CPU costs on the machine, and may affect other connection's apparent
14136 response time. Certain operations like compression, complex regex matching or
14137 heavy Lua operations may directly affect this value, and having it in the
14138 logs will make it easier to spot the faulty processing that needs to be fixed
14139 to recover decent performance. The value may be artificially high due to a
14140 high cpu_calls count, for example when processing many HTTP chunks, and for
14141 this reason it is often preferred to log cpu_ns_avg instead.
14142
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014143date([<offset>]) : integer
14144 Returns the current date as the epoch (number of seconds since 01/01/1970).
14145 If an offset value is specified, then it is a number of seconds that is added
14146 to the current date before returning the value. This is particularly useful
14147 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014148 It is useful combined with the http_date converter.
14149
14150 Example :
14151
14152 # set an expires header to now+1 hour in every response
14153 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014154
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014155date_us : integer
14156 Return the microseconds part of the date (the "second" part is returned by
14157 date sample). This sample is coherent with the date sample as it is comes
14158 from the same timeval structure.
14159
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014160distcc_body(<token>[,<occ>]) : binary
14161 Parses a distcc message and returns the body associated to occurrence #<occ>
14162 of the token <token>. Occurrences start at 1, and when unspecified, any may
14163 match though in practice only the first one is checked for now. This can be
14164 used to extract file names or arguments in files built using distcc through
14165 haproxy. Please refer to distcc's protocol documentation for the complete
14166 list of supported tokens.
14167
14168distcc_param(<token>[,<occ>]) : integer
14169 Parses a distcc message and returns the parameter associated to occurrence
14170 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14171 any may match though in practice only the first one is checked for now. This
14172 can be used to extract certain information such as the protocol version, the
14173 file size or the argument in files built using distcc through haproxy.
14174 Another use case consists in waiting for the start of the preprocessed file
14175 contents before connecting to the server to avoid keeping idle connections.
14176 Please refer to distcc's protocol documentation for the complete list of
14177 supported tokens.
14178
14179 Example :
14180 # wait up to 20s for the pre-processed file to be uploaded
14181 tcp-request inspect-delay 20s
14182 tcp-request content accept if { distcc_param(DOTI) -m found }
14183 # send large files to the big farm
14184 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14185
Willy Tarreau595ec542013-06-12 21:34:28 +020014186env(<name>) : string
14187 Returns a string containing the value of environment variable <name>. As a
14188 reminder, environment variables are per-process and are sampled when the
14189 process starts. This can be useful to pass some information to a next hop
14190 server, or with ACLs to take specific action when the process is started a
14191 certain way.
14192
14193 Examples :
14194 # Pass the Via header to next hop with the local hostname in it
14195 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14196
14197 # reject cookie-less requests when the STOP environment variable is set
14198 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14199
Willy Tarreau74ca5042013-06-11 23:12:07 +020014200fe_conn([<frontend>]) : integer
14201 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014202 possibly including the connection being evaluated. If no frontend name is
14203 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014204 frontend. It can be used to return a sorry page before hard-blocking, or to
14205 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014206 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014207 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14208 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014209
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014210fe_req_rate([<frontend>]) : integer
14211 Returns an integer value corresponding to the number of HTTP requests per
14212 second sent to a frontend. This number can differ from "fe_sess_rate" in
14213 situations where client-side keep-alive is enabled.
14214
Willy Tarreau74ca5042013-06-11 23:12:07 +020014215fe_sess_rate([<frontend>]) : integer
14216 Returns an integer value corresponding to the sessions creation rate on the
14217 frontend, in number of new sessions per second. This is used with ACLs to
14218 limit the incoming session rate to an acceptable range in order to prevent
14219 abuse of service at the earliest moment, for example when combined with other
14220 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14221 down below the limit. It can also be useful to add this element to logs using
14222 a log-format directive. See also the "rate-limit sessions" directive for use
14223 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014224
14225 Example :
14226 # This frontend limits incoming mails to 10/s with a max of 100
14227 # concurrent connections. We accept any connection below 10/s, and
14228 # force excess clients to wait for 100 ms. Since clients are limited to
14229 # 100 max, there cannot be more than 10 incoming mails per second.
14230 frontend mail
14231 bind :25
14232 mode tcp
14233 maxconn 100
14234 acl too_fast fe_sess_rate ge 10
14235 tcp-request inspect-delay 100ms
14236 tcp-request content accept if ! too_fast
14237 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014238
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014239hostname : string
14240 Returns the system hostname.
14241
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014242int(<integer>) : signed integer
14243 Returns a signed integer.
14244
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014245ipv4(<ipv4>) : ipv4
14246 Returns an ipv4.
14247
14248ipv6(<ipv6>) : ipv6
14249 Returns an ipv6.
14250
Willy Tarreau70fe9442018-11-22 16:07:39 +010014251lat_ns_avg : integer
14252 Returns the average number of nanoseconds spent between the moment the task
14253 handling the stream is woken up and the moment it is effectively called. This
14254 number is reset for each new request on the same connections in case of HTTP
14255 keep-alive. This value indicates the overall latency inflicted to the current
14256 request by all other requests being processed in parallel, and is a direct
14257 indicator of perceived performance due to noisy neighbours. In order to keep
14258 the value low, it is possible to reduce the scheduler's run queue depth using
14259 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14260 once using "tune.maxpollevents", to decrease the stream's nice value using
14261 the "nice" option on the "bind" lines or in the frontend, or to look for
14262 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14263 whose processing needs to be adjusted or fixed. Compression of large buffers
14264 could be a culprit, like heavy regex or long lists of regex.
14265 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14266
14267lat_ns_tot : integer
14268 Returns the total number of nanoseconds spent between the moment the task
14269 handling the stream is woken up and the moment it is effectively called. This
14270 number is reset for each new request on the same connections in case of HTTP
14271 keep-alive. This value indicates the overall latency inflicted to the current
14272 request by all other requests being processed in parallel, and is a direct
14273 indicator of perceived performance due to noisy neighbours. In order to keep
14274 the value low, it is possible to reduce the scheduler's run queue depth using
14275 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14276 once using "tune.maxpollevents", to decrease the stream's nice value using
14277 the "nice" option on the "bind" lines or in the frontend, or to look for
14278 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14279 whose processing needs to be adjusted or fixed. Compression of large buffers
14280 could be a culprit, like heavy regex or long lists of regex. Note: while it
14281 may intuitively seem that the total latency adds to a transfer time, it is
14282 almost never true because while a task waits for the CPU, network buffers
14283 continue to fill up and the next call will process more at once. The value
14284 may be artificially high due to a high cpu_calls count, for example when
14285 processing many HTTP chunks, and for this reason it is often preferred to log
14286 lat_ns_avg instead, which is a more relevant performance indicator.
14287
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014288meth(<method>) : method
14289 Returns a method.
14290
Willy Tarreau0f30d262014-11-24 16:02:05 +010014291nbproc : integer
14292 Returns an integer value corresponding to the number of processes that were
14293 started (it equals the global "nbproc" setting). This is useful for logging
14294 and debugging purposes.
14295
Willy Tarreau74ca5042013-06-11 23:12:07 +020014296nbsrv([<backend>]) : integer
14297 Returns an integer value corresponding to the number of usable servers of
14298 either the current backend or the named backend. This is mostly used with
14299 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014300 switch to an alternate backend when the number of servers is too low to
14301 to handle some load. It is useful to report a failure when combined with
14302 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014303
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014304prio_class : integer
14305 Returns the priority class of the current session for http mode or connection
14306 for tcp mode. The value will be that set by the last call to "http-request
14307 set-priority-class" or "tcp-request content set-priority-class".
14308
14309prio_offset : integer
14310 Returns the priority offset of the current session for http mode or
14311 connection for tcp mode. The value will be that set by the last call to
14312 "http-request set-priority-offset" or "tcp-request content
14313 set-priority-offset".
14314
Willy Tarreau0f30d262014-11-24 16:02:05 +010014315proc : integer
14316 Returns an integer value corresponding to the position of the process calling
14317 the function, between 1 and global.nbproc. This is useful for logging and
14318 debugging purposes.
14319
Willy Tarreau74ca5042013-06-11 23:12:07 +020014320queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014321 Returns the total number of queued connections of the designated backend,
14322 including all the connections in server queues. If no backend name is
14323 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014324 one. This is useful with ACLs or to pass statistics to backend servers. This
14325 can be used to take actions when queuing goes above a known level, generally
14326 indicating a surge of traffic or a massive slowdown on the servers. One
14327 possible action could be to reject new users but still accept old ones. See
14328 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14329
Willy Tarreau84310e22014-02-14 11:59:04 +010014330rand([<range>]) : integer
14331 Returns a random integer value within a range of <range> possible values,
14332 starting at zero. If the range is not specified, it defaults to 2^32, which
14333 gives numbers between 0 and 4294967295. It can be useful to pass some values
14334 needed to take some routing decisions for example, or just for debugging
14335 purposes. This random must not be used for security purposes.
14336
Luca Schimweg8a694b82019-09-10 15:42:52 +020014337uuid([<version>]) : string
14338 Returns a UUID following the RFC4122 standard. If the version is not
14339 specified, a UUID version 4 (fully random) is returned.
14340 Currently, only version 4 is supported.
14341
Willy Tarreau74ca5042013-06-11 23:12:07 +020014342srv_conn([<backend>/]<server>) : integer
14343 Returns an integer value corresponding to the number of currently established
14344 connections on the designated server, possibly including the connection being
14345 evaluated. If <backend> is omitted, then the server is looked up in the
14346 current backend. It can be used to use a specific farm when one server is
14347 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014348 connections with it. See also the "fe_conn", "be_conn", "queue", and
14349 "srv_conn_free" fetch methods.
14350
14351srv_conn_free([<backend>/]<server>) : integer
14352 Returns an integer value corresponding to the number of available connections
14353 on the designated server, possibly including the connection being evaluated.
14354 The value does not include queue slots. If <backend> is omitted, then the
14355 server is looked up in the current backend. It can be used to use a specific
14356 farm when one server is full, or to inform the server about our view of the
14357 number of active connections with it. See also the "be_conn_free" and
14358 "srv_conn" fetch methods.
14359
14360 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14361 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014362
14363srv_is_up([<backend>/]<server>) : boolean
14364 Returns true when the designated server is UP, and false when it is either
14365 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14366 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014367 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014368 availability). Another possible use which is more of a hack consists in
14369 using dummy servers as boolean variables that can be enabled or disabled from
14370 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14371
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014372srv_queue([<backend>/]<server>) : integer
14373 Returns an integer value corresponding to the number of connections currently
14374 pending in the designated server's queue. If <backend> is omitted, then the
14375 server is looked up in the current backend. It can sometimes be used together
14376 with the "use-server" directive to force to use a known faster server when it
14377 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14378 fetch methods.
14379
Willy Tarreau74ca5042013-06-11 23:12:07 +020014380srv_sess_rate([<backend>/]<server>) : integer
14381 Returns an integer corresponding to the sessions creation rate on the
14382 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014383 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014384 used with ACLs but can make sense with logs too. This is used to switch to an
14385 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014386 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014387 overloading servers).
14388
14389 Example :
14390 # Redirect to a separate back
14391 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14392 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14393 use_backend be2 if srv1_full or srv2_full
14394
Willy Tarreau0f30d262014-11-24 16:02:05 +010014395stopping : boolean
14396 Returns TRUE if the process calling the function is currently stopping. This
14397 can be useful for logging, or for relaxing certain checks or helping close
14398 certain connections upon graceful shutdown.
14399
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014400str(<string>) : string
14401 Returns a string.
14402
Willy Tarreau74ca5042013-06-11 23:12:07 +020014403table_avl([<table>]) : integer
14404 Returns the total number of available entries in the current proxy's
14405 stick-table or in the designated stick-table. See also table_cnt.
14406
14407table_cnt([<table>]) : integer
14408 Returns the total number of entries currently in use in the current proxy's
14409 stick-table or in the designated stick-table. See also src_conn_cnt and
14410 table_avl for other entry counting methods.
14411
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014412thread : integer
14413 Returns an integer value corresponding to the position of the thread calling
14414 the function, between 0 and (global.nbthread-1). This is useful for logging
14415 and debugging purposes.
14416
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014417var(<var-name>) : undefined
14418 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014419 sample fetch fails. The name of the variable starts with an indication
14420 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014421 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014422 "sess" : the variable is shared with the whole session
14423 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014424 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014425 "req" : the variable is shared only during request processing,
14426 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014427 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014428 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014429
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200144307.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014431----------------------------------
14432
14433The layer 4 usually describes just the transport layer which in haproxy is
14434closest to the connection, where no content is yet made available. The fetch
14435methods described here are usable as low as the "tcp-request connection" rule
14436sets unless they require some future information. Those generally include
14437TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014438the incoming connection. For retrieving a value from a sticky counters, the
14439counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014440"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14441used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14442can be specified as the first integer argument when using the "sc_" prefix.
14443Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14444table may be specified with the "sc*" form, in which case the currently
14445tracked key will be looked up into this alternate table instead of the table
14446currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014447
Jérôme Magnin35e53a62019-01-16 14:38:37 +010014448bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010014449 Returns the backend connection's HTTP major version encoding, which may be 1
14450 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14451 encoding and not the version present in the request header.
14452
Willy Tarreau74ca5042013-06-11 23:12:07 +020014453be_id : integer
14454 Returns an integer containing the current backend's id. It can be used in
14455 frontends with responses to check which backend processed the request.
14456
Marcin Deranekd2471c22016-12-12 14:08:05 +010014457be_name : string
14458 Returns a string containing the current backend's name. It can be used in
14459 frontends with responses to check which backend processed the request.
14460
Willy Tarreau74ca5042013-06-11 23:12:07 +020014461dst : ip
14462 This is the destination IPv4 address of the connection on the client side,
14463 which is the address the client connected to. It can be useful when running
14464 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14465 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014466 RFC 4291. When the incoming connection passed through address translation or
14467 redirection involving connection tracking, the original destination address
14468 before the redirection will be reported. On Linux systems, the source and
14469 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
14470 is set, because a late response may reopen a timed out connection and switch
14471 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014472
14473dst_conn : integer
14474 Returns an integer value corresponding to the number of currently established
14475 connections on the same socket including the one being evaluated. It is
14476 normally used with ACLs but can as well be used to pass the information to
14477 servers in an HTTP header or in logs. It can be used to either return a sorry
14478 page before hard-blocking, or to use a specific backend to drain new requests
14479 when the socket is considered saturated. This offers the ability to assign
14480 different limits to different listening ports or addresses. See also the
14481 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014482
Willy Tarreau16e01562016-08-09 16:46:18 +020014483dst_is_local : boolean
14484 Returns true if the destination address of the incoming connection is local
14485 to the system, or false if the address doesn't exist on the system, meaning
14486 that it was intercepted in transparent mode. It can be useful to apply
14487 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014488 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014489 be delivered only on this address, or SSH access could be locally redirected.
14490 Please note that the check involves a few system calls, so it's better to do
14491 it only once per connection.
14492
Willy Tarreau74ca5042013-06-11 23:12:07 +020014493dst_port : integer
14494 Returns an integer value corresponding to the destination TCP port of the
14495 connection on the client side, which is the port the client connected to.
14496 This might be used when running in transparent mode, when assigning dynamic
14497 ports to some clients for a whole application session, to stick all users to
14498 a same server, or to pass the destination port information to a server using
14499 an HTTP header.
14500
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014501fc_http_major : integer
14502 Reports the front connection's HTTP major version encoding, which may be 1
14503 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14504 encoding and not on the version present in the request header.
14505
Geoff Simmons7185b782019-08-27 18:31:16 +020014506fc_pp_authority : string
14507 Returns the authority TLV sent by the client in the PROXY protocol header,
14508 if any.
14509
Emeric Brun4f603012017-01-05 15:11:44 +010014510fc_rcvd_proxy : boolean
14511 Returns true if the client initiated the connection with a PROXY protocol
14512 header.
14513
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014514fc_rtt(<unit>) : integer
14515 Returns the Round Trip Time (RTT) measured by the kernel for the client
14516 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14517 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14518 connection is not established, if the connection is not TCP or if the
14519 operating system does not support TCP_INFO, for example Linux kernels before
14520 2.4, the sample fetch fails.
14521
14522fc_rttvar(<unit>) : integer
14523 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14524 client connection. <unit> is facultative, by default the unit is milliseconds.
14525 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14526 server connection is not established, if the connection is not TCP or if the
14527 operating system does not support TCP_INFO, for example Linux kernels before
14528 2.4, the sample fetch fails.
14529
Joe Williams30fcd392016-08-10 07:06:44 -070014530fc_unacked(<unit>) : integer
14531 Returns the unacked counter measured by the kernel for the client connection.
14532 If the server connection is not established, if the connection is not TCP or
14533 if the operating system does not support TCP_INFO, for example Linux kernels
14534 before 2.4, the sample fetch fails.
14535
14536fc_sacked(<unit>) : integer
14537 Returns the sacked counter measured by the kernel for the client connection.
14538 If the server connection is not established, if the connection is not TCP or
14539 if the operating system does not support TCP_INFO, for example Linux kernels
14540 before 2.4, the sample fetch fails.
14541
14542fc_retrans(<unit>) : integer
14543 Returns the retransmits counter measured by the kernel for the client
14544 connection. If the server connection is not established, if the connection is
14545 not TCP or if the operating system does not support TCP_INFO, for example
14546 Linux kernels before 2.4, the sample fetch fails.
14547
14548fc_fackets(<unit>) : integer
14549 Returns the fack counter measured by the kernel for the client
14550 connection. If the server connection is not established, if the connection is
14551 not TCP or if the operating system does not support TCP_INFO, for example
14552 Linux kernels before 2.4, the sample fetch fails.
14553
14554fc_lost(<unit>) : integer
14555 Returns the lost counter measured by the kernel for the client
14556 connection. If the server connection is not established, if the connection is
14557 not TCP or if the operating system does not support TCP_INFO, for example
14558 Linux kernels before 2.4, the sample fetch fails.
14559
14560fc_reordering(<unit>) : integer
14561 Returns the reordering counter measured by the kernel for the client
14562 connection. If the server connection is not established, if the connection is
14563 not TCP or if the operating system does not support TCP_INFO, for example
14564 Linux kernels before 2.4, the sample fetch fails.
14565
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014566fe_defbe : string
14567 Returns a string containing the frontend's default backend name. It can be
14568 used in frontends to check which backend will handle requests by default.
14569
Willy Tarreau74ca5042013-06-11 23:12:07 +020014570fe_id : integer
14571 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014572 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014573 coming via a same frontend to the same server.
14574
Marcin Deranekd2471c22016-12-12 14:08:05 +010014575fe_name : string
14576 Returns a string containing the current frontend's name. It can be used in
14577 backends to check from which frontend it was called, or to stick all users
14578 coming via a same frontend to the same server.
14579
Cyril Bonté62ba8702014-04-22 23:52:25 +020014580sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014581sc0_bytes_in_rate([<table>]) : integer
14582sc1_bytes_in_rate([<table>]) : integer
14583sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014584 Returns the average client-to-server bytes rate from the currently tracked
14585 counters, measured in amount of bytes over the period configured in the
14586 table. See also src_bytes_in_rate.
14587
Cyril Bonté62ba8702014-04-22 23:52:25 +020014588sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014589sc0_bytes_out_rate([<table>]) : integer
14590sc1_bytes_out_rate([<table>]) : integer
14591sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014592 Returns the average server-to-client bytes rate from the currently tracked
14593 counters, measured in amount of bytes over the period configured in the
14594 table. See also src_bytes_out_rate.
14595
Cyril Bonté62ba8702014-04-22 23:52:25 +020014596sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014597sc0_clr_gpc0([<table>]) : integer
14598sc1_clr_gpc0([<table>]) : integer
14599sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014600 Clears the first General Purpose Counter associated to the currently tracked
14601 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014602 stored value is zero, so first invocation will always return zero. This is
14603 typically used as a second ACL in an expression in order to mark a connection
14604 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014605
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014606 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014607 # block if 5 consecutive requests continue to come faster than 10 sess
14608 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014609 acl abuse sc0_http_req_rate gt 10
14610 acl kill sc0_inc_gpc0 gt 5
14611 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014612 tcp-request connection accept if !abuse save
14613 tcp-request connection reject if abuse kill
14614
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014615sc_clr_gpc1(<ctr>[,<table>]) : integer
14616sc0_clr_gpc1([<table>]) : integer
14617sc1_clr_gpc1([<table>]) : integer
14618sc2_clr_gpc1([<table>]) : integer
14619 Clears the second General Purpose Counter associated to the currently tracked
14620 counters, and returns its previous value. Before the first invocation, the
14621 stored value is zero, so first invocation will always return zero. This is
14622 typically used as a second ACL in an expression in order to mark a connection
14623 when a first ACL was verified.
14624
Cyril Bonté62ba8702014-04-22 23:52:25 +020014625sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014626sc0_conn_cnt([<table>]) : integer
14627sc1_conn_cnt([<table>]) : integer
14628sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014629 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014630 counters. See also src_conn_cnt.
14631
Cyril Bonté62ba8702014-04-22 23:52:25 +020014632sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014633sc0_conn_cur([<table>]) : integer
14634sc1_conn_cur([<table>]) : integer
14635sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014636 Returns the current amount of concurrent connections tracking the same
14637 tracked counters. This number is automatically incremented when tracking
14638 begins and decremented when tracking stops. See also src_conn_cur.
14639
Cyril Bonté62ba8702014-04-22 23:52:25 +020014640sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014641sc0_conn_rate([<table>]) : integer
14642sc1_conn_rate([<table>]) : integer
14643sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014644 Returns the average connection rate from the currently tracked counters,
14645 measured in amount of connections over the period configured in the table.
14646 See also src_conn_rate.
14647
Cyril Bonté62ba8702014-04-22 23:52:25 +020014648sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014649sc0_get_gpc0([<table>]) : integer
14650sc1_get_gpc0([<table>]) : integer
14651sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014652 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014653 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014654
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014655sc_get_gpc1(<ctr>[,<table>]) : integer
14656sc0_get_gpc1([<table>]) : integer
14657sc1_get_gpc1([<table>]) : integer
14658sc2_get_gpc1([<table>]) : integer
14659 Returns the value of the second General Purpose Counter associated to the
14660 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14661
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014662sc_get_gpt0(<ctr>[,<table>]) : integer
14663sc0_get_gpt0([<table>]) : integer
14664sc1_get_gpt0([<table>]) : integer
14665sc2_get_gpt0([<table>]) : integer
14666 Returns the value of the first General Purpose Tag associated to the
14667 currently tracked counters. See also src_get_gpt0.
14668
Cyril Bonté62ba8702014-04-22 23:52:25 +020014669sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014670sc0_gpc0_rate([<table>]) : integer
14671sc1_gpc0_rate([<table>]) : integer
14672sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014673 Returns the average increment rate of the first General Purpose Counter
14674 associated to the currently tracked counters. It reports the frequency
14675 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014676 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14677 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14678 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014679
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014680sc_gpc1_rate(<ctr>[,<table>]) : integer
14681sc0_gpc1_rate([<table>]) : integer
14682sc1_gpc1_rate([<table>]) : integer
14683sc2_gpc1_rate([<table>]) : integer
14684 Returns the average increment rate of the second General Purpose Counter
14685 associated to the currently tracked counters. It reports the frequency
14686 which the gpc1 counter was incremented over the configured period. See also
14687 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14688 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14689 be returned, as "gpc1" only holds the event count.
14690
Cyril Bonté62ba8702014-04-22 23:52:25 +020014691sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014692sc0_http_err_cnt([<table>]) : integer
14693sc1_http_err_cnt([<table>]) : integer
14694sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014695 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014696 counters. This includes the both request errors and 4xx error responses.
14697 See also src_http_err_cnt.
14698
Cyril Bonté62ba8702014-04-22 23:52:25 +020014699sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014700sc0_http_err_rate([<table>]) : integer
14701sc1_http_err_rate([<table>]) : integer
14702sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014703 Returns the average rate of HTTP errors from the currently tracked counters,
14704 measured in amount of errors over the period configured in the table. This
14705 includes the both request errors and 4xx error responses. See also
14706 src_http_err_rate.
14707
Cyril Bonté62ba8702014-04-22 23:52:25 +020014708sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014709sc0_http_req_cnt([<table>]) : integer
14710sc1_http_req_cnt([<table>]) : integer
14711sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014712 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014713 counters. This includes every started request, valid or not. See also
14714 src_http_req_cnt.
14715
Cyril Bonté62ba8702014-04-22 23:52:25 +020014716sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014717sc0_http_req_rate([<table>]) : integer
14718sc1_http_req_rate([<table>]) : integer
14719sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014720 Returns the average rate of HTTP requests from the currently tracked
14721 counters, measured in amount of requests over the period configured in
14722 the table. This includes every started request, valid or not. See also
14723 src_http_req_rate.
14724
Cyril Bonté62ba8702014-04-22 23:52:25 +020014725sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014726sc0_inc_gpc0([<table>]) : integer
14727sc1_inc_gpc0([<table>]) : integer
14728sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014729 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014730 tracked counters, and returns its new value. Before the first invocation,
14731 the stored value is zero, so first invocation will increase it to 1 and will
14732 return 1. This is typically used as a second ACL in an expression in order
14733 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014734
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014735 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014736 acl abuse sc0_http_req_rate gt 10
14737 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014738 tcp-request connection reject if abuse kill
14739
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014740sc_inc_gpc1(<ctr>[,<table>]) : integer
14741sc0_inc_gpc1([<table>]) : integer
14742sc1_inc_gpc1([<table>]) : integer
14743sc2_inc_gpc1([<table>]) : integer
14744 Increments the second General Purpose Counter associated to the currently
14745 tracked counters, and returns its new value. Before the first invocation,
14746 the stored value is zero, so first invocation will increase it to 1 and will
14747 return 1. This is typically used as a second ACL in an expression in order
14748 to mark a connection when a first ACL was verified.
14749
Cyril Bonté62ba8702014-04-22 23:52:25 +020014750sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014751sc0_kbytes_in([<table>]) : integer
14752sc1_kbytes_in([<table>]) : integer
14753sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014754 Returns the total amount of client-to-server data from the currently tracked
14755 counters, measured in kilobytes. The test is currently performed on 32-bit
14756 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014757
Cyril Bonté62ba8702014-04-22 23:52:25 +020014758sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014759sc0_kbytes_out([<table>]) : integer
14760sc1_kbytes_out([<table>]) : integer
14761sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014762 Returns the total amount of server-to-client data from the currently tracked
14763 counters, measured in kilobytes. The test is currently performed on 32-bit
14764 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014765
Cyril Bonté62ba8702014-04-22 23:52:25 +020014766sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014767sc0_sess_cnt([<table>]) : integer
14768sc1_sess_cnt([<table>]) : integer
14769sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014770 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014771 into sessions, which means that they were accepted by a "tcp-request
14772 connection" rule, from the currently tracked counters. A backend may count
14773 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014774 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014775 with the client. See also src_sess_cnt.
14776
Cyril Bonté62ba8702014-04-22 23:52:25 +020014777sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014778sc0_sess_rate([<table>]) : integer
14779sc1_sess_rate([<table>]) : integer
14780sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014781 Returns the average session rate from the currently tracked counters,
14782 measured in amount of sessions over the period configured in the table. A
14783 session is a connection that got past the early "tcp-request connection"
14784 rules. A backend may count more sessions than connections because each
14785 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014786 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014787
Cyril Bonté62ba8702014-04-22 23:52:25 +020014788sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014789sc0_tracked([<table>]) : boolean
14790sc1_tracked([<table>]) : boolean
14791sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014792 Returns true if the designated session counter is currently being tracked by
14793 the current session. This can be useful when deciding whether or not we want
14794 to set some values in a header passed to the server.
14795
Cyril Bonté62ba8702014-04-22 23:52:25 +020014796sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014797sc0_trackers([<table>]) : integer
14798sc1_trackers([<table>]) : integer
14799sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014800 Returns the current amount of concurrent connections tracking the same
14801 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014802 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014803 that it does not rely on any stored information but on the table's reference
14804 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014805 may sometimes be more suited for layer7 tracking. It can be used to tell a
14806 server how many concurrent connections there are from a given address for
14807 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014808
Willy Tarreau74ca5042013-06-11 23:12:07 +020014809so_id : integer
14810 Returns an integer containing the current listening socket's id. It is useful
14811 in frontends involving many "bind" lines, or to stick all users coming via a
14812 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014813
Willy Tarreau74ca5042013-06-11 23:12:07 +020014814src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014815 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014816 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14817 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14818 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014819 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14820 directive is used, it can be the address of a client behind another
14821 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014822 "tcp-request connection" which sees the real address. When the incoming
14823 connection passed through address translation or redirection involving
14824 connection tracking, the original destination address before the redirection
14825 will be reported. On Linux systems, the source and destination may seldom
14826 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
14827 response may reopen a timed out connection and switch what is believed to be
14828 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014829
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014830 Example:
14831 # add an HTTP header in requests with the originating address' country
14832 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14833
Willy Tarreau74ca5042013-06-11 23:12:07 +020014834src_bytes_in_rate([<table>]) : integer
14835 Returns the average bytes rate from the incoming connection's source address
14836 in the current proxy's stick-table or in the designated stick-table, measured
14837 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014838 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014839
Willy Tarreau74ca5042013-06-11 23:12:07 +020014840src_bytes_out_rate([<table>]) : integer
14841 Returns the average bytes rate to the incoming connection's source address in
14842 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014843 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014844 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014845
Willy Tarreau74ca5042013-06-11 23:12:07 +020014846src_clr_gpc0([<table>]) : integer
14847 Clears the first General Purpose Counter associated to the incoming
14848 connection's source address in the current proxy's stick-table or in the
14849 designated stick-table, and returns its previous value. If the address is not
14850 found, an entry is created and 0 is returned. This is typically used as a
14851 second ACL in an expression in order to mark a connection when a first ACL
14852 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014853
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014854 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014855 # block if 5 consecutive requests continue to come faster than 10 sess
14856 # per second, and reset the counter as soon as the traffic slows down.
14857 acl abuse src_http_req_rate gt 10
14858 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014859 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014860 tcp-request connection accept if !abuse save
14861 tcp-request connection reject if abuse kill
14862
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014863src_clr_gpc1([<table>]) : integer
14864 Clears the second General Purpose Counter associated to the incoming
14865 connection's source address in the current proxy's stick-table or in the
14866 designated stick-table, and returns its previous value. If the address is not
14867 found, an entry is created and 0 is returned. This is typically used as a
14868 second ACL in an expression in order to mark a connection when a first ACL
14869 was verified.
14870
Willy Tarreau74ca5042013-06-11 23:12:07 +020014871src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014872 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014873 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014874 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014875 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014876
Willy Tarreau74ca5042013-06-11 23:12:07 +020014877src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014878 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014879 current incoming connection's source address in the current proxy's
14880 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014881 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014882
Willy Tarreau74ca5042013-06-11 23:12:07 +020014883src_conn_rate([<table>]) : integer
14884 Returns the average connection rate from the incoming connection's source
14885 address in the current proxy's stick-table or in the designated stick-table,
14886 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014887 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014888
Willy Tarreau74ca5042013-06-11 23:12:07 +020014889src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014890 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014891 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014892 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014893 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014894
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014895src_get_gpc1([<table>]) : integer
14896 Returns the value of the second General Purpose Counter associated to the
14897 incoming connection's source address in the current proxy's stick-table or in
14898 the designated stick-table. If the address is not found, zero is returned.
14899 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14900
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014901src_get_gpt0([<table>]) : integer
14902 Returns the value of the first General Purpose Tag associated to the
14903 incoming connection's source address in the current proxy's stick-table or in
14904 the designated stick-table. If the address is not found, zero is returned.
14905 See also sc/sc0/sc1/sc2_get_gpt0.
14906
Willy Tarreau74ca5042013-06-11 23:12:07 +020014907src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014908 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014909 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014910 stick-table or in the designated stick-table. It reports the frequency
14911 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014912 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14913 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14914 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014915
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014916src_gpc1_rate([<table>]) : integer
14917 Returns the average increment rate of the second General Purpose Counter
14918 associated to the incoming connection's source address in the current proxy's
14919 stick-table or in the designated stick-table. It reports the frequency
14920 which the gpc1 counter was incremented over the configured period. See also
14921 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14922 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14923 be returned, as "gpc1" only holds the event count.
14924
Willy Tarreau74ca5042013-06-11 23:12:07 +020014925src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014926 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014927 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014928 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014929 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014930 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014931
Willy Tarreau74ca5042013-06-11 23:12:07 +020014932src_http_err_rate([<table>]) : integer
14933 Returns the average rate of HTTP errors from the incoming connection's source
14934 address in the current proxy's stick-table or in the designated stick-table,
14935 measured in amount of errors over the period configured in the table. This
14936 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014937 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014938
Willy Tarreau74ca5042013-06-11 23:12:07 +020014939src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014940 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014941 source address in the current proxy's stick-table or in the designated stick-
14942 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014943 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014944
Willy Tarreau74ca5042013-06-11 23:12:07 +020014945src_http_req_rate([<table>]) : integer
14946 Returns the average rate of HTTP requests from the incoming connection's
14947 source address in the current proxy's stick-table or in the designated stick-
14948 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014949 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014950 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014951
Willy Tarreau74ca5042013-06-11 23:12:07 +020014952src_inc_gpc0([<table>]) : integer
14953 Increments the first General Purpose Counter associated to the incoming
14954 connection's source address in the current proxy's stick-table or in the
14955 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014956 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014957 This is typically used as a second ACL in an expression in order to mark a
14958 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014959
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014960 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014961 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014962 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014963 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014964
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014965src_inc_gpc1([<table>]) : integer
14966 Increments the second General Purpose Counter associated to the incoming
14967 connection's source address in the current proxy's stick-table or in the
14968 designated stick-table, and returns its new value. If the address is not
14969 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14970 This is typically used as a second ACL in an expression in order to mark a
14971 connection when a first ACL was verified.
14972
Willy Tarreau16e01562016-08-09 16:46:18 +020014973src_is_local : boolean
14974 Returns true if the source address of the incoming connection is local to the
14975 system, or false if the address doesn't exist on the system, meaning that it
14976 comes from a remote machine. Note that UNIX addresses are considered local.
14977 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014978 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014979 note that the check involves a few system calls, so it's better to do it only
14980 once per connection.
14981
Willy Tarreau74ca5042013-06-11 23:12:07 +020014982src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014983 Returns the total amount of data received from the incoming connection's
14984 source address in the current proxy's stick-table or in the designated
14985 stick-table, measured in kilobytes. If the address is not found, zero is
14986 returned. The test is currently performed on 32-bit integers, which limits
14987 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014988
Willy Tarreau74ca5042013-06-11 23:12:07 +020014989src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014990 Returns the total amount of data sent to the incoming connection's source
14991 address in the current proxy's stick-table or in the designated stick-table,
14992 measured in kilobytes. If the address is not found, zero is returned. The
14993 test is currently performed on 32-bit integers, which limits values to 4
14994 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014995
Willy Tarreau74ca5042013-06-11 23:12:07 +020014996src_port : integer
14997 Returns an integer value corresponding to the TCP source port of the
14998 connection on the client side, which is the port the client connected from.
14999 Usage of this function is very limited as modern protocols do not care much
15000 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010015001
Willy Tarreau74ca5042013-06-11 23:12:07 +020015002src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015003 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020015004 connection's source IPv4 address in the current proxy's stick-table or in the
15005 designated stick-table, that were transformed into sessions, which means that
15006 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015007 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015008
Willy Tarreau74ca5042013-06-11 23:12:07 +020015009src_sess_rate([<table>]) : integer
15010 Returns the average session rate from the incoming connection's source
15011 address in the current proxy's stick-table or in the designated stick-table,
15012 measured in amount of sessions over the period configured in the table. A
15013 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015014 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015015
Willy Tarreau74ca5042013-06-11 23:12:07 +020015016src_updt_conn_cnt([<table>]) : integer
15017 Creates or updates the entry associated to the incoming connection's source
15018 address in the current proxy's stick-table or in the designated stick-table.
15019 This table must be configured to store the "conn_cnt" data type, otherwise
15020 the match will be ignored. The current count is incremented by one, and the
15021 expiration timer refreshed. The updated count is returned, so this match
15022 can't return zero. This was used to reject service abusers based on their
15023 source address. Note: it is recommended to use the more complete "track-sc*"
15024 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015025
15026 Example :
15027 # This frontend limits incoming SSH connections to 3 per 10 second for
15028 # each source address, and rejects excess connections until a 10 second
15029 # silence is observed. At most 20 addresses are tracked.
15030 listen ssh
15031 bind :22
15032 mode tcp
15033 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020015034 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020015035 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015036 server local 127.0.0.1:22
15037
Willy Tarreau74ca5042013-06-11 23:12:07 +020015038srv_id : integer
15039 Returns an integer containing the server's id when processing the response.
15040 While it's almost only used with ACLs, it may be used for logging or
15041 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020015042
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200150437.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020015044----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020015045
Willy Tarreau74ca5042013-06-11 23:12:07 +020015046The layer 5 usually describes just the session layer which in haproxy is
15047closest to the session once all the connection handshakes are finished, but
15048when no content is yet made available. The fetch methods described here are
15049usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015050future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020015051
Ben Shillitof25e8e52016-12-02 14:25:37 +00001505251d.all(<prop>[,<prop>*]) : string
15053 Returns values for the properties requested as a string, where values are
15054 separated by the delimiter specified with "51degrees-property-separator".
15055 The device is identified using all the important HTTP headers from the
15056 request. The function can be passed up to five property names, and if a
15057 property name can't be found, the value "NoData" is returned.
15058
15059 Example :
15060 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
15061 # containing the three properties requested using all relevant headers from
15062 # the request.
15063 frontend http-in
15064 bind *:8081
15065 default_backend servers
15066 http-request set-header X-51D-DeviceTypeMobileTablet \
15067 %[51d.all(DeviceType,IsMobile,IsTablet)]
15068
Emeric Brun645ae792014-04-30 14:21:06 +020015069ssl_bc : boolean
15070 Returns true when the back connection was made via an SSL/TLS transport
15071 layer and is locally deciphered. This means the outgoing connection was made
15072 other a server with the "ssl" option.
15073
15074ssl_bc_alg_keysize : integer
15075 Returns the symmetric cipher key size supported in bits when the outgoing
15076 connection was made over an SSL/TLS transport layer.
15077
Olivier Houchard6b77f492018-11-22 18:18:29 +010015078ssl_bc_alpn : string
15079 This extracts the Application Layer Protocol Negotiation field from an
15080 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020015081 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010015082 server. The SSL library must have been built with support for TLS
15083 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15084 not advertised unless the "alpn" keyword on the "server" line specifies a
15085 protocol list. Also, nothing forces the server to pick a protocol from this
15086 list, any other one may be requested. The TLS ALPN extension is meant to
15087 replace the TLS NPN extension. See also "ssl_bc_npn".
15088
Emeric Brun645ae792014-04-30 14:21:06 +020015089ssl_bc_cipher : string
15090 Returns the name of the used cipher when the outgoing connection was made
15091 over an SSL/TLS transport layer.
15092
Patrick Hemmer65674662019-06-04 08:13:03 -040015093ssl_bc_client_random : binary
15094 Returns the client random of the back connection when the incoming connection
15095 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15096 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15097
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015098ssl_bc_is_resumed : boolean
15099 Returns true when the back connection was made over an SSL/TLS transport
15100 layer and the newly created SSL session was resumed using a cached
15101 session or a TLS ticket.
15102
Olivier Houchard6b77f492018-11-22 18:18:29 +010015103ssl_bc_npn : string
15104 This extracts the Next Protocol Negotiation field from an outgoing connection
15105 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020015106 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010015107 built with support for TLS extensions enabled (check haproxy -vv). Note that
15108 the TLS NPN extension is not advertised unless the "npn" keyword on the
15109 "server" line specifies a protocol list. Also, nothing forces the server to
15110 pick a protocol from this list, any other one may be used. Please note that
15111 the TLS NPN extension was replaced with ALPN.
15112
Emeric Brun645ae792014-04-30 14:21:06 +020015113ssl_bc_protocol : string
15114 Returns the name of the used protocol when the outgoing connection was made
15115 over an SSL/TLS transport layer.
15116
Emeric Brunb73a9b02014-04-30 18:49:19 +020015117ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015118 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015119 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15120 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015121
Patrick Hemmer65674662019-06-04 08:13:03 -040015122ssl_bc_server_random : binary
15123 Returns the server random of the back connection when the incoming connection
15124 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15125 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15126
Emeric Brun645ae792014-04-30 14:21:06 +020015127ssl_bc_session_id : binary
15128 Returns the SSL ID of the back connection when the outgoing connection was
15129 made over an SSL/TLS transport layer. It is useful to log if we want to know
15130 if session was reused or not.
15131
Patrick Hemmere0275472018-04-28 19:15:51 -040015132ssl_bc_session_key : binary
15133 Returns the SSL session master key of the back connection when the outgoing
15134 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15135 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15136 BoringSSL.
15137
Emeric Brun645ae792014-04-30 14:21:06 +020015138ssl_bc_use_keysize : integer
15139 Returns the symmetric cipher key size used in bits when the outgoing
15140 connection was made over an SSL/TLS transport layer.
15141
Willy Tarreau74ca5042013-06-11 23:12:07 +020015142ssl_c_ca_err : integer
15143 When the incoming connection was made over an SSL/TLS transport layer,
15144 returns the ID of the first error detected during verification of the client
15145 certificate at depth > 0, or 0 if no error was encountered during this
15146 verification process. Please refer to your SSL library's documentation to
15147 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015148
Willy Tarreau74ca5042013-06-11 23:12:07 +020015149ssl_c_ca_err_depth : integer
15150 When the incoming connection was made over an SSL/TLS transport layer,
15151 returns the depth in the CA chain of the first error detected during the
15152 verification of the client certificate. If no error is encountered, 0 is
15153 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015154
Emeric Brun43e79582014-10-29 19:03:26 +010015155ssl_c_der : binary
15156 Returns the DER formatted certificate presented by the client when the
15157 incoming connection was made over an SSL/TLS transport layer. When used for
15158 an ACL, the value(s) to match against can be passed in hexadecimal form.
15159
Willy Tarreau74ca5042013-06-11 23:12:07 +020015160ssl_c_err : integer
15161 When the incoming connection was made over an SSL/TLS transport layer,
15162 returns the ID of the first error detected during verification at depth 0, or
15163 0 if no error was encountered during this verification process. Please refer
15164 to your SSL library's documentation to find the exhaustive list of error
15165 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015166
Willy Tarreau74ca5042013-06-11 23:12:07 +020015167ssl_c_i_dn([<entry>[,<occ>]]) : string
15168 When the incoming connection was made over an SSL/TLS transport layer,
15169 returns the full distinguished name of the issuer of the certificate
15170 presented by the client when no <entry> is specified, or the value of the
15171 first given entry found from the beginning of the DN. If a positive/negative
15172 occurrence number is specified as the optional second argument, it returns
15173 the value of the nth given entry value from the beginning/end of the DN.
15174 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15175 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020015176
Willy Tarreau74ca5042013-06-11 23:12:07 +020015177ssl_c_key_alg : string
15178 Returns the name of the algorithm used to generate the key of the certificate
15179 presented by the client when the incoming connection was made over an SSL/TLS
15180 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015181
Willy Tarreau74ca5042013-06-11 23:12:07 +020015182ssl_c_notafter : string
15183 Returns the end date presented by the client as a formatted string
15184 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15185 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015186
Willy Tarreau74ca5042013-06-11 23:12:07 +020015187ssl_c_notbefore : string
15188 Returns the start date presented by the client as a formatted string
15189 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15190 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015191
Willy Tarreau74ca5042013-06-11 23:12:07 +020015192ssl_c_s_dn([<entry>[,<occ>]]) : string
15193 When the incoming connection was made over an SSL/TLS transport layer,
15194 returns the full distinguished name of the subject of the certificate
15195 presented by the client when no <entry> is specified, or the value of the
15196 first given entry found from the beginning of the DN. If a positive/negative
15197 occurrence number is specified as the optional second argument, it returns
15198 the value of the nth given entry value from the beginning/end of the DN.
15199 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15200 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015201
Willy Tarreau74ca5042013-06-11 23:12:07 +020015202ssl_c_serial : binary
15203 Returns the serial of the certificate presented by the client when the
15204 incoming connection was made over an SSL/TLS transport layer. When used for
15205 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015206
Willy Tarreau74ca5042013-06-11 23:12:07 +020015207ssl_c_sha1 : binary
15208 Returns the SHA-1 fingerprint of the certificate presented by the client when
15209 the incoming connection was made over an SSL/TLS transport layer. This can be
15210 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015211 Note that the output is binary, so if you want to pass that signature to the
15212 server, you need to encode it in hex or base64, such as in the example below:
15213
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015214 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015215 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015216
Willy Tarreau74ca5042013-06-11 23:12:07 +020015217ssl_c_sig_alg : string
15218 Returns the name of the algorithm used to sign the certificate presented by
15219 the client when the incoming connection was made over an SSL/TLS transport
15220 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015221
Willy Tarreau74ca5042013-06-11 23:12:07 +020015222ssl_c_used : boolean
15223 Returns true if current SSL session uses a client certificate even if current
15224 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015225
Willy Tarreau74ca5042013-06-11 23:12:07 +020015226ssl_c_verify : integer
15227 Returns the verify result error ID when the incoming connection was made over
15228 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15229 refer to your SSL library's documentation for an exhaustive list of error
15230 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015231
Willy Tarreau74ca5042013-06-11 23:12:07 +020015232ssl_c_version : integer
15233 Returns the version of the certificate presented by the client when the
15234 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015235
Emeric Brun43e79582014-10-29 19:03:26 +010015236ssl_f_der : binary
15237 Returns the DER formatted certificate presented by the frontend when the
15238 incoming connection was made over an SSL/TLS transport layer. When used for
15239 an ACL, the value(s) to match against can be passed in hexadecimal form.
15240
Willy Tarreau74ca5042013-06-11 23:12:07 +020015241ssl_f_i_dn([<entry>[,<occ>]]) : string
15242 When the incoming connection was made over an SSL/TLS transport layer,
15243 returns the full distinguished name of the issuer of the certificate
15244 presented by the frontend when no <entry> is specified, or the value of the
15245 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015246 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015247 the value of the nth given entry value from the beginning/end of the DN.
15248 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15249 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020015250
Willy Tarreau74ca5042013-06-11 23:12:07 +020015251ssl_f_key_alg : string
15252 Returns the name of the algorithm used to generate the key of the certificate
15253 presented by the frontend when the incoming connection was made over an
15254 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015255
Willy Tarreau74ca5042013-06-11 23:12:07 +020015256ssl_f_notafter : string
15257 Returns the end date presented by the frontend as a formatted string
15258 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15259 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015260
Willy Tarreau74ca5042013-06-11 23:12:07 +020015261ssl_f_notbefore : string
15262 Returns the start date presented by the frontend as a formatted string
15263 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15264 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015265
Willy Tarreau74ca5042013-06-11 23:12:07 +020015266ssl_f_s_dn([<entry>[,<occ>]]) : string
15267 When the incoming connection was made over an SSL/TLS transport layer,
15268 returns the full distinguished name of the subject of the certificate
15269 presented by the frontend when no <entry> is specified, or the value of the
15270 first given entry found from the beginning of the DN. If a positive/negative
15271 occurrence number is specified as the optional second argument, it returns
15272 the value of the nth given entry value from the beginning/end of the DN.
15273 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15274 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020015275
Willy Tarreau74ca5042013-06-11 23:12:07 +020015276ssl_f_serial : binary
15277 Returns the serial of the certificate presented by the frontend when the
15278 incoming connection was made over an SSL/TLS transport layer. When used for
15279 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015280
Emeric Brun55f4fa82014-04-30 17:11:25 +020015281ssl_f_sha1 : binary
15282 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15283 when the incoming connection was made over an SSL/TLS transport layer. This
15284 can be used to know which certificate was chosen using SNI.
15285
Willy Tarreau74ca5042013-06-11 23:12:07 +020015286ssl_f_sig_alg : string
15287 Returns the name of the algorithm used to sign the certificate presented by
15288 the frontend when the incoming connection was made over an SSL/TLS transport
15289 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015290
Willy Tarreau74ca5042013-06-11 23:12:07 +020015291ssl_f_version : integer
15292 Returns the version of the certificate presented by the frontend when the
15293 incoming connection was made over an SSL/TLS transport layer.
15294
15295ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015296 Returns true when the front connection was made via an SSL/TLS transport
15297 layer and is locally deciphered. This means it has matched a socket declared
15298 with a "bind" line having the "ssl" option.
15299
Willy Tarreau74ca5042013-06-11 23:12:07 +020015300 Example :
15301 # This passes "X-Proto: https" to servers when client connects over SSL
15302 listen http-https
15303 bind :80
15304 bind :443 ssl crt /etc/haproxy.pem
15305 http-request add-header X-Proto https if { ssl_fc }
15306
15307ssl_fc_alg_keysize : integer
15308 Returns the symmetric cipher key size supported in bits when the incoming
15309 connection was made over an SSL/TLS transport layer.
15310
15311ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015312 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015313 incoming connection made via a TLS transport layer and locally deciphered by
15314 haproxy. The result is a string containing the protocol name advertised by
15315 the client. The SSL library must have been built with support for TLS
15316 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15317 not advertised unless the "alpn" keyword on the "bind" line specifies a
15318 protocol list. Also, nothing forces the client to pick a protocol from this
15319 list, any other one may be requested. The TLS ALPN extension is meant to
15320 replace the TLS NPN extension. See also "ssl_fc_npn".
15321
Willy Tarreau74ca5042013-06-11 23:12:07 +020015322ssl_fc_cipher : string
15323 Returns the name of the used cipher when the incoming connection was made
15324 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015325
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015326ssl_fc_cipherlist_bin : binary
15327 Returns the binary form of the client hello cipher list. The maximum returned
15328 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015329 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015330
15331ssl_fc_cipherlist_hex : string
15332 Returns the binary form of the client hello cipher list encoded as
15333 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015334 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015335
15336ssl_fc_cipherlist_str : string
15337 Returns the decoded text form of the client hello cipher list. The maximum
15338 number of ciphers returned is according with the value of
15339 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015340 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015341 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015342
15343ssl_fc_cipherlist_xxh : integer
15344 Returns a xxh64 of the cipher list. This hash can be return only is the value
15345 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015346 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015347
Patrick Hemmer65674662019-06-04 08:13:03 -040015348ssl_fc_client_random : binary
15349 Returns the client random of the front connection when the incoming connection
15350 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15351 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15352
Willy Tarreau74ca5042013-06-11 23:12:07 +020015353ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015354 Returns true if a client certificate is present in an incoming connection over
15355 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015356 Note: on SSL session resumption with Session ID or TLS ticket, client
15357 certificate is not present in the current connection but may be retrieved
15358 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15359 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015360
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015361ssl_fc_has_early : boolean
15362 Returns true if early data were sent, and the handshake didn't happen yet. As
15363 it has security implications, it is useful to be able to refuse those, or
15364 wait until the handshake happened.
15365
Willy Tarreau74ca5042013-06-11 23:12:07 +020015366ssl_fc_has_sni : boolean
15367 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015368 in an incoming connection was made over an SSL/TLS transport layer. Returns
15369 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050015370 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015371 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015372
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015373ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015374 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015375 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15376 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015377
Willy Tarreau74ca5042013-06-11 23:12:07 +020015378ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015379 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015380 made via a TLS transport layer and locally deciphered by haproxy. The result
15381 is a string containing the protocol name advertised by the client. The SSL
15382 library must have been built with support for TLS extensions enabled (check
15383 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15384 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15385 forces the client to pick a protocol from this list, any other one may be
15386 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015387
Willy Tarreau74ca5042013-06-11 23:12:07 +020015388ssl_fc_protocol : string
15389 Returns the name of the used protocol when the incoming connection was made
15390 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015391
Emeric Brunb73a9b02014-04-30 18:49:19 +020015392ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015393 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015394 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15395 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040015396
Patrick Hemmer65674662019-06-04 08:13:03 -040015397ssl_fc_server_random : binary
15398 Returns the server random of the front connection when the incoming connection
15399 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15400 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15401
Willy Tarreau74ca5042013-06-11 23:12:07 +020015402ssl_fc_session_id : binary
15403 Returns the SSL ID of the front connection when the incoming connection was
15404 made over an SSL/TLS transport layer. It is useful to stick a given client to
15405 a server. It is important to note that some browsers refresh their session ID
15406 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020015407
Patrick Hemmere0275472018-04-28 19:15:51 -040015408ssl_fc_session_key : binary
15409 Returns the SSL session master key of the front connection when the incoming
15410 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15411 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15412 BoringSSL.
15413
15414
Willy Tarreau74ca5042013-06-11 23:12:07 +020015415ssl_fc_sni : string
15416 This extracts the Server Name Indication TLS extension (SNI) field from an
15417 incoming connection made via an SSL/TLS transport layer and locally
15418 deciphered by haproxy. The result (when present) typically is a string
15419 matching the HTTPS host name (253 chars or less). The SSL library must have
15420 been built with support for TLS extensions enabled (check haproxy -vv).
15421
15422 This fetch is different from "req_ssl_sni" above in that it applies to the
15423 connection being deciphered by haproxy and not to SSL contents being blindly
15424 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050015425 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020015426 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020015427
Willy Tarreau74ca5042013-06-11 23:12:07 +020015428 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020015429 ssl_fc_sni_end : suffix match
15430 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020015431
Willy Tarreau74ca5042013-06-11 23:12:07 +020015432ssl_fc_use_keysize : integer
15433 Returns the symmetric cipher key size used in bits when the incoming
15434 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015435
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015436
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200154377.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015438------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015439
Willy Tarreau74ca5042013-06-11 23:12:07 +020015440Fetching samples from buffer contents is a bit different from the previous
15441sample fetches above because the sampled data are ephemeral. These data can
15442only be used when they're available and will be lost when they're forwarded.
15443For this reason, samples fetched from buffer contents during a request cannot
15444be used in a response for example. Even while the data are being fetched, they
15445can change. Sometimes it is necessary to set some delays or combine multiple
15446sample fetch methods to ensure that the expected data are complete and usable,
15447for example through TCP request content inspection. Please see the "tcp-request
15448content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015449
Willy Tarreau74ca5042013-06-11 23:12:07 +020015450payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015451 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015452 "stick on", "stick match"), and for "res.payload" when used in the context of
15453 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015454
Willy Tarreau74ca5042013-06-11 23:12:07 +020015455payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15456 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015457 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015458 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015459
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015460req.hdrs : string
15461 Returns the current request headers as string including the last empty line
15462 separating headers from the request body. The last empty line can be used to
15463 detect a truncated header block. This sample fetch is useful for some SPOE
15464 headers analyzers and for advanced logging.
15465
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015466req.hdrs_bin : binary
15467 Returns the current request headers contained in preparsed binary form. This
15468 is useful for offloading some processing with SPOE. Each string is described
15469 by a length followed by the number of bytes indicated in the length. The
15470 length is represented using the variable integer encoding detailed in the
15471 SPOE documentation. The end of the list is marked by a couple of empty header
15472 names and values (length of 0 for both).
15473
15474 *(<str:header-name><str:header-value>)<empty string><empty string>
15475
15476 int: refer to the SPOE documentation for the encoding
15477 str: <int:length><bytes>
15478
Willy Tarreau74ca5042013-06-11 23:12:07 +020015479req.len : integer
15480req_len : integer (deprecated)
15481 Returns an integer value corresponding to the number of bytes present in the
15482 request buffer. This is mostly used in ACL. It is important to understand
15483 that this test does not return false as long as the buffer is changing. This
15484 means that a check with equality to zero will almost always immediately match
15485 at the beginning of the session, while a test for more data will wait for
15486 that data to come in and return false only when haproxy is certain that no
15487 more data will come in. This test was designed to be used with TCP request
15488 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015489
Willy Tarreau74ca5042013-06-11 23:12:07 +020015490req.payload(<offset>,<length>) : binary
15491 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015492 in the request buffer. As a special case, if the <length> argument is zero,
15493 the the whole buffer from <offset> to the end is extracted. This can be used
15494 with ACLs in order to check for the presence of some content in a buffer at
15495 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015496
Willy Tarreau74ca5042013-06-11 23:12:07 +020015497 ACL alternatives :
15498 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015499
Willy Tarreau74ca5042013-06-11 23:12:07 +020015500req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15501 This extracts a binary block whose size is specified at <offset1> for <length>
15502 bytes, and which starts at <offset2> if specified or just after the length in
15503 the request buffer. The <offset2> parameter also supports relative offsets if
15504 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015505
Willy Tarreau74ca5042013-06-11 23:12:07 +020015506 ACL alternatives :
15507 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015508
Willy Tarreau74ca5042013-06-11 23:12:07 +020015509 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015510
Willy Tarreau74ca5042013-06-11 23:12:07 +020015511req.proto_http : boolean
15512req_proto_http : boolean (deprecated)
15513 Returns true when data in the request buffer look like HTTP and correctly
15514 parses as such. It is the same parser as the common HTTP request parser which
15515 is used so there should be no surprises. The test does not match until the
15516 request is complete, failed or timed out. This test may be used to report the
15517 protocol in TCP logs, but the biggest use is to block TCP request analysis
15518 until a complete HTTP request is present in the buffer, for example to track
15519 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015520
Willy Tarreau74ca5042013-06-11 23:12:07 +020015521 Example:
15522 # track request counts per "base" (concatenation of Host+URL)
15523 tcp-request inspect-delay 10s
15524 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015525 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015526
Willy Tarreau74ca5042013-06-11 23:12:07 +020015527req.rdp_cookie([<name>]) : string
15528rdp_cookie([<name>]) : string (deprecated)
15529 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15530 <name>, or any cookie if unspecified. The parser only checks for the first
15531 cookie, as illustrated in the RDP protocol specification. The cookie name is
15532 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15533 contain the user name of the client connecting to the server if properly
15534 configured on the client. The "MSTSHASH" cookie is often used as well for
15535 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015536
Willy Tarreau74ca5042013-06-11 23:12:07 +020015537 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15538 used and thus the distribution of clients to backend servers is not linked to
15539 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15540 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15541 distribution of clients to backend servers than the hash used by "balance
15542 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015543
Willy Tarreau74ca5042013-06-11 23:12:07 +020015544 ACL derivatives :
15545 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015546
Willy Tarreau74ca5042013-06-11 23:12:07 +020015547 Example :
15548 listen tse-farm
15549 bind 0.0.0.0:3389
15550 # wait up to 5s for an RDP cookie in the request
15551 tcp-request inspect-delay 5s
15552 tcp-request content accept if RDP_COOKIE
15553 # apply RDP cookie persistence
15554 persist rdp-cookie
15555 # Persist based on the mstshash cookie
15556 # This is only useful makes sense if
15557 # balance rdp-cookie is not used
15558 stick-table type string size 204800
15559 stick on req.rdp_cookie(mstshash)
15560 server srv1 1.1.1.1:3389
15561 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015562
Willy Tarreau74ca5042013-06-11 23:12:07 +020015563 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15564 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015565
Willy Tarreau74ca5042013-06-11 23:12:07 +020015566req.rdp_cookie_cnt([name]) : integer
15567rdp_cookie_cnt([name]) : integer (deprecated)
15568 Tries to parse the request buffer as RDP protocol, then returns an integer
15569 corresponding to the number of RDP cookies found. If an optional cookie name
15570 is passed, only cookies matching this name are considered. This is mostly
15571 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015572
Willy Tarreau74ca5042013-06-11 23:12:07 +020015573 ACL derivatives :
15574 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015575
Alex Zorin4afdd132018-12-30 13:56:28 +110015576req.ssl_alpn : string
15577 Returns a string containing the values of the Application-Layer Protocol
15578 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
15579 ClientHello message. Note that this only applies to raw contents found in the
15580 request buffer and not to the contents deciphered via an SSL data layer, so
15581 this will not work with "bind" lines having the "ssl" option. This is useful
15582 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020015583 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110015584
15585 Examples :
15586 # Wait for a client hello for at most 5 seconds
15587 tcp-request inspect-delay 5s
15588 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020015589 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110015590 default_backend bk_default
15591
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015592req.ssl_ec_ext : boolean
15593 Returns a boolean identifying if client sent the Supported Elliptic Curves
15594 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015595 message. This can be used to present ECC compatible clients with EC
15596 certificate and to use RSA for all others, on the same IP address. Note that
15597 this only applies to raw contents found in the request buffer and not to
15598 contents deciphered via an SSL data layer, so this will not work with "bind"
15599 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015600
Willy Tarreau74ca5042013-06-11 23:12:07 +020015601req.ssl_hello_type : integer
15602req_ssl_hello_type : integer (deprecated)
15603 Returns an integer value containing the type of the SSL hello message found
15604 in the request buffer if the buffer contains data that parse as a complete
15605 SSL (v3 or superior) client hello message. Note that this only applies to raw
15606 contents found in the request buffer and not to contents deciphered via an
15607 SSL data layer, so this will not work with "bind" lines having the "ssl"
15608 option. This is mostly used in ACL to detect presence of an SSL hello message
15609 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015610
Willy Tarreau74ca5042013-06-11 23:12:07 +020015611req.ssl_sni : string
15612req_ssl_sni : string (deprecated)
15613 Returns a string containing the value of the Server Name TLS extension sent
15614 by a client in a TLS stream passing through the request buffer if the buffer
15615 contains data that parse as a complete SSL (v3 or superior) client hello
15616 message. Note that this only applies to raw contents found in the request
15617 buffer and not to contents deciphered via an SSL data layer, so this will not
15618 work with "bind" lines having the "ssl" option. SNI normally contains the
15619 name of the host the client tries to connect to (for recent browsers). SNI is
15620 useful for allowing or denying access to certain hosts when SSL/TLS is used
15621 by the client. This test was designed to be used with TCP request content
15622 inspection. If content switching is needed, it is recommended to first wait
15623 for a complete client hello (type 1), like in the example below. See also
15624 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015625
Willy Tarreau74ca5042013-06-11 23:12:07 +020015626 ACL derivatives :
15627 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015628
Willy Tarreau74ca5042013-06-11 23:12:07 +020015629 Examples :
15630 # Wait for a client hello for at most 5 seconds
15631 tcp-request inspect-delay 5s
15632 tcp-request content accept if { req_ssl_hello_type 1 }
15633 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15634 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015635
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015636req.ssl_st_ext : integer
15637 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15638 Returns 1 if the client sent SessionTicket TLS Extension
15639 Returns 2 if the client also sent non-zero length TLS SessionTicket
15640 Note that this only applies to raw contents found in the request buffer and
15641 not to contents deciphered via an SSL data layer, so this will not work with
15642 "bind" lines having the "ssl" option. This can for example be used to detect
15643 whether the client sent a SessionTicket or not and stick it accordingly, if
15644 no SessionTicket then stick on SessionID or don't stick as there's no server
15645 side state is there when SessionTickets are in use.
15646
Willy Tarreau74ca5042013-06-11 23:12:07 +020015647req.ssl_ver : integer
15648req_ssl_ver : integer (deprecated)
15649 Returns an integer value containing the version of the SSL/TLS protocol of a
15650 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15651 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15652 composed of the major version multiplied by 65536, added to the minor
15653 version. Note that this only applies to raw contents found in the request
15654 buffer and not to contents deciphered via an SSL data layer, so this will not
15655 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015656 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015657 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015658
Willy Tarreau74ca5042013-06-11 23:12:07 +020015659 ACL derivatives :
15660 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015661
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015662res.len : integer
15663 Returns an integer value corresponding to the number of bytes present in the
15664 response buffer. This is mostly used in ACL. It is important to understand
15665 that this test does not return false as long as the buffer is changing. This
15666 means that a check with equality to zero will almost always immediately match
15667 at the beginning of the session, while a test for more data will wait for
15668 that data to come in and return false only when haproxy is certain that no
15669 more data will come in. This test was designed to be used with TCP response
15670 content inspection.
15671
Willy Tarreau74ca5042013-06-11 23:12:07 +020015672res.payload(<offset>,<length>) : binary
15673 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015674 in the response buffer. As a special case, if the <length> argument is zero,
15675 the the whole buffer from <offset> to the end is extracted. This can be used
15676 with ACLs in order to check for the presence of some content in a buffer at
15677 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015678
Willy Tarreau74ca5042013-06-11 23:12:07 +020015679res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15680 This extracts a binary block whose size is specified at <offset1> for <length>
15681 bytes, and which starts at <offset2> if specified or just after the length in
15682 the response buffer. The <offset2> parameter also supports relative offsets
15683 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015684
Willy Tarreau74ca5042013-06-11 23:12:07 +020015685 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015686
Willy Tarreau971f7b62015-09-29 14:06:59 +020015687res.ssl_hello_type : integer
15688rep_ssl_hello_type : integer (deprecated)
15689 Returns an integer value containing the type of the SSL hello message found
15690 in the response buffer if the buffer contains data that parses as a complete
15691 SSL (v3 or superior) hello message. Note that this only applies to raw
15692 contents found in the response buffer and not to contents deciphered via an
15693 SSL data layer, so this will not work with "server" lines having the "ssl"
15694 option. This is mostly used in ACL to detect presence of an SSL hello message
15695 that is supposed to contain an SSL session ID usable for stickiness.
15696
Willy Tarreau74ca5042013-06-11 23:12:07 +020015697wait_end : boolean
15698 This fetch either returns true when the inspection period is over, or does
15699 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015700 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015701 actions, such as a delayed reject for some special addresses. Since it either
15702 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015703 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015704 "WAIT_END" is always usable without prior declaration. This test was designed
15705 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015706
Willy Tarreau74ca5042013-06-11 23:12:07 +020015707 Examples :
15708 # delay every incoming request by 2 seconds
15709 tcp-request inspect-delay 2s
15710 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015711
Willy Tarreau74ca5042013-06-11 23:12:07 +020015712 # don't immediately tell bad guys they are rejected
15713 tcp-request inspect-delay 10s
15714 acl goodguys src 10.0.0.0/24
15715 acl badguys src 10.0.1.0/24
15716 tcp-request content accept if goodguys
15717 tcp-request content reject if badguys WAIT_END
15718 tcp-request content reject
15719
15720
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200157217.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015722--------------------------------------
15723
15724It is possible to fetch samples from HTTP contents, requests and responses.
15725This application layer is also called layer 7. It is only possible to fetch the
15726data in this section when a full HTTP request or response has been parsed from
15727its respective request or response buffer. This is always the case with all
15728HTTP specific rules and for sections running with "mode http". When using TCP
15729content inspection, it may be necessary to support an inspection delay in order
15730to let the request or response come in first. These fetches may require a bit
15731more CPU resources than the layer 4 ones, but not much since the request and
15732response are indexed.
15733
15734base : string
15735 This returns the concatenation of the first Host header and the path part of
15736 the request, which starts at the first slash and ends before the question
15737 mark. It can be useful in virtual hosted environments to detect URL abuses as
15738 well as to improve shared caches efficiency. Using this with a limited size
15739 stick table also allows one to collect statistics about most commonly
15740 requested objects by host/path. With ACLs it can allow simple content
15741 switching rules involving the host and the path at the same time, such as
15742 "www.example.com/favicon.ico". See also "path" and "uri".
15743
15744 ACL derivatives :
15745 base : exact string match
15746 base_beg : prefix match
15747 base_dir : subdir match
15748 base_dom : domain match
15749 base_end : suffix match
15750 base_len : length match
15751 base_reg : regex match
15752 base_sub : substring match
15753
15754base32 : integer
15755 This returns a 32-bit hash of the value returned by the "base" fetch method
15756 above. This is useful to track per-URL activity on high traffic sites without
15757 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015758 memory. The output type is an unsigned integer. The hash function used is
15759 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15760 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015761
15762base32+src : binary
15763 This returns the concatenation of the base32 fetch above and the src fetch
15764 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15765 depending on the source address family. This can be used to track per-IP,
15766 per-URL counters.
15767
William Lallemand65ad6e12014-01-31 15:08:02 +010015768capture.req.hdr(<idx>) : string
15769 This extracts the content of the header captured by the "capture request
15770 header", idx is the position of the capture keyword in the configuration.
15771 The first entry is an index of 0. See also: "capture request header".
15772
15773capture.req.method : string
15774 This extracts the METHOD of an HTTP request. It can be used in both request
15775 and response. Unlike "method", it can be used in both request and response
15776 because it's allocated.
15777
15778capture.req.uri : string
15779 This extracts the request's URI, which starts at the first slash and ends
15780 before the first space in the request (without the host part). Unlike "path"
15781 and "url", it can be used in both request and response because it's
15782 allocated.
15783
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015784capture.req.ver : string
15785 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15786 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15787 logs because it relies on a persistent flag.
15788
William Lallemand65ad6e12014-01-31 15:08:02 +010015789capture.res.hdr(<idx>) : string
15790 This extracts the content of the header captured by the "capture response
15791 header", idx is the position of the capture keyword in the configuration.
15792 The first entry is an index of 0.
15793 See also: "capture response header"
15794
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015795capture.res.ver : string
15796 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15797 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15798 persistent flag.
15799
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015800req.body : binary
15801 This returns the HTTP request's available body as a block of data. It
15802 requires that the request body has been buffered made available using
15803 "option http-buffer-request". In case of chunked-encoded body, currently only
15804 the first chunk is analyzed.
15805
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015806req.body_param([<name>) : string
15807 This fetch assumes that the body of the POST request is url-encoded. The user
15808 can check if the "content-type" contains the value
15809 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15810 parameter <name> in the body, which ends before '&'. The parameter name is
15811 case-sensitive. If no name is given, any parameter will match, and the first
15812 one will be returned. The result is a string corresponding to the value of the
15813 parameter <name> as presented in the request body (no URL decoding is
15814 performed). Note that the ACL version of this fetch iterates over multiple
15815 parameters and will iteratively report all parameters values if no name is
15816 given.
15817
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015818req.body_len : integer
15819 This returns the length of the HTTP request's available body in bytes. It may
15820 be lower than the advertised length if the body is larger than the buffer. It
15821 requires that the request body has been buffered made available using
15822 "option http-buffer-request".
15823
15824req.body_size : integer
15825 This returns the advertised length of the HTTP request's body in bytes. It
15826 will represent the advertised Content-Length header, or the size of the first
15827 chunk in case of chunked encoding. In order to parse the chunks, it requires
15828 that the request body has been buffered made available using
15829 "option http-buffer-request".
15830
Willy Tarreau74ca5042013-06-11 23:12:07 +020015831req.cook([<name>]) : string
15832cook([<name>]) : string (deprecated)
15833 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15834 header line from the request, and returns its value as string. If no name is
15835 specified, the first cookie value is returned. When used with ACLs, all
15836 matching cookies are evaluated. Spaces around the name and the value are
15837 ignored as requested by the Cookie header specification (RFC6265). The cookie
15838 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15839 well return an empty value if it is present. Use the "found" match to detect
15840 presence. Use the res.cook() variant for response cookies sent by the server.
15841
15842 ACL derivatives :
15843 cook([<name>]) : exact string match
15844 cook_beg([<name>]) : prefix match
15845 cook_dir([<name>]) : subdir match
15846 cook_dom([<name>]) : domain match
15847 cook_end([<name>]) : suffix match
15848 cook_len([<name>]) : length match
15849 cook_reg([<name>]) : regex match
15850 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015851
Willy Tarreau74ca5042013-06-11 23:12:07 +020015852req.cook_cnt([<name>]) : integer
15853cook_cnt([<name>]) : integer (deprecated)
15854 Returns an integer value representing the number of occurrences of the cookie
15855 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015856
Willy Tarreau74ca5042013-06-11 23:12:07 +020015857req.cook_val([<name>]) : integer
15858cook_val([<name>]) : integer (deprecated)
15859 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15860 header line from the request, and converts its value to an integer which is
15861 returned. If no name is specified, the first cookie value is returned. When
15862 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015863
Willy Tarreau74ca5042013-06-11 23:12:07 +020015864cookie([<name>]) : string (deprecated)
15865 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15866 header line from the request, or a "Set-Cookie" header from the response, and
15867 returns its value as a string. A typical use is to get multiple clients
15868 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015869 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015870 multi-peer synchronization and state keeping across restarts. If no name is
15871 specified, the first cookie value is returned. This fetch should not be used
15872 anymore and should be replaced by req.cook() or res.cook() instead as it
15873 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015874
Willy Tarreau74ca5042013-06-11 23:12:07 +020015875hdr([<name>[,<occ>]]) : string
15876 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15877 used on responses. Please refer to these respective fetches for more details.
15878 In case of doubt about the fetch direction, please use the explicit ones.
15879 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015880 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015881
Willy Tarreau74ca5042013-06-11 23:12:07 +020015882req.fhdr(<name>[,<occ>]) : string
15883 This extracts the last occurrence of header <name> in an HTTP request. When
15884 used from an ACL, all occurrences are iterated over until a match is found.
15885 Optionally, a specific occurrence might be specified as a position number.
15886 Positive values indicate a position from the first occurrence, with 1 being
15887 the first one. Negative values indicate positions relative to the last one,
15888 with -1 being the last one. It differs from req.hdr() in that any commas
15889 present in the value are returned and are not used as delimiters. This is
15890 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015891
Willy Tarreau74ca5042013-06-11 23:12:07 +020015892req.fhdr_cnt([<name>]) : integer
15893 Returns an integer value representing the number of occurrences of request
15894 header field name <name>, or the total number of header fields if <name> is
15895 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15896 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015897
Willy Tarreau74ca5042013-06-11 23:12:07 +020015898req.hdr([<name>[,<occ>]]) : string
15899 This extracts the last occurrence of header <name> in an HTTP request. When
15900 used from an ACL, all occurrences are iterated over until a match is found.
15901 Optionally, a specific occurrence might be specified as a position number.
15902 Positive values indicate a position from the first occurrence, with 1 being
15903 the first one. Negative values indicate positions relative to the last one,
15904 with -1 being the last one. A typical use is with the X-Forwarded-For header
15905 once converted to IP, associated with an IP stick-table. The function
15906 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015907 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015908 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015909 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015910
Willy Tarreau74ca5042013-06-11 23:12:07 +020015911 ACL derivatives :
15912 hdr([<name>[,<occ>]]) : exact string match
15913 hdr_beg([<name>[,<occ>]]) : prefix match
15914 hdr_dir([<name>[,<occ>]]) : subdir match
15915 hdr_dom([<name>[,<occ>]]) : domain match
15916 hdr_end([<name>[,<occ>]]) : suffix match
15917 hdr_len([<name>[,<occ>]]) : length match
15918 hdr_reg([<name>[,<occ>]]) : regex match
15919 hdr_sub([<name>[,<occ>]]) : substring match
15920
15921req.hdr_cnt([<name>]) : integer
15922hdr_cnt([<header>]) : integer (deprecated)
15923 Returns an integer value representing the number of occurrences of request
15924 header field name <name>, or the total number of header field values if
15925 <name> is not specified. It is important to remember that one header line may
15926 count as several headers if it has several values. The function considers any
15927 comma as a delimiter for distinct values. If full-line headers are desired
15928 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15929 detect presence, absence or abuse of a specific header, as well as to block
15930 request smuggling attacks by rejecting requests which contain more than one
15931 of certain headers. See "req.hdr" for more information on header matching.
15932
15933req.hdr_ip([<name>[,<occ>]]) : ip
15934hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15935 This extracts the last occurrence of header <name> in an HTTP request,
15936 converts it to an IPv4 or IPv6 address and returns this address. When used
15937 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15938 of every header is checked. Optionally, a specific occurrence might be
15939 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015940 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015941 positions relative to the last one, with -1 being the last one. A typical use
15942 is with the X-Forwarded-For and X-Client-IP headers.
15943
15944req.hdr_val([<name>[,<occ>]]) : integer
15945hdr_val([<name>[,<occ>]]) : integer (deprecated)
15946 This extracts the last occurrence of header <name> in an HTTP request, and
15947 converts it to an integer value. When used with ACLs, all occurrences are
15948 checked, and if <name> is omitted, every value of every header is checked.
15949 Optionally, a specific occurrence might be specified as a position number.
15950 Positive values indicate a position from the first occurrence, with 1 being
15951 the first one. Negative values indicate positions relative to the last one,
15952 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15953
Frédéric Lécailleec891192019-02-26 15:02:35 +010015954
15955
Willy Tarreau74ca5042013-06-11 23:12:07 +020015956http_auth(<userlist>) : boolean
15957 Returns a boolean indicating whether the authentication data received from
15958 the client match a username & password stored in the specified userlist. This
15959 fetch function is not really useful outside of ACLs. Currently only http
15960 basic auth is supported.
15961
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015962http_auth_group(<userlist>) : string
15963 Returns a string corresponding to the user name found in the authentication
15964 data received from the client if both the user name and password are valid
15965 according to the specified userlist. The main purpose is to use it in ACLs
15966 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015967 This fetch function is not really useful outside of ACLs. Currently only http
15968 basic auth is supported.
15969
15970 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015971 http_auth_group(<userlist>) : group ...
15972 Returns true when the user extracted from the request and whose password is
15973 valid according to the specified userlist belongs to at least one of the
15974 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015975
Christopher Fauleta4063562019-08-02 11:51:37 +020015976http_auth_pass : string
15977 Returns the user's password found in the authentication data received from
15978 the client, as supplied in the Authorization header. Not checks are
15979 performed by this sample fetch. Only Basic authentication is supported.
15980
15981http_auth_type : string
15982 Returns the authentication method found in the authentication data received from
15983 the client, as supplied in the Authorization header. Not checks are
15984 performed by this sample fetch. Only Basic authentication is supported.
15985
15986http_auth_user : string
15987 Returns the user name found in the authentication data received from the
15988 client, as supplied in the Authorization header. Not checks are performed by
15989 this sample fetch. Only Basic authentication is supported.
15990
Willy Tarreau74ca5042013-06-11 23:12:07 +020015991http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015992 Returns true when the request being processed is the first one of the
15993 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015994 from some requests when a request is not the first one, or to help grouping
15995 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015996
Willy Tarreau74ca5042013-06-11 23:12:07 +020015997method : integer + string
15998 Returns an integer value corresponding to the method in the HTTP request. For
15999 example, "GET" equals 1 (check sources to establish the matching). Value 9
16000 means "other method" and may be converted to a string extracted from the
16001 stream. This should not be used directly as a sample, this is only meant to
16002 be used from ACLs, which transparently convert methods from patterns to these
16003 integer + string values. Some predefined ACL already check for most common
16004 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016005
Willy Tarreau74ca5042013-06-11 23:12:07 +020016006 ACL derivatives :
16007 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016008
Willy Tarreau74ca5042013-06-11 23:12:07 +020016009 Example :
16010 # only accept GET and HEAD requests
16011 acl valid_method method GET HEAD
16012 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020016013
Willy Tarreau74ca5042013-06-11 23:12:07 +020016014path : string
16015 This extracts the request's URL path, which starts at the first slash and
16016 ends before the question mark (without the host part). A typical use is with
16017 prefetch-capable caches, and with portals which need to aggregate multiple
16018 information from databases and keep them in caches. Note that with outgoing
16019 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010016020 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020016021 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016022
Willy Tarreau74ca5042013-06-11 23:12:07 +020016023 ACL derivatives :
16024 path : exact string match
16025 path_beg : prefix match
16026 path_dir : subdir match
16027 path_dom : domain match
16028 path_end : suffix match
16029 path_len : length match
16030 path_reg : regex match
16031 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016032
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016033query : string
16034 This extracts the request's query string, which starts after the first
16035 question mark. If no question mark is present, this fetch returns nothing. If
16036 a question mark is present but nothing follows, it returns an empty string.
16037 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010016038 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016039 which stops before the question mark.
16040
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016041req.hdr_names([<delim>]) : string
16042 This builds a string made from the concatenation of all header names as they
16043 appear in the request when the rule is evaluated. The default delimiter is
16044 the comma (',') but it may be overridden as an optional argument <delim>. In
16045 this case, only the first character of <delim> is considered.
16046
Willy Tarreau74ca5042013-06-11 23:12:07 +020016047req.ver : string
16048req_ver : string (deprecated)
16049 Returns the version string from the HTTP request, for example "1.1". This can
16050 be useful for logs, but is mostly there for ACL. Some predefined ACL already
16051 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016052
Willy Tarreau74ca5042013-06-11 23:12:07 +020016053 ACL derivatives :
16054 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020016055
Willy Tarreau74ca5042013-06-11 23:12:07 +020016056res.comp : boolean
16057 Returns the boolean "true" value if the response has been compressed by
16058 HAProxy, otherwise returns boolean "false". This may be used to add
16059 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016060
Willy Tarreau74ca5042013-06-11 23:12:07 +020016061res.comp_algo : string
16062 Returns a string containing the name of the algorithm used if the response
16063 was compressed by HAProxy, for example : "deflate". This may be used to add
16064 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016065
Willy Tarreau74ca5042013-06-11 23:12:07 +020016066res.cook([<name>]) : string
16067scook([<name>]) : string (deprecated)
16068 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16069 header line from the response, and returns its value as string. If no name is
16070 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016071
Willy Tarreau74ca5042013-06-11 23:12:07 +020016072 ACL derivatives :
16073 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016074
Willy Tarreau74ca5042013-06-11 23:12:07 +020016075res.cook_cnt([<name>]) : integer
16076scook_cnt([<name>]) : integer (deprecated)
16077 Returns an integer value representing the number of occurrences of the cookie
16078 <name> in the response, or all cookies if <name> is not specified. This is
16079 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016080
Willy Tarreau74ca5042013-06-11 23:12:07 +020016081res.cook_val([<name>]) : integer
16082scook_val([<name>]) : integer (deprecated)
16083 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16084 header line from the response, and converts its value to an integer which is
16085 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016086
Willy Tarreau74ca5042013-06-11 23:12:07 +020016087res.fhdr([<name>[,<occ>]]) : string
16088 This extracts the last occurrence of header <name> in an HTTP response, or of
16089 the last header if no <name> is specified. Optionally, a specific occurrence
16090 might be specified as a position number. Positive values indicate a position
16091 from the first occurrence, with 1 being the first one. Negative values
16092 indicate positions relative to the last one, with -1 being the last one. It
16093 differs from res.hdr() in that any commas present in the value are returned
16094 and are not used as delimiters. If this is not desired, the res.hdr() fetch
16095 should be used instead. This is sometimes useful with headers such as Date or
16096 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016097
Willy Tarreau74ca5042013-06-11 23:12:07 +020016098res.fhdr_cnt([<name>]) : integer
16099 Returns an integer value representing the number of occurrences of response
16100 header field name <name>, or the total number of header fields if <name> is
16101 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
16102 the number of full line headers and does not stop on commas. If this is not
16103 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016104
Willy Tarreau74ca5042013-06-11 23:12:07 +020016105res.hdr([<name>[,<occ>]]) : string
16106shdr([<name>[,<occ>]]) : string (deprecated)
16107 This extracts the last occurrence of header <name> in an HTTP response, or of
16108 the last header if no <name> is specified. Optionally, a specific occurrence
16109 might be specified as a position number. Positive values indicate a position
16110 from the first occurrence, with 1 being the first one. Negative values
16111 indicate positions relative to the last one, with -1 being the last one. This
16112 can be useful to learn some data into a stick-table. The function considers
16113 any comma as a delimiter for distinct values. If this is not desired, the
16114 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016115
Willy Tarreau74ca5042013-06-11 23:12:07 +020016116 ACL derivatives :
16117 shdr([<name>[,<occ>]]) : exact string match
16118 shdr_beg([<name>[,<occ>]]) : prefix match
16119 shdr_dir([<name>[,<occ>]]) : subdir match
16120 shdr_dom([<name>[,<occ>]]) : domain match
16121 shdr_end([<name>[,<occ>]]) : suffix match
16122 shdr_len([<name>[,<occ>]]) : length match
16123 shdr_reg([<name>[,<occ>]]) : regex match
16124 shdr_sub([<name>[,<occ>]]) : substring match
16125
16126res.hdr_cnt([<name>]) : integer
16127shdr_cnt([<name>]) : integer (deprecated)
16128 Returns an integer value representing the number of occurrences of response
16129 header field name <name>, or the total number of header fields if <name> is
16130 not specified. The function considers any comma as a delimiter for distinct
16131 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16132 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016133
Willy Tarreau74ca5042013-06-11 23:12:07 +020016134res.hdr_ip([<name>[,<occ>]]) : ip
16135shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16136 This extracts the last occurrence of header <name> in an HTTP response,
16137 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16138 specific occurrence might be specified as a position number. Positive values
16139 indicate a position from the first occurrence, with 1 being the first one.
16140 Negative values indicate positions relative to the last one, with -1 being
16141 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016142
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016143res.hdr_names([<delim>]) : string
16144 This builds a string made from the concatenation of all header names as they
16145 appear in the response when the rule is evaluated. The default delimiter is
16146 the comma (',') but it may be overridden as an optional argument <delim>. In
16147 this case, only the first character of <delim> is considered.
16148
Willy Tarreau74ca5042013-06-11 23:12:07 +020016149res.hdr_val([<name>[,<occ>]]) : integer
16150shdr_val([<name>[,<occ>]]) : integer (deprecated)
16151 This extracts the last occurrence of header <name> in an HTTP response, and
16152 converts it to an integer value. Optionally, a specific occurrence might be
16153 specified as a position number. Positive values indicate a position from the
16154 first occurrence, with 1 being the first one. Negative values indicate
16155 positions relative to the last one, with -1 being the last one. This can be
16156 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016157
Willy Tarreau74ca5042013-06-11 23:12:07 +020016158res.ver : string
16159resp_ver : string (deprecated)
16160 Returns the version string from the HTTP response, for example "1.1". This
16161 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016162
Willy Tarreau74ca5042013-06-11 23:12:07 +020016163 ACL derivatives :
16164 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016165
Willy Tarreau74ca5042013-06-11 23:12:07 +020016166set-cookie([<name>]) : string (deprecated)
16167 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16168 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016169 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016170 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016171
Willy Tarreau74ca5042013-06-11 23:12:07 +020016172 This fetch function is deprecated and has been superseded by the "res.cook"
16173 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016174
Willy Tarreau74ca5042013-06-11 23:12:07 +020016175status : integer
16176 Returns an integer containing the HTTP status code in the HTTP response, for
16177 example, 302. It is mostly used within ACLs and integer ranges, for example,
16178 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016179
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016180unique-id : string
16181 Returns the unique-id attached to the request. The directive
16182 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16183 fails. Note that the unique-id is usually used with HTTP requests, however this
16184 sample fetch can be used with other protocols. Obviously, if it is used with
16185 other protocols than HTTP, the unique-id-format directive must not contain
16186 HTTP parts. See: unique-id-format and unique-id-header
16187
Willy Tarreau74ca5042013-06-11 23:12:07 +020016188url : string
16189 This extracts the request's URL as presented in the request. A typical use is
16190 with prefetch-capable caches, and with portals which need to aggregate
16191 multiple information from databases and keep them in caches. With ACLs, using
16192 "path" is preferred over using "url", because clients may send a full URL as
16193 is normally done with proxies. The only real use is to match "*" which does
16194 not match in "path", and for which there is already a predefined ACL. See
16195 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016196
Willy Tarreau74ca5042013-06-11 23:12:07 +020016197 ACL derivatives :
16198 url : exact string match
16199 url_beg : prefix match
16200 url_dir : subdir match
16201 url_dom : domain match
16202 url_end : suffix match
16203 url_len : length match
16204 url_reg : regex match
16205 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016206
Willy Tarreau74ca5042013-06-11 23:12:07 +020016207url_ip : ip
16208 This extracts the IP address from the request's URL when the host part is
16209 presented as an IP address. Its use is very limited. For instance, a
16210 monitoring system might use this field as an alternative for the source IP in
16211 order to test what path a given source address would follow, or to force an
16212 entry in a table for a given source address. With ACLs it can be used to
16213 restrict access to certain systems through a proxy, for example when combined
16214 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016215
Willy Tarreau74ca5042013-06-11 23:12:07 +020016216url_port : integer
16217 This extracts the port part from the request's URL. Note that if the port is
16218 not specified in the request, port 80 is assumed. With ACLs it can be used to
16219 restrict access to certain systems through a proxy, for example when combined
16220 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016221
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016222urlp([<name>[,<delim>]]) : string
16223url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016224 This extracts the first occurrence of the parameter <name> in the query
16225 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016226 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16227 any parameter will match, and the first one will be returned. The result is
16228 a string corresponding to the value of the parameter <name> as presented in
16229 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016230 stickiness based on a client ID, to extract an application cookie passed as a
16231 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016232 this fetch iterates over multiple parameters and will iteratively report all
16233 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016234
Willy Tarreau74ca5042013-06-11 23:12:07 +020016235 ACL derivatives :
16236 urlp(<name>[,<delim>]) : exact string match
16237 urlp_beg(<name>[,<delim>]) : prefix match
16238 urlp_dir(<name>[,<delim>]) : subdir match
16239 urlp_dom(<name>[,<delim>]) : domain match
16240 urlp_end(<name>[,<delim>]) : suffix match
16241 urlp_len(<name>[,<delim>]) : length match
16242 urlp_reg(<name>[,<delim>]) : regex match
16243 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016244
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016245
Willy Tarreau74ca5042013-06-11 23:12:07 +020016246 Example :
16247 # match http://example.com/foo?PHPSESSIONID=some_id
16248 stick on urlp(PHPSESSIONID)
16249 # match http://example.com/foo;JSESSIONID=some_id
16250 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016251
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016252urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016253 See "urlp" above. This one extracts the URL parameter <name> in the request
16254 and converts it to an integer value. This can be used for session stickiness
16255 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016256
Dragan Dosen0070cd52016-06-16 12:19:49 +020016257url32 : integer
16258 This returns a 32-bit hash of the value obtained by concatenating the first
16259 Host header and the whole URL including parameters (not only the path part of
16260 the request, as in the "base32" fetch above). This is useful to track per-URL
16261 activity. A shorter hash is stored, saving a lot of memory. The output type
16262 is an unsigned integer.
16263
16264url32+src : binary
16265 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16266 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16267 the source address family. This can be used to track per-IP, per-URL counters.
16268
Willy Tarreau198a7442008-01-17 12:05:32 +010016269
Willy Tarreau74ca5042013-06-11 23:12:07 +0200162707.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016271---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016272
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016273Some predefined ACLs are hard-coded so that they do not have to be declared in
16274every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016275order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016276
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016277ACL name Equivalent to Usage
16278---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016279FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016280HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016281HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16282HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016283HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16284HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16285HTTP_URL_SLASH url_beg / match URL beginning with "/"
16286HTTP_URL_STAR url * match URL equal to "*"
16287LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016288METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016289METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016290METH_GET method GET HEAD match HTTP GET or HEAD method
16291METH_HEAD method HEAD match HTTP HEAD method
16292METH_OPTIONS method OPTIONS match HTTP OPTIONS method
16293METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016294METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016295METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020016296RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016297REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016298TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016299WAIT_END wait_end wait for end of content analysis
16300---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016301
Willy Tarreaub937b7e2010-01-12 15:27:54 +010016302
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163038. Logging
16304----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010016305
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016306One of HAProxy's strong points certainly lies is its precise logs. It probably
16307provides the finest level of information available for such a product, which is
16308very important for troubleshooting complex environments. Standard information
16309provided in logs include client ports, TCP/HTTP state timers, precise session
16310state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016311to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016312headers.
16313
16314In order to improve administrators reactivity, it offers a great transparency
16315about encountered problems, both internal and external, and it is possible to
16316send logs to different sources at the same time with different level filters :
16317
16318 - global process-level logs (system errors, start/stop, etc..)
16319 - per-instance system and internal errors (lack of resource, bugs, ...)
16320 - per-instance external troubles (servers up/down, max connections)
16321 - per-instance activity (client connections), either at the establishment or
16322 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016323 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016324 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016325
16326The ability to distribute different levels of logs to different log servers
16327allow several production teams to interact and to fix their problems as soon
16328as possible. For example, the system team might monitor system-wide errors,
16329while the application team might be monitoring the up/down for their servers in
16330real time, and the security team might analyze the activity logs with one hour
16331delay.
16332
16333
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163348.1. Log levels
16335---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016336
Simon Hormandf791f52011-05-29 15:01:10 +090016337TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016338source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016339HTTP request, HTTP return code, number of bytes transmitted, conditions
16340in which the session ended, and even exchanged cookies values. For example
16341track a particular user's problems. All messages may be sent to up to two
16342syslog servers. Check the "log" keyword in section 4.2 for more information
16343about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016344
16345
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163468.2. Log formats
16347----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016348
William Lallemand48940402012-01-30 16:47:22 +010016349HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016350and will be detailed in the following sections. A few of them may vary
16351slightly with the configuration, due to indicators specific to certain
16352options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016353
16354 - the default format, which is very basic and very rarely used. It only
16355 provides very basic information about the incoming connection at the moment
16356 it is accepted : source IP:port, destination IP:port, and frontend-name.
16357 This mode will eventually disappear so it will not be described to great
16358 extents.
16359
16360 - the TCP format, which is more advanced. This format is enabled when "option
16361 tcplog" is set on the frontend. HAProxy will then usually wait for the
16362 connection to terminate before logging. This format provides much richer
16363 information, such as timers, connection counts, queue size, etc... This
16364 format is recommended for pure TCP proxies.
16365
16366 - the HTTP format, which is the most advanced for HTTP proxying. This format
16367 is enabled when "option httplog" is set on the frontend. It provides the
16368 same information as the TCP format with some HTTP-specific fields such as
16369 the request, the status code, and captures of headers and cookies. This
16370 format is recommended for HTTP proxies.
16371
Emeric Brun3a058f32009-06-30 18:26:00 +020016372 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16373 fields arranged in the same order as the CLF format. In this mode, all
16374 timers, captures, flags, etc... appear one per field after the end of the
16375 common fields, in the same order they appear in the standard HTTP format.
16376
William Lallemand48940402012-01-30 16:47:22 +010016377 - the custom log format, allows you to make your own log line.
16378
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016379Next sections will go deeper into details for each of these formats. Format
16380specification will be performed on a "field" basis. Unless stated otherwise, a
16381field is a portion of text delimited by any number of spaces. Since syslog
16382servers are susceptible of inserting fields at the beginning of a line, it is
16383always assumed that the first field is the one containing the process name and
16384identifier.
16385
16386Note : Since log lines may be quite long, the log examples in sections below
16387 might be broken into multiple lines. The example log lines will be
16388 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16389 broken into multiple lines, each non-final line will end with a
16390 backslash ('\') and the next line will start indented by two characters.
16391
16392
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163938.2.1. Default log format
16394-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016395
16396This format is used when no specific option is set. The log is emitted as soon
16397as the connection is accepted. One should note that this currently is the only
16398format which logs the request's destination IP and ports.
16399
16400 Example :
16401 listen www
16402 mode http
16403 log global
16404 server srv1 127.0.0.1:8000
16405
16406 >>> Feb 6 12:12:09 localhost \
16407 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16408 (www/HTTP)
16409
16410 Field Format Extract from the example above
16411 1 process_name '[' pid ']:' haproxy[14385]:
16412 2 'Connect from' Connect from
16413 3 source_ip ':' source_port 10.0.1.2:33312
16414 4 'to' to
16415 5 destination_ip ':' destination_port 10.0.3.31:8012
16416 6 '(' frontend_name '/' mode ')' (www/HTTP)
16417
16418Detailed fields description :
16419 - "source_ip" is the IP address of the client which initiated the connection.
16420 - "source_port" is the TCP port of the client which initiated the connection.
16421 - "destination_ip" is the IP address the client connected to.
16422 - "destination_port" is the TCP port the client connected to.
16423 - "frontend_name" is the name of the frontend (or listener) which received
16424 and processed the connection.
16425 - "mode is the mode the frontend is operating (TCP or HTTP).
16426
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016427In case of a UNIX socket, the source and destination addresses are marked as
16428"unix:" and the ports reflect the internal ID of the socket which accepted the
16429connection (the same ID as reported in the stats).
16430
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016431It is advised not to use this deprecated format for newer installations as it
16432will eventually disappear.
16433
16434
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200164358.2.2. TCP log format
16436---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016437
16438The TCP format is used when "option tcplog" is specified in the frontend, and
16439is the recommended format for pure TCP proxies. It provides a lot of precious
16440information for troubleshooting. Since this format includes timers and byte
16441counts, the log is normally emitted at the end of the session. It can be
16442emitted earlier if "option logasap" is specified, which makes sense in most
16443environments with long sessions such as remote terminals. Sessions which match
16444the "monitor" rules are never logged. It is also possible not to emit logs for
16445sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016446specifying "option dontlognull" in the frontend. Successful connections will
16447not be logged if "option dontlog-normal" is specified in the frontend. A few
16448fields may slightly vary depending on some configuration options, those are
16449marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016450
16451 Example :
16452 frontend fnt
16453 mode tcp
16454 option tcplog
16455 log global
16456 default_backend bck
16457
16458 backend bck
16459 server srv1 127.0.0.1:8000
16460
16461 >>> Feb 6 12:12:56 localhost \
16462 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
16463 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
16464
16465 Field Format Extract from the example above
16466 1 process_name '[' pid ']:' haproxy[14387]:
16467 2 client_ip ':' client_port 10.0.1.2:33313
16468 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
16469 4 frontend_name fnt
16470 5 backend_name '/' server_name bck/srv1
16471 6 Tw '/' Tc '/' Tt* 0/0/5007
16472 7 bytes_read* 212
16473 8 termination_state --
16474 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16475 10 srv_queue '/' backend_queue 0/0
16476
16477Detailed fields description :
16478 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016479 connection to haproxy. If the connection was accepted on a UNIX socket
16480 instead, the IP address would be replaced with the word "unix". Note that
16481 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016482 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016483 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016484 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016485
16486 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016487 If the connection was accepted on a UNIX socket instead, the port would be
16488 replaced with the ID of the accepting socket, which is also reported in the
16489 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016490
16491 - "accept_date" is the exact date when the connection was received by haproxy
16492 (which might be very slightly different from the date observed on the
16493 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016494 the same date which may appear in any upstream firewall's log. When used in
16495 HTTP mode, the accept_date field will be reset to the first moment the
16496 connection is ready to receive a new request (end of previous response for
16497 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016498
16499 - "frontend_name" is the name of the frontend (or listener) which received
16500 and processed the connection.
16501
16502 - "backend_name" is the name of the backend (or listener) which was selected
16503 to manage the connection to the server. This will be the same as the
16504 frontend if no switching rule has been applied, which is common for TCP
16505 applications.
16506
16507 - "server_name" is the name of the last server to which the connection was
16508 sent, which might differ from the first one if there were connection errors
16509 and a redispatch occurred. Note that this server belongs to the backend
16510 which processed the request. If the connection was aborted before reaching
16511 a server, "<NOSRV>" is indicated instead of a server name.
16512
16513 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16514 It can be "-1" if the connection was aborted before reaching the queue.
16515 See "Timers" below for more details.
16516
16517 - "Tc" is the total time in milliseconds spent waiting for the connection to
16518 establish to the final server, including retries. It can be "-1" if the
16519 connection was aborted before a connection could be established. See
16520 "Timers" below for more details.
16521
16522 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016523 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016524 "option logasap" was specified, then the time counting stops at the moment
16525 the log is emitted. In this case, a '+' sign is prepended before the value,
16526 indicating that the final one will be larger. See "Timers" below for more
16527 details.
16528
16529 - "bytes_read" is the total number of bytes transmitted from the server to
16530 the client when the log is emitted. If "option logasap" is specified, the
16531 this value will be prefixed with a '+' sign indicating that the final one
16532 may be larger. Please note that this value is a 64-bit counter, so log
16533 analysis tools must be able to handle it without overflowing.
16534
16535 - "termination_state" is the condition the session was in when the session
16536 ended. This indicates the session state, which side caused the end of
16537 session to happen, and for what reason (timeout, error, ...). The normal
16538 flags should be "--", indicating the session was closed by either end with
16539 no data remaining in buffers. See below "Session state at disconnection"
16540 for more details.
16541
16542 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016543 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016544 limits have been reached. For instance, if actconn is close to 512 when
16545 multiple connection errors occur, chances are high that the system limits
16546 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016547 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016548
16549 - "feconn" is the total number of concurrent connections on the frontend when
16550 the session was logged. It is useful to estimate the amount of resource
16551 required to sustain high loads, and to detect when the frontend's "maxconn"
16552 has been reached. Most often when this value increases by huge jumps, it is
16553 because there is congestion on the backend servers, but sometimes it can be
16554 caused by a denial of service attack.
16555
16556 - "beconn" is the total number of concurrent connections handled by the
16557 backend when the session was logged. It includes the total number of
16558 concurrent connections active on servers as well as the number of
16559 connections pending in queues. It is useful to estimate the amount of
16560 additional servers needed to support high loads for a given application.
16561 Most often when this value increases by huge jumps, it is because there is
16562 congestion on the backend servers, but sometimes it can be caused by a
16563 denial of service attack.
16564
16565 - "srv_conn" is the total number of concurrent connections still active on
16566 the server when the session was logged. It can never exceed the server's
16567 configured "maxconn" parameter. If this value is very often close or equal
16568 to the server's "maxconn", it means that traffic regulation is involved a
16569 lot, meaning that either the server's maxconn value is too low, or that
16570 there aren't enough servers to process the load with an optimal response
16571 time. When only one of the server's "srv_conn" is high, it usually means
16572 that this server has some trouble causing the connections to take longer to
16573 be processed than on other servers.
16574
16575 - "retries" is the number of connection retries experienced by this session
16576 when trying to connect to the server. It must normally be zero, unless a
16577 server is being stopped at the same moment the connection was attempted.
16578 Frequent retries generally indicate either a network problem between
16579 haproxy and the server, or a misconfigured system backlog on the server
16580 preventing new connections from being queued. This field may optionally be
16581 prefixed with a '+' sign, indicating that the session has experienced a
16582 redispatch after the maximal retry count has been reached on the initial
16583 server. In this case, the server name appearing in the log is the one the
16584 connection was redispatched to, and not the first one, though both may
16585 sometimes be the same in case of hashing for instance. So as a general rule
16586 of thumb, when a '+' is present in front of the retry count, this count
16587 should not be attributed to the logged server.
16588
16589 - "srv_queue" is the total number of requests which were processed before
16590 this one in the server queue. It is zero when the request has not gone
16591 through the server queue. It makes it possible to estimate the approximate
16592 server's response time by dividing the time spent in queue by the number of
16593 requests in the queue. It is worth noting that if a session experiences a
16594 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016595 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016596 backend queue unless a redispatch occurs.
16597
16598 - "backend_queue" is the total number of requests which were processed before
16599 this one in the backend's global queue. It is zero when the request has not
16600 gone through the global queue. It makes it possible to estimate the average
16601 queue length, which easily translates into a number of missing servers when
16602 divided by a server's "maxconn" parameter. It is worth noting that if a
16603 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016604 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016605 through both the server queue and the backend queue unless a redispatch
16606 occurs.
16607
16608
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200166098.2.3. HTTP log format
16610----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016611
16612The HTTP format is the most complete and the best suited for HTTP proxies. It
16613is enabled by when "option httplog" is specified in the frontend. It provides
16614the same level of information as the TCP format with additional features which
16615are specific to the HTTP protocol. Just like the TCP format, the log is usually
16616emitted at the end of the session, unless "option logasap" is specified, which
16617generally only makes sense for download sites. A session which matches the
16618"monitor" rules will never logged. It is also possible not to log sessions for
16619which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016620frontend. Successful connections will not be logged if "option dontlog-normal"
16621is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016622
16623Most fields are shared with the TCP log, some being different. A few fields may
16624slightly vary depending on some configuration options. Those ones are marked
16625with a star ('*') after the field name below.
16626
16627 Example :
16628 frontend http-in
16629 mode http
16630 option httplog
16631 log global
16632 default_backend bck
16633
16634 backend static
16635 server srv1 127.0.0.1:8000
16636
16637 >>> Feb 6 12:14:14 localhost \
16638 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16639 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 +010016640 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016641
16642 Field Format Extract from the example above
16643 1 process_name '[' pid ']:' haproxy[14389]:
16644 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016645 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016646 4 frontend_name http-in
16647 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016648 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016649 7 status_code 200
16650 8 bytes_read* 2750
16651 9 captured_request_cookie -
16652 10 captured_response_cookie -
16653 11 termination_state ----
16654 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16655 13 srv_queue '/' backend_queue 0/0
16656 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16657 15 '{' captured_response_headers* '}' {}
16658 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016659
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016660Detailed fields description :
16661 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016662 connection to haproxy. If the connection was accepted on a UNIX socket
16663 instead, the IP address would be replaced with the word "unix". Note that
16664 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016665 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016666 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016667 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016668
16669 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016670 If the connection was accepted on a UNIX socket instead, the port would be
16671 replaced with the ID of the accepting socket, which is also reported in the
16672 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016673
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016674 - "request_date" is the exact date when the first byte of the HTTP request
16675 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016676
16677 - "frontend_name" is the name of the frontend (or listener) which received
16678 and processed the connection.
16679
16680 - "backend_name" is the name of the backend (or listener) which was selected
16681 to manage the connection to the server. This will be the same as the
16682 frontend if no switching rule has been applied.
16683
16684 - "server_name" is the name of the last server to which the connection was
16685 sent, which might differ from the first one if there were connection errors
16686 and a redispatch occurred. Note that this server belongs to the backend
16687 which processed the request. If the request was aborted before reaching a
16688 server, "<NOSRV>" is indicated instead of a server name. If the request was
16689 intercepted by the stats subsystem, "<STATS>" is indicated instead.
16690
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016691 - "TR" is the total time in milliseconds spent waiting for a full HTTP
16692 request from the client (not counting body) after the first byte was
16693 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050016694 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016695 always be very small because a request generally fits in one single packet.
16696 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020016697 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
16698 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016699
16700 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16701 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020016702 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016703
16704 - "Tc" is the total time in milliseconds spent waiting for the connection to
16705 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020016706 request was aborted before a connection could be established. See section
16707 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016708
16709 - "Tr" is the total time in milliseconds spent waiting for the server to send
16710 a full HTTP response, not counting data. It can be "-1" if the request was
16711 aborted before a complete response could be received. It generally matches
16712 the server's processing time for the request, though it may be altered by
16713 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020016714 "GET" requests generally indicate an overloaded server. See section 8.4
16715 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016716
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016717 - "Ta" is the time the request remained active in haproxy, which is the total
16718 time in milliseconds elapsed between the first byte of the request was
16719 received and the last byte of response was sent. It covers all possible
16720 processing except the handshake (see Th) and idle time (see Ti). There is
16721 one exception, if "option logasap" was specified, then the time counting
16722 stops at the moment the log is emitted. In this case, a '+' sign is
16723 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020016724 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016725
16726 - "status_code" is the HTTP status code returned to the client. This status
16727 is generally set by the server, but it might also be set by haproxy when
16728 the server cannot be reached or when its response is blocked by haproxy.
16729
16730 - "bytes_read" is the total number of bytes transmitted to the client when
16731 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050016732 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016733 the final one may be larger. Please note that this value is a 64-bit
16734 counter, so log analysis tools must be able to handle it without
16735 overflowing.
16736
16737 - "captured_request_cookie" is an optional "name=value" entry indicating that
16738 the client had this cookie in the request. The cookie name and its maximum
16739 length are defined by the "capture cookie" statement in the frontend
16740 configuration. The field is a single dash ('-') when the option is not
16741 set. Only one cookie may be captured, it is generally used to track session
16742 ID exchanges between a client and a server to detect session crossing
16743 between clients due to application bugs. For more details, please consult
16744 the section "Capturing HTTP headers and cookies" below.
16745
16746 - "captured_response_cookie" is an optional "name=value" entry indicating
16747 that the server has returned a cookie with its response. The cookie name
16748 and its maximum length are defined by the "capture cookie" statement in the
16749 frontend configuration. The field is a single dash ('-') when the option is
16750 not set. Only one cookie may be captured, it is generally used to track
16751 session ID exchanges between a client and a server to detect session
16752 crossing between clients due to application bugs. For more details, please
16753 consult the section "Capturing HTTP headers and cookies" below.
16754
16755 - "termination_state" is the condition the session was in when the session
16756 ended. This indicates the session state, which side caused the end of
16757 session to happen, for what reason (timeout, error, ...), just like in TCP
16758 logs, and information about persistence operations on cookies in the last
16759 two characters. The normal flags should begin with "--", indicating the
16760 session was closed by either end with no data remaining in buffers. See
16761 below "Session state at disconnection" for more details.
16762
16763 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016764 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016765 limits have been reached. For instance, if actconn is close to 512 or 1024
16766 when multiple connection errors occur, chances are high that the system
16767 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016768 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016769 system.
16770
16771 - "feconn" is the total number of concurrent connections on the frontend when
16772 the session was logged. It is useful to estimate the amount of resource
16773 required to sustain high loads, and to detect when the frontend's "maxconn"
16774 has been reached. Most often when this value increases by huge jumps, it is
16775 because there is congestion on the backend servers, but sometimes it can be
16776 caused by a denial of service attack.
16777
16778 - "beconn" is the total number of concurrent connections handled by the
16779 backend when the session was logged. It includes the total number of
16780 concurrent connections active on servers as well as the number of
16781 connections pending in queues. It is useful to estimate the amount of
16782 additional servers needed to support high loads for a given application.
16783 Most often when this value increases by huge jumps, it is because there is
16784 congestion on the backend servers, but sometimes it can be caused by a
16785 denial of service attack.
16786
16787 - "srv_conn" is the total number of concurrent connections still active on
16788 the server when the session was logged. It can never exceed the server's
16789 configured "maxconn" parameter. If this value is very often close or equal
16790 to the server's "maxconn", it means that traffic regulation is involved a
16791 lot, meaning that either the server's maxconn value is too low, or that
16792 there aren't enough servers to process the load with an optimal response
16793 time. When only one of the server's "srv_conn" is high, it usually means
16794 that this server has some trouble causing the requests to take longer to be
16795 processed than on other servers.
16796
16797 - "retries" is the number of connection retries experienced by this session
16798 when trying to connect to the server. It must normally be zero, unless a
16799 server is being stopped at the same moment the connection was attempted.
16800 Frequent retries generally indicate either a network problem between
16801 haproxy and the server, or a misconfigured system backlog on the server
16802 preventing new connections from being queued. This field may optionally be
16803 prefixed with a '+' sign, indicating that the session has experienced a
16804 redispatch after the maximal retry count has been reached on the initial
16805 server. In this case, the server name appearing in the log is the one the
16806 connection was redispatched to, and not the first one, though both may
16807 sometimes be the same in case of hashing for instance. So as a general rule
16808 of thumb, when a '+' is present in front of the retry count, this count
16809 should not be attributed to the logged server.
16810
16811 - "srv_queue" is the total number of requests which were processed before
16812 this one in the server queue. It is zero when the request has not gone
16813 through the server queue. It makes it possible to estimate the approximate
16814 server's response time by dividing the time spent in queue by the number of
16815 requests in the queue. It is worth noting that if a session experiences a
16816 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016817 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016818 backend queue unless a redispatch occurs.
16819
16820 - "backend_queue" is the total number of requests which were processed before
16821 this one in the backend's global queue. It is zero when the request has not
16822 gone through the global queue. It makes it possible to estimate the average
16823 queue length, which easily translates into a number of missing servers when
16824 divided by a server's "maxconn" parameter. It is worth noting that if a
16825 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016826 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016827 through both the server queue and the backend queue unless a redispatch
16828 occurs.
16829
16830 - "captured_request_headers" is a list of headers captured in the request due
16831 to the presence of the "capture request header" statement in the frontend.
16832 Multiple headers can be captured, they will be delimited by a vertical bar
16833 ('|'). When no capture is enabled, the braces do not appear, causing a
16834 shift of remaining fields. It is important to note that this field may
16835 contain spaces, and that using it requires a smarter log parser than when
16836 it's not used. Please consult the section "Capturing HTTP headers and
16837 cookies" below for more details.
16838
16839 - "captured_response_headers" is a list of headers captured in the response
16840 due to the presence of the "capture response header" statement in the
16841 frontend. Multiple headers can be captured, they will be delimited by a
16842 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16843 causing a shift of remaining fields. It is important to note that this
16844 field may contain spaces, and that using it requires a smarter log parser
16845 than when it's not used. Please consult the section "Capturing HTTP headers
16846 and cookies" below for more details.
16847
16848 - "http_request" is the complete HTTP request line, including the method,
16849 request and HTTP version string. Non-printable characters are encoded (see
16850 below the section "Non-printable characters"). This is always the last
16851 field, and it is always delimited by quotes and is the only one which can
16852 contain quotes. If new fields are added to the log format, they will be
16853 added before this field. This field might be truncated if the request is
16854 huge and does not fit in the standard syslog buffer (1024 characters). This
16855 is the reason why this field must always remain the last one.
16856
16857
Cyril Bontédc4d9032012-04-08 21:57:39 +0200168588.2.4. Custom log format
16859------------------------
William Lallemand48940402012-01-30 16:47:22 +010016860
Willy Tarreau2beef582012-12-20 17:22:52 +010016861The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016862mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016863
Davor Ocelice9ed2812017-12-25 17:49:28 +010016864HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016865Variables can take arguments using braces ('{}'), and multiple arguments are
16866separated by commas within the braces. Flags may be added or removed by
16867prefixing them with a '+' or '-' sign.
16868
16869Special variable "%o" may be used to propagate its flags to all other
16870variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016871("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016872
Willy Tarreauc8368452012-12-21 00:09:23 +010016873If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016874as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016875less common information such as the client's SSL certificate's DN, or to log
16876the key that would be used to store an entry into a stick table.
16877
William Lallemand48940402012-01-30 16:47:22 +010016878Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016879In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016880in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016881
Dragan Dosen835b9212016-02-12 13:23:03 +010016882Note: when using the RFC5424 syslog message format, the characters '"',
16883'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16884https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16885such cases, the use of the flag "E" should be considered.
16886
William Lallemand48940402012-01-30 16:47:22 +010016887Flags are :
16888 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016889 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016890 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16891 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016892
16893 Example:
16894
16895 log-format %T\ %t\ Some\ Text
16896 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16897
Dragan Dosen835b9212016-02-12 13:23:03 +010016898 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16899
William Lallemand48940402012-01-30 16:47:22 +010016900At the moment, the default HTTP format is defined this way :
16901
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016902 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16903 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016904
William Lallemandbddd4fd2012-02-27 11:23:10 +010016905the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016906
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016907 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16908 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16909 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016910
William Lallemandbddd4fd2012-02-27 11:23:10 +010016911and the default TCP format is defined this way :
16912
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016913 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16914 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016915
William Lallemand48940402012-01-30 16:47:22 +010016916Please refer to the table below for currently defined variables :
16917
William Lallemandbddd4fd2012-02-27 11:23:10 +010016918 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016919 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016920 +---+------+-----------------------------------------------+-------------+
16921 | | %o | special variable, apply flags on all next var | |
16922 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016923 | | %B | bytes_read (from server to client) | numeric |
16924 | H | %CC | captured_request_cookie | string |
16925 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016926 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016927 | H | %HM | HTTP method (ex: POST) | string |
16928 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016929 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016930 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16931 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016932 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016933 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016934 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016935 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016936 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016937 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016938 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016939 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16940 | H | %Ti | idle time before the HTTP request | numeric |
16941 | H | %Tq | Th + Ti + TR | numeric |
16942 | H | %TR | time to receive the full request from 1st byte| numeric |
16943 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016944 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016945 | | %Tt | Tt | numeric |
16946 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016947 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016948 | | %ac | actconn | numeric |
16949 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016950 | | %bc | beconn (backend concurrent connections) | numeric |
16951 | | %bi | backend_source_ip (connecting address) | IP |
16952 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016953 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016954 | | %ci | client_ip (accepted address) | IP |
16955 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016956 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016957 | | %fc | feconn (frontend concurrent connections) | numeric |
16958 | | %fi | frontend_ip (accepting address) | IP |
16959 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016960 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016961 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016962 | | %hr | captured_request_headers default style | string |
16963 | | %hrl | captured_request_headers CLF style | string list |
16964 | | %hs | captured_response_headers default style | string |
16965 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016966 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016967 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016968 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016969 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016970 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016971 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016972 | | %sc | srv_conn (server concurrent connections) | numeric |
16973 | | %si | server_IP (target address) | IP |
16974 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016975 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016976 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16977 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016978 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016979 | H | %tr | date_time of HTTP request | date |
16980 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020016981 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016982 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016983 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016984 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016985
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016986 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016987
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016988
169898.2.5. Error log format
16990-----------------------
16991
16992When an incoming connection fails due to an SSL handshake or an invalid PROXY
16993protocol header, haproxy will log the event using a shorter, fixed line format.
16994By default, logs are emitted at the LOG_INFO level, unless the option
16995"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016996will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016997logged if the "dontlognull" option is set.
16998
16999The format looks like this :
17000
17001 >>> Dec 3 18:27:14 localhost \
17002 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
17003 Connection error during SSL handshake
17004
17005 Field Format Extract from the example above
17006 1 process_name '[' pid ']:' haproxy[6103]:
17007 2 client_ip ':' client_port 127.0.0.1:56059
17008 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
17009 4 frontend_name "/" bind_name ":" frt/f1:
17010 5 message Connection error during SSL handshake
17011
17012These fields just provide minimal information to help debugging connection
17013failures.
17014
17015
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170168.3. Advanced logging options
17017-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017018
17019Some advanced logging options are often looked for but are not easy to find out
17020just by looking at the various options. Here is an entry point for the few
17021options which can enable better logging. Please refer to the keywords reference
17022for more information about their usage.
17023
17024
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170258.3.1. Disabling logging of external tests
17026------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017027
17028It is quite common to have some monitoring tools perform health checks on
17029haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
17030commercial load-balancer, and sometimes it will simply be a more complete
17031monitoring system such as Nagios. When the tests are very frequent, users often
17032ask how to disable logging for those checks. There are three possibilities :
17033
17034 - if connections come from everywhere and are just TCP probes, it is often
17035 desired to simply disable logging of connections without data exchange, by
17036 setting "option dontlognull" in the frontend. It also disables logging of
17037 port scans, which may or may not be desired.
17038
17039 - if the connection come from a known source network, use "monitor-net" to
17040 declare this network as monitoring only. Any host in this network will then
17041 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017042 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017043 such as other load-balancers.
17044
17045 - if the tests are performed on a known URI, use "monitor-uri" to declare
17046 this URI as dedicated to monitoring. Any host sending this request will
17047 only get the result of a health-check, and the request will not be logged.
17048
17049
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170508.3.2. Logging before waiting for the session to terminate
17051----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017052
17053The problem with logging at end of connection is that you have no clue about
17054what is happening during very long sessions, such as remote terminal sessions
17055or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010017056"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017057just before data transfer begins. This means that in case of TCP, it will still
17058log the connection status to the server, and in case of HTTP, it will log just
17059after processing the server headers. In this case, the number of bytes reported
17060is the number of header bytes sent to the client. In order to avoid confusion
17061with normal logs, the total time field and the number of bytes are prefixed
17062with a '+' sign which means that real numbers are certainly larger.
17063
17064
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170658.3.3. Raising log level upon errors
17066------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017067
17068Sometimes it is more convenient to separate normal traffic from errors logs,
17069for instance in order to ease error monitoring from log files. When the option
17070"log-separate-errors" is used, connections which experience errors, timeouts,
17071retries, redispatches or HTTP status codes 5xx will see their syslog level
17072raised from "info" to "err". This will help a syslog daemon store the log in
17073a separate file. It is very important to keep the errors in the normal traffic
17074file too, so that log ordering is not altered. You should also be careful if
17075you already have configured your syslog daemon to store all logs higher than
17076"notice" in an "admin" file, because the "err" level is higher than "notice".
17077
17078
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170798.3.4. Disabling logging of successful connections
17080--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017081
17082Although this may sound strange at first, some large sites have to deal with
17083multiple thousands of logs per second and are experiencing difficulties keeping
17084them intact for a long time or detecting errors within them. If the option
17085"dontlog-normal" is set on the frontend, all normal connections will not be
17086logged. In this regard, a normal connection is defined as one without any
17087error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
17088and a response with a status 5xx is not considered normal and will be logged
17089too. Of course, doing is is really discouraged as it will remove most of the
17090useful information from the logs. Do this only if you have no other
17091alternative.
17092
17093
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170948.4. Timing events
17095------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017096
17097Timers provide a great help in troubleshooting network problems. All values are
17098reported in milliseconds (ms). These timers should be used in conjunction with
17099the session termination flags. In TCP mode with "option tcplog" set on the
17100frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017101mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
17102addition, three other measures are provided, "Th", "Ti", and "Tq".
17103
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010017104Timings events in HTTP mode:
17105
17106 first request 2nd request
17107 |<-------------------------------->|<-------------- ...
17108 t tr t tr ...
17109 ---|----|----|----|----|----|----|----|----|--
17110 : Th Ti TR Tw Tc Tr Td : Ti ...
17111 :<---- Tq ---->: :
17112 :<-------------- Tt -------------->:
17113 :<--------- Ta --------->:
17114
17115Timings events in TCP mode:
17116
17117 TCP session
17118 |<----------------->|
17119 t t
17120 ---|----|----|----|----|---
17121 | Th Tw Tc Td |
17122 |<------ Tt ------->|
17123
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017124 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017125 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017126 only happen once during the whole connection's lifetime. A large time here
17127 may indicate that the client only pre-established the connection without
17128 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017129 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017130 SSL handshake was very expensive to compute. Please note that this time is
17131 reported only before the first request, so it is safe to average it over
17132 all request to calculate the amortized value. The second and subsequent
17133 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017134
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017135 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17136 counts between the end of the handshakes and the first byte of the HTTP
17137 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017138 to count after the end of the transmission the previous response. When a
17139 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17140 after the previous request. Some browsers pre-establish connections to a
17141 server in order to reduce the latency of a future request, and keep them
17142 pending until they need it. This delay will be reported as the idle time. A
17143 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017144
17145 - TR: total time to get the client request (HTTP mode only). It's the time
17146 elapsed between the first bytes received and the moment the proxy received
17147 the empty line marking the end of the HTTP headers. The value "-1"
17148 indicates that the end of headers has never been seen. This happens when
17149 the client closes prematurely or times out. This time is usually very short
17150 since most requests fit in a single packet. A large time may indicate a
17151 request typed by hand during a test.
17152
17153 - Tq: total time to get the client request from the accept date or since the
17154 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017155 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 +020017156 returns -1 as well. This timer used to be very useful before the arrival of
17157 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17158 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17159 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017160
17161 - Tw: total time spent in the queues waiting for a connection slot. It
17162 accounts for backend queue as well as the server queues, and depends on the
17163 queue size, and the time needed for the server to complete previous
17164 requests. The value "-1" means that the request was killed before reaching
17165 the queue, which is generally what happens with invalid or denied requests.
17166
17167 - Tc: total time to establish the TCP connection to the server. It's the time
17168 elapsed between the moment the proxy sent the connection request, and the
17169 moment it was acknowledged by the server, or between the TCP SYN packet and
17170 the matching SYN/ACK packet in return. The value "-1" means that the
17171 connection never established.
17172
17173 - Tr: server response time (HTTP mode only). It's the time elapsed between
17174 the moment the TCP connection was established to the server and the moment
17175 the server sent its complete response headers. It purely shows its request
17176 processing time, without the network overhead due to the data transmission.
17177 It is worth noting that when the client has data to send to the server, for
17178 instance during a POST request, the time already runs, and this can distort
17179 apparent response time. For this reason, it's generally wise not to trust
17180 too much this field for POST requests initiated from clients behind an
17181 untrusted network. A value of "-1" here means that the last the response
17182 header (empty line) was never seen, most likely because the server timeout
17183 stroke before the server managed to process the request.
17184
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017185 - Ta: total active time for the HTTP request, between the moment the proxy
17186 received the first byte of the request header and the emission of the last
17187 byte of the response body. The exception is when the "logasap" option is
17188 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17189 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17190 by subtracting other timers when valid :
17191
17192 Td = Ta - (TR + Tw + Tc + Tr)
17193
17194 Timers with "-1" values have to be excluded from this equation. Note that
17195 "Ta" can never be negative.
17196
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017197 - Tt: total session duration time, between the moment the proxy accepted it
17198 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017199 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17200 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017201 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017202
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017203 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017204
17205 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017206 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17207 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017208
17209These timers provide precious indications on trouble causes. Since the TCP
17210protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17211that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017212due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17213"Tt" is close to a timeout value specified in the configuration, it often means
17214that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017215
17216Most common cases :
17217
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017218 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17219 the client and the proxy. This is very rare on local networks but might
17220 happen when clients are on far remote networks and send large requests. It
17221 may happen that values larger than usual appear here without any network
17222 cause. Sometimes, during an attack or just after a resource starvation has
17223 ended, haproxy may accept thousands of connections in a few milliseconds.
17224 The time spent accepting these connections will inevitably slightly delay
17225 processing of other connections, and it can happen that request times in the
17226 order of a few tens of milliseconds are measured after a few thousands of
17227 new connections have been accepted at once. Using one of the keep-alive
17228 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017229 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017230
17231 - If "Tc" is close to 3000, a packet has probably been lost between the
17232 server and the proxy during the server connection phase. This value should
17233 always be very low, such as 1 ms on local networks and less than a few tens
17234 of ms on remote networks.
17235
Willy Tarreau55165fe2009-05-10 12:02:55 +020017236 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17237 to be the average majored by 3000, there are probably some packets lost
17238 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017239
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017240 - If "Ta" is large even for small byte counts, it generally is because
17241 neither the client nor the server decides to close the connection while
17242 haproxy is running in tunnel mode and both have agreed on a keep-alive
17243 connection mode. In order to solve this issue, it will be needed to specify
17244 one of the HTTP options to manipulate keep-alive or close options on either
17245 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17246 important when connection regulation is used with the "maxconn" option on
17247 the servers, since no new connection will be sent to the server until
17248 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017249
17250Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17251
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017252 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017253 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017254 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017255
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017256 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017257 or it aborted too early. Check the session termination flags
17258 then "timeout http-request" and "timeout client" settings.
17259
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017260 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017261 servers were out of order, because the request was invalid
17262 or forbidden by ACL rules. Check the session termination
17263 flags.
17264
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017265 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17266 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017267 Check the session termination flags, then check the
17268 "timeout connect" setting. Note that the tarpit action might
17269 return similar-looking patterns, with "Tw" equal to the time
17270 the client connection was maintained open.
17271
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017272 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017273 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017274 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017275 termination flags, then check the "timeout server" setting.
17276
17277
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172788.5. Session state at disconnection
17279-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017280
17281TCP and HTTP logs provide a session termination indicator in the
17282"termination_state" field, just before the number of active connections. It is
172832-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17284each of which has a special meaning :
17285
17286 - On the first character, a code reporting the first event which caused the
17287 session to terminate :
17288
17289 C : the TCP session was unexpectedly aborted by the client.
17290
17291 S : the TCP session was unexpectedly aborted by the server, or the
17292 server explicitly refused it.
17293
17294 P : the session was prematurely aborted by the proxy, because of a
17295 connection limit enforcement, because a DENY filter was matched,
17296 because of a security check which detected and blocked a dangerous
17297 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010017298 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020017299
17300 L : the session was locally processed by haproxy and was not passed to
17301 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017302
17303 R : a resource on the proxy has been exhausted (memory, sockets, source
17304 ports, ...). Usually, this appears during the connection phase, and
17305 system logs should contain a copy of the precise error. If this
17306 happens, it must be considered as a very serious anomaly which
17307 should be fixed as soon as possible by any means.
17308
17309 I : an internal error was identified by the proxy during a self-check.
17310 This should NEVER happen, and you are encouraged to report any log
17311 containing this, because this would almost certainly be a bug. It
17312 would be wise to preventively restart the process after such an
17313 event too, in case it would be caused by memory corruption.
17314
Simon Horman752dc4a2011-06-21 14:34:59 +090017315 D : the session was killed by haproxy because the server was detected
17316 as down and was configured to kill all connections when going down.
17317
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070017318 U : the session was killed by haproxy on this backup server because an
17319 active server was detected as up and was configured to kill all
17320 backup connections when going up.
17321
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017322 K : the session was actively killed by an admin operating on haproxy.
17323
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017324 c : the client-side timeout expired while waiting for the client to
17325 send or receive data.
17326
17327 s : the server-side timeout expired while waiting for the server to
17328 send or receive data.
17329
17330 - : normal session completion, both the client and the server closed
17331 with nothing left in the buffers.
17332
17333 - on the second character, the TCP or HTTP session state when it was closed :
17334
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017335 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017336 (HTTP mode only). Nothing was sent to any server.
17337
17338 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17339 only happen when servers have a 'maxconn' parameter set. It can
17340 also happen in the global queue after a redispatch consecutive to
17341 a failed attempt to connect to a dying server. If no redispatch is
17342 reported, then no connection attempt was made to any server.
17343
17344 C : the proxy was waiting for the CONNECTION to establish on the
17345 server. The server might at most have noticed a connection attempt.
17346
17347 H : the proxy was waiting for complete, valid response HEADERS from the
17348 server (HTTP only).
17349
17350 D : the session was in the DATA phase.
17351
17352 L : the proxy was still transmitting LAST data to the client while the
17353 server had already finished. This one is very rare as it can only
17354 happen when the client dies while receiving the last packets.
17355
17356 T : the request was tarpitted. It has been held open with the client
17357 during the whole "timeout tarpit" duration or until the client
17358 closed, both of which will be reported in the "Tw" timer.
17359
17360 - : normal session completion after end of data transfer.
17361
17362 - the third character tells whether the persistence cookie was provided by
17363 the client (only in HTTP mode) :
17364
17365 N : the client provided NO cookie. This is usually the case for new
17366 visitors, so counting the number of occurrences of this flag in the
17367 logs generally indicate a valid trend for the site frequentation.
17368
17369 I : the client provided an INVALID cookie matching no known server.
17370 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017371 cookies between HTTP/HTTPS sites, persistence conditionally
17372 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017373
17374 D : the client provided a cookie designating a server which was DOWN,
17375 so either "option persist" was used and the client was sent to
17376 this server, or it was not set and the client was redispatched to
17377 another server.
17378
Willy Tarreau996a92c2010-10-13 19:30:47 +020017379 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017380 server.
17381
Willy Tarreau996a92c2010-10-13 19:30:47 +020017382 E : the client provided a valid cookie, but with a last date which was
17383 older than what is allowed by the "maxidle" cookie parameter, so
17384 the cookie is consider EXPIRED and is ignored. The request will be
17385 redispatched just as if there was no cookie.
17386
17387 O : the client provided a valid cookie, but with a first date which was
17388 older than what is allowed by the "maxlife" cookie parameter, so
17389 the cookie is consider too OLD and is ignored. The request will be
17390 redispatched just as if there was no cookie.
17391
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017392 U : a cookie was present but was not used to select the server because
17393 some other server selection mechanism was used instead (typically a
17394 "use-server" rule).
17395
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017396 - : does not apply (no cookie set in configuration).
17397
17398 - the last character reports what operations were performed on the persistence
17399 cookie returned by the server (only in HTTP mode) :
17400
17401 N : NO cookie was provided by the server, and none was inserted either.
17402
17403 I : no cookie was provided by the server, and the proxy INSERTED one.
17404 Note that in "cookie insert" mode, if the server provides a cookie,
17405 it will still be overwritten and reported as "I" here.
17406
Willy Tarreau996a92c2010-10-13 19:30:47 +020017407 U : the proxy UPDATED the last date in the cookie that was presented by
17408 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017409 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017410 date indicated in the cookie. If any other change happens, such as
17411 a redispatch, then the cookie will be marked as inserted instead.
17412
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017413 P : a cookie was PROVIDED by the server and transmitted as-is.
17414
17415 R : the cookie provided by the server was REWRITTEN by the proxy, which
17416 happens in "cookie rewrite" or "cookie prefix" modes.
17417
17418 D : the cookie provided by the server was DELETED by the proxy.
17419
17420 - : does not apply (no cookie set in configuration).
17421
Willy Tarreau996a92c2010-10-13 19:30:47 +020017422The combination of the two first flags gives a lot of information about what
17423was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017424helpful to detect server saturation, network troubles, local system resource
17425starvation, attacks, etc...
17426
17427The most common termination flags combinations are indicated below. They are
17428alphabetically sorted, with the lowercase set just after the upper case for
17429easier finding and understanding.
17430
17431 Flags Reason
17432
17433 -- Normal termination.
17434
17435 CC The client aborted before the connection could be established to the
17436 server. This can happen when haproxy tries to connect to a recently
17437 dead (or unchecked) server, and the client aborts while haproxy is
17438 waiting for the server to respond or for "timeout connect" to expire.
17439
17440 CD The client unexpectedly aborted during data transfer. This can be
17441 caused by a browser crash, by an intermediate equipment between the
17442 client and haproxy which decided to actively break the connection,
17443 by network routing issues between the client and haproxy, or by a
17444 keep-alive session between the server and the client terminated first
17445 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010017446
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017447 cD The client did not send nor acknowledge any data for as long as the
17448 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020017449 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017450
17451 CH The client aborted while waiting for the server to start responding.
17452 It might be the server taking too long to respond or the client
17453 clicking the 'Stop' button too fast.
17454
17455 cH The "timeout client" stroke while waiting for client data during a
17456 POST request. This is sometimes caused by too large TCP MSS values
17457 for PPPoE networks which cannot transport full-sized packets. It can
17458 also happen when client timeout is smaller than server timeout and
17459 the server takes too long to respond.
17460
17461 CQ The client aborted while its session was queued, waiting for a server
17462 with enough empty slots to accept it. It might be that either all the
17463 servers were saturated or that the assigned server was taking too
17464 long a time to respond.
17465
17466 CR The client aborted before sending a full HTTP request. Most likely
17467 the request was typed by hand using a telnet client, and aborted
17468 too early. The HTTP status code is likely a 400 here. Sometimes this
17469 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017470 and the client. "option http-ignore-probes" can be used to ignore
17471 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017472
17473 cR The "timeout http-request" stroke before the client sent a full HTTP
17474 request. This is sometimes caused by too large TCP MSS values on the
17475 client side for PPPoE networks which cannot transport full-sized
17476 packets, or by clients sending requests by hand and not typing fast
17477 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017478 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017479 some browsers started to implement a "pre-connect" feature consisting
17480 in speculatively connecting to some recently visited web sites just
17481 in case the user would like to visit them. This results in many
17482 connections being established to web sites, which end up in 408
17483 Request Timeout if the timeout strikes first, or 400 Bad Request when
17484 the browser decides to close them first. These ones pollute the log
17485 and feed the error counters. Some versions of some browsers have even
17486 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017487 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017488 http-ignore-probes" in the frontend, resulting in connections with
17489 zero data transfer to be totally ignored. This will definitely hide
17490 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017491
17492 CT The client aborted while its session was tarpitted. It is important to
17493 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017494 wrong tarpit rules have been written. If a lot of them happen, it
17495 might make sense to lower the "timeout tarpit" value to something
17496 closer to the average reported "Tw" timer, in order not to consume
17497 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017498
Willy Tarreau570f2212013-06-10 16:42:09 +020017499 LR The request was intercepted and locally handled by haproxy. Generally
17500 it means that this was a redirect or a stats request.
17501
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017502 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017503 the TCP connection (the proxy received a TCP RST or an ICMP message
17504 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017505 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017506 or no ARP response on local network). When this happens in HTTP mode,
17507 the status code is likely a 502 or 503 here.
17508
17509 sC The "timeout connect" stroke before a connection to the server could
17510 complete. When this happens in HTTP mode, the status code is likely a
17511 503 or 504 here.
17512
17513 SD The connection to the server died with an error during the data
17514 transfer. This usually means that haproxy has received an RST from
17515 the server or an ICMP message from an intermediate equipment while
17516 exchanging data with the server. This can be caused by a server crash
17517 or by a network issue on an intermediate equipment.
17518
17519 sD The server did not send nor acknowledge any data for as long as the
17520 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017521 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017522 load-balancers, ...), as well as keep-alive sessions maintained
17523 between the client and the server expiring first on haproxy.
17524
17525 SH The server aborted before sending its full HTTP response headers, or
17526 it crashed while processing the request. Since a server aborting at
17527 this moment is very rare, it would be wise to inspect its logs to
17528 control whether it crashed and why. The logged request may indicate a
17529 small set of faulty requests, demonstrating bugs in the application.
17530 Sometimes this might also be caused by an IDS killing the connection
17531 between haproxy and the server.
17532
17533 sH The "timeout server" stroke before the server could return its
17534 response headers. This is the most common anomaly, indicating too
17535 long transactions, probably caused by server or database saturation.
17536 The immediate workaround consists in increasing the "timeout server"
17537 setting, but it is important to keep in mind that the user experience
17538 will suffer from these long response times. The only long term
17539 solution is to fix the application.
17540
17541 sQ The session spent too much time in queue and has been expired. See
17542 the "timeout queue" and "timeout connect" settings to find out how to
17543 fix this if it happens too often. If it often happens massively in
17544 short periods, it may indicate general problems on the affected
17545 servers due to I/O or database congestion, or saturation caused by
17546 external attacks.
17547
17548 PC The proxy refused to establish a connection to the server because the
17549 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017550 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017551 so that it does not happen anymore. This status is very rare and
17552 might happen when the global "ulimit-n" parameter is forced by hand.
17553
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017554 PD The proxy blocked an incorrectly formatted chunked encoded message in
17555 a request or a response, after the server has emitted its headers. In
17556 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017557 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017558 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017559
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017560 PH The proxy blocked the server's response, because it was invalid,
17561 incomplete, dangerous (cache control), or matched a security filter.
17562 In any case, an HTTP 502 error is sent to the client. One possible
17563 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017564 containing unauthorized characters. It is also possible but quite
17565 rare, that the proxy blocked a chunked-encoding request from the
17566 client due to an invalid syntax, before the server responded. In this
17567 case, an HTTP 400 error is sent to the client and reported in the
17568 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017569
17570 PR The proxy blocked the client's HTTP request, either because of an
17571 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17572 the client, or because a deny filter matched, in which case it
17573 returned an HTTP 403 error.
17574
17575 PT The proxy blocked the client's request and has tarpitted its
17576 connection before returning it a 500 server error. Nothing was sent
17577 to the server. The connection was maintained open for as long as
17578 reported by the "Tw" timer field.
17579
17580 RC A local resource has been exhausted (memory, sockets, source ports)
17581 preventing the connection to the server from establishing. The error
17582 logs will tell precisely what was missing. This is very rare and can
17583 only be solved by proper system tuning.
17584
Willy Tarreau996a92c2010-10-13 19:30:47 +020017585The combination of the two last flags gives a lot of information about how
17586persistence was handled by the client, the server and by haproxy. This is very
17587important to troubleshoot disconnections, when users complain they have to
17588re-authenticate. The commonly encountered flags are :
17589
17590 -- Persistence cookie is not enabled.
17591
17592 NN No cookie was provided by the client, none was inserted in the
17593 response. For instance, this can be in insert mode with "postonly"
17594 set on a GET request.
17595
17596 II A cookie designating an invalid server was provided by the client,
17597 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017598 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017599 value can be presented by a client when no other server knows it.
17600
17601 NI No cookie was provided by the client, one was inserted in the
17602 response. This typically happens for first requests from every user
17603 in "insert" mode, which makes it an easy way to count real users.
17604
17605 VN A cookie was provided by the client, none was inserted in the
17606 response. This happens for most responses for which the client has
17607 already got a cookie.
17608
17609 VU A cookie was provided by the client, with a last visit date which is
17610 not completely up-to-date, so an updated cookie was provided in
17611 response. This can also happen if there was no date at all, or if
17612 there was a date but the "maxidle" parameter was not set, so that the
17613 cookie can be switched to unlimited time.
17614
17615 EI A cookie was provided by the client, with a last visit date which is
17616 too old for the "maxidle" parameter, so the cookie was ignored and a
17617 new cookie was inserted in the response.
17618
17619 OI A cookie was provided by the client, with a first visit date which is
17620 too old for the "maxlife" parameter, so the cookie was ignored and a
17621 new cookie was inserted in the response.
17622
17623 DI The server designated by the cookie was down, a new server was
17624 selected and a new cookie was emitted in the response.
17625
17626 VI The server designated by the cookie was not marked dead but could not
17627 be reached. A redispatch happened and selected another one, which was
17628 then advertised in the response.
17629
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017630
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176318.6. Non-printable characters
17632-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017633
17634In order not to cause trouble to log analysis tools or terminals during log
17635consulting, non-printable characters are not sent as-is into log files, but are
17636converted to the two-digits hexadecimal representation of their ASCII code,
17637prefixed by the character '#'. The only characters that can be logged without
17638being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17639escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17640is the same for the character '"' which becomes "#22", as well as '{', '|' and
17641'}' when logging headers.
17642
17643Note that the space character (' ') is not encoded in headers, which can cause
17644issues for tools relying on space count to locate fields. A typical header
17645containing spaces is "User-Agent".
17646
17647Last, it has been observed that some syslog daemons such as syslog-ng escape
17648the quote ('"') with a backslash ('\'). The reverse operation can safely be
17649performed since no quote may appear anywhere else in the logs.
17650
17651
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176528.7. Capturing HTTP cookies
17653---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017654
17655Cookie capture simplifies the tracking a complete user session. This can be
17656achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017657section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017658cookie will simultaneously be checked in the request ("Cookie:" header) and in
17659the response ("Set-Cookie:" header). The respective values will be reported in
17660the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017661locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017662not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17663user switches to a new session for example, because the server will reassign it
17664a new cookie. It is also possible to detect if a server unexpectedly sets a
17665wrong cookie to a client, leading to session crossing.
17666
17667 Examples :
17668 # capture the first cookie whose name starts with "ASPSESSION"
17669 capture cookie ASPSESSION len 32
17670
17671 # capture the first cookie whose name is exactly "vgnvisitor"
17672 capture cookie vgnvisitor= len 32
17673
17674
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176758.8. Capturing HTTP headers
17676---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017677
17678Header captures are useful to track unique request identifiers set by an upper
17679proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
17680the response, one can search for information about the response length, how the
17681server asked the cache to behave, or an object location during a redirection.
17682
17683Header captures are performed using the "capture request header" and "capture
17684response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017685section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017686
17687It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017688time. Non-existent headers are logged as empty strings, and if one header
17689appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017690are grouped within braces '{' and '}' in the same order as they were declared,
17691and delimited with a vertical bar '|' without any space. Response headers
17692follow the same representation, but are displayed after a space following the
17693request headers block. These blocks are displayed just before the HTTP request
17694in the logs.
17695
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017696As a special case, it is possible to specify an HTTP header capture in a TCP
17697frontend. The purpose is to enable logging of headers which will be parsed in
17698an HTTP backend if the request is then switched to this HTTP backend.
17699
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017700 Example :
17701 # This instance chains to the outgoing proxy
17702 listen proxy-out
17703 mode http
17704 option httplog
17705 option logasap
17706 log global
17707 server cache1 192.168.1.1:3128
17708
17709 # log the name of the virtual server
17710 capture request header Host len 20
17711
17712 # log the amount of data uploaded during a POST
17713 capture request header Content-Length len 10
17714
17715 # log the beginning of the referrer
17716 capture request header Referer len 20
17717
17718 # server name (useful for outgoing proxies only)
17719 capture response header Server len 20
17720
17721 # logging the content-length is useful with "option logasap"
17722 capture response header Content-Length len 10
17723
Davor Ocelice9ed2812017-12-25 17:49:28 +010017724 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017725 capture response header Cache-Control len 8
17726
17727 # the Via header will report the next proxy's name
17728 capture response header Via len 20
17729
17730 # log the URL location during a redirection
17731 capture response header Location len 20
17732
17733 >>> Aug 9 20:26:09 localhost \
17734 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
17735 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
17736 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
17737 "GET http://fr.adserver.yahoo.com/"
17738
17739 >>> Aug 9 20:30:46 localhost \
17740 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
17741 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
17742 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017743 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017744
17745 >>> Aug 9 20:30:46 localhost \
17746 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
17747 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
17748 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
17749 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017750 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017751
17752
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177538.9. Examples of logs
17754---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017755
17756These are real-world examples of logs accompanied with an explanation. Some of
17757them have been made up by hand. The syslog part has been removed for better
17758reading. Their sole purpose is to explain how to decipher them.
17759
17760 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17761 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17762 "HEAD / HTTP/1.0"
17763
17764 => long request (6.5s) entered by hand through 'telnet'. The server replied
17765 in 147 ms, and the session ended normally ('----')
17766
17767 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17768 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17769 0/9 "HEAD / HTTP/1.0"
17770
17771 => Idem, but the request was queued in the global queue behind 9 other
17772 requests, and waited there for 1230 ms.
17773
17774 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17775 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17776 "GET /image.iso HTTP/1.0"
17777
17778 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017779 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017780 14 ms, 243 bytes of headers were sent to the client, and total time from
17781 accept to first data byte is 30 ms.
17782
17783 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17784 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17785 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17786
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020017787 => the proxy blocked a server response either because of an "http-response
17788 deny" rule, or because the response was improperly formatted and not
17789 HTTP-compliant, or because it blocked sensitive information which risked
17790 being cached. In this case, the response is replaced with a "502 bad
17791 gateway". The flags ("PH--") tell us that it was haproxy who decided to
17792 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017793
17794 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017795 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 +010017796
17797 => the client never completed its request and aborted itself ("C---") after
17798 8.5s, while the proxy was waiting for the request headers ("-R--").
17799 Nothing was sent to any server.
17800
17801 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17802 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17803
17804 => The client never completed its request, which was aborted by the
17805 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017806 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017807 send a 408 return code to the client.
17808
17809 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17810 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17811
17812 => This log was produced with "option tcplog". The client timed out after
17813 5 seconds ("c----").
17814
17815 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17816 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017817 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017818
17819 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017820 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017821 (config says 'retries 3'), and no redispatch (otherwise we would have
17822 seen "/+3"). Status code 503 was returned to the client. There were 115
17823 connections on this server, 202 connections on this proxy, and 205 on
17824 the global process. It is possible that the server refused the
17825 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017826
Willy Tarreau52b2d222011-09-07 23:48:48 +020017827
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200178289. Supported filters
17829--------------------
17830
17831Here are listed officially supported filters with the list of parameters they
17832accept. Depending on compile options, some of these filters might be
17833unavailable. The list of available filters is reported in haproxy -vv.
17834
17835See also : "filter"
17836
178379.1. Trace
17838----------
17839
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017840filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017841
17842 Arguments:
17843 <name> is an arbitrary name that will be reported in
17844 messages. If no name is provided, "TRACE" is used.
17845
17846 <random-parsing> enables the random parsing of data exchanged between
17847 the client and the server. By default, this filter
17848 parses all available data. With this parameter, it
17849 only parses a random amount of the available data.
17850
Davor Ocelice9ed2812017-12-25 17:49:28 +010017851 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017852 default, this filter forwards all previously parsed
17853 data. With this parameter, it only forwards a random
17854 amount of the parsed data.
17855
Davor Ocelice9ed2812017-12-25 17:49:28 +010017856 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017857
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017858This filter can be used as a base to develop new filters. It defines all
17859callbacks and print a message on the standard error stream (stderr) with useful
17860information for all of them. It may be useful to debug the activity of other
17861filters or, quite simply, HAProxy's activity.
17862
17863Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17864tests the behavior of a filter that parses data exchanged between a client and
17865a server by adding some latencies in the processing.
17866
17867
178689.2. HTTP compression
17869---------------------
17870
17871filter compression
17872
17873The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17874keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020017875when no other filter is used, it is enough. When used with the cache or the
17876fcgi-app enabled, it is also enough. In this case, the compression is always
17877done after the response is stored in the cache. But it is mandatory to
17878explicitly use a filter line to enable the HTTP compression when at least one
17879filter other than the cache or the fcgi-app is used for the same
17880listener/frontend/backend. This is important to know the filters evaluation
17881order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017882
Christopher Fauletb30b3102019-09-12 23:03:09 +020017883See also : "compression", section 9.4 about the cache filter and section 9.5
17884 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017885
17886
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200178879.3. Stream Processing Offload Engine (SPOE)
17888--------------------------------------------
17889
17890filter spoe [engine <name>] config <file>
17891
17892 Arguments :
17893
17894 <name> is the engine name that will be used to find the right scope in
17895 the configuration file. If not provided, all the file will be
17896 parsed.
17897
17898 <file> is the path of the engine configuration file. This file can
17899 contain configuration of several engines. In this case, each
17900 part must be placed in its own scope.
17901
17902The Stream Processing Offload Engine (SPOE) is a filter communicating with
17903external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017904streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017905exchanged with them are configured in dedicated files, for the main part. It
17906also requires dedicated backends, defined in HAProxy configuration.
17907
17908SPOE communicates with external components using an in-house binary protocol,
17909the Stream Processing Offload Protocol (SPOP).
17910
Tim Düsterhus4896c442016-11-29 02:15:19 +010017911For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017912"doc/SPOE.txt".
17913
17914Important note:
17915 The SPOE filter is highly experimental for now and was not heavily
17916 tested. It is really not production ready. So use it carefully.
17917
Christopher Faulet99a17a22018-12-11 09:18:27 +0100179189.4. Cache
17919----------
17920
17921filter cache <name>
17922
17923 Arguments :
17924
17925 <name> is name of the cache section this filter will use.
17926
17927The cache uses a filter to store cacheable responses. The HTTP rules
17928"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050017929cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020017930other filters than fcgi-app or compression are used, it is enough. In such
17931case, the compression filter is always evaluated after the cache filter. But it
17932is mandatory to explicitly use a filter line to use a cache when at least one
17933filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010017934listener/frontend/backend. This is important to know the filters evaluation
17935order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010017936
Christopher Fauletb30b3102019-09-12 23:03:09 +020017937See also : section 9.2 about the compression filter, section 9.5 about the
17938 fcgi-app filter and section 6 about cache.
17939
17940
179419.5. Fcgi-app
17942-------------
17943
17944filter fcg-app <name>
17945
17946 Arguments :
17947
17948 <name> is name of the fcgi-app section this filter will use.
17949
17950The FastCGI application uses a filter to evaluate all custom parameters on the
17951request path, and to process the headers on the response path. the <name> must
17952reference an existing fcgi-app section. The directive "use-fcgi-app" should be
17953used to define the application to use. By default the corresponding filter is
17954implicitly defined. And when no other filters than cache or compression are
17955used, it is enough. But it is mandatory to explicitly use a filter line to a
17956fcgi-app when at least one filter other than the compression or the cache is
17957used for the same backend. This is important to know the filters evaluation
17958order.
17959
17960See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
17961 about the cache filter and section 10 about FastCGI application.
17962
17963
1796410. FastCGI applications
17965-------------------------
17966
17967HAProxy is able to send HTTP requests to Responder FastCGI applications. This
17968feature was added in HAProxy 2.1. To do so, servers must be configured to use
17969the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
17970FastCGI application must be configured and used by the backend managing these
17971servers (using the keyword "use-fcgi-app" into the proxy section). Several
17972FastCGI applications may be defined, but only one can be used at a time by a
17973backend.
17974
17975HAProxy implements all features of the FastCGI specification for Responder
17976application. Especially it is able to multiplex several requests on a simple
17977connection.
17978
1797910.1. Setup
17980-----------
17981
1798210.1.1. Fcgi-app section
17983--------------------------
17984
17985fcgi-app <name>
17986 Declare a FastCGI application named <name>. To be valid, at least the
17987 document root must be defined.
17988
17989acl <aclname> <criterion> [flags] [operator] <value> ...
17990 Declare or complete an access list.
17991
17992 See "acl" keyword in section 4.2 and section 7 about ACL usage for
17993 details. ACLs defined for a FastCGI application are private. They cannot be
17994 used by any other application or by any proxy. In the same way, ACLs defined
17995 in any other section are not usable by a FastCGI application. However,
17996 Pre-defined ACLs are available.
17997
17998docroot <path>
17999 Define the document root on the remote host. <path> will be used to build
18000 the default value of FastCGI parameters SCRIPT_FILENAME and
18001 PATH_TRANSLATED. It is a mandatory setting.
18002
18003index <script-name>
18004 Define the script name that will be appended after an URI that ends with a
18005 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
18006 is an optional setting.
18007
18008 Example :
18009 index index.php
18010
18011log-stderr global
18012log-stderr <address> [len <length>] [format <format>]
18013 [sample <ranges>:<smp_size>] <facility> [<level> [<minlevel>]]
18014 Enable logging of STDERR messages reported by the FastCGI application.
18015
18016 See "log" keyword in section 4.2 for details. It is an optional setting. By
18017 default STDERR messages are ignored.
18018
18019pass-header <name> [ { if | unless } <condition> ]
18020 Specify the name of a request header which will be passed to the FastCGI
18021 application. It may optionally be followed by an ACL-based condition, in
18022 which case it will only be evaluated if the condition is true.
18023
18024 Most request headers are already available to the FastCGI application,
18025 prefixed with "HTTP_". Thus, this directive is only required to pass headers
18026 that are purposefully omitted. Currently, the headers "Authorization",
18027 "Proxy-Authorization" and hop-by-hop headers are omitted.
18028
18029 Note that the headers "Content-type" and "Content-length" are never passed to
18030 the FastCGI application because they are already converted into parameters.
18031
18032path-info <regex>
18033 Define a regular expression to extract the script-name and the path-info
18034 from the URI. Thus, <regex> should have two captures: the first one to
18035 capture the script name and the second one to capture the path-info. It is an
18036 optional setting. If it is not defined, no matching is performed on the
18037 URI. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not filled.
18038
18039 Example :
18040 path-info ^(/.+\.php)(/.*)?$
18041
18042option get-values
18043no option get-values
18044 Enable or disable the retrieve of variables about connection management.
18045
18046 HAproxy is able to send the record FCGI_GET_VALUES on connection
18047 establishment to retrieve the value for following variables:
18048
18049 * FCGI_MAX_REQS The maximum number of concurrent requests this
18050 application will accept.
18051
18052 * FCGI_MPXS_CONNS “0” if this application does not multiplex connections,
18053 “1” otherwise.
18054
18055 Some FastCGI applications does not support this feature. Some others close
18056 the connexion immediately after sending their response. So, by default, this
18057 option is disabled.
18058
18059 Note that the maximum number of concurrent requests accepted by a FastCGI
18060 application is a connection variable. It only limits the number of streams
18061 per connection. If the global load must be limited on the application, the
18062 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
18063 an application does not support connection multiplexing, the maximum number
18064 of concurrent requests is automatically set to 1.
18065
18066option keep-conn
18067no option keep-conn
18068 Instruct the FastCGI application to keep the connection open or not after
18069 sending a response.
18070
18071 If disabled, the FastCGI application closes the connection after responding
18072 to this request. By default, this option is enabled.
18073
18074option max-reqs <reqs>
18075 Define the maximum number of concurrent requests this application will
18076 accept.
18077
18078 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
18079 during connection establishment. Furthermore, if the application does not
18080 support connection multiplexing, this option will be ignored. By default set
18081 to 1.
18082
18083option mpxs-conns
18084no option mpxs-conns
18085 Enable or disable the support of connection multiplexing.
18086
18087 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
18088 during connection establishment. It is disabled by default.
18089
18090set-param <name> <fmt> [ { if | unless } <condition> ]
18091 Set a FastCGI parameter that should be passed to this application. Its
18092 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
18093 "Custom Log format"). It may optionally be followed by an ACL-based
18094 condition, in which case it will only be evaluated if the condition is true.
18095
18096 With this directive, it is possible to overwrite the value of default FastCGI
18097 parameters. If the value is evaluated to an empty string, the rule is
18098 ignored. These directives are evaluated in their declaration order.
18099
18100 Example :
18101 # PHP only, required if PHP was built with --enable-force-cgi-redirect
18102 set-param REDIRECT_STATUS 200
18103
18104 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
18105
18106
1810710.1.2. Proxy section
18108---------------------
18109
18110use-fcgi-app <name>
18111 Define the FastCGI application to use for the backend.
18112
18113 Arguments :
18114 <name> is the name of the FastCGI application to use.
18115
18116 This keyword is only available for HTTP proxies with the backend capability
18117 and with at least one FastCGI server. However, FastCGI servers can be mixed
18118 with HTTP servers. But except there is a good reason to do so, it is not
18119 recommended (see section 10.3 about the limitations for details). Only one
18120 application may be defined at a time per backend.
18121
18122 Note that, once a FastCGI application is referenced for a backend, depending
18123 on the configuration some processing may be done even if the request is not
18124 sent to a FastCGI server. Rules to set parameters or pass headers to an
18125 application are evaluated.
18126
18127
1812810.1.3. Example
18129---------------
18130
18131 frontend front-http
18132 mode http
18133 bind *:80
18134 bind *:
18135
18136 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
18137 default_backend back-static
18138
18139 backend back-static
18140 mode http
18141 server www A.B.C.D:80
18142
18143 backend back-dynamic
18144 mode http
18145 use-fcgi-app php-fpm
18146 server php-fpm A.B.C.D:9000 proto fcgi
18147
18148 fcgi-app php-fpm
18149 log-stderr global
18150 option keep-conn
18151
18152 docroot /var/www/my-app
18153 index index.php
18154 path-info ^(/.+\.php)(/.*)?$
18155
18156
1815710.2. Default parameters
18158------------------------
18159
18160A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
18161the CGI/1.1 specification (RFC3875), several variables must be passed to the
18162scipt. So HAProxy set them and some others commonly used by FastCGI
18163applications. All these variables may be overwritten, with caution though.
18164
18165 +-------------------+-----------------------------------------------------+
18166 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
18167 | | to authenticate the user. Concretely, only the |
18168 | | BASIC authentication mechanism is supported. |
18169 | | |
18170 +-------------------+-----------------------------------------------------+
18171 | CONTENT_LENGTH | Contains the size of the message-body attached to |
18172 | | the request. It means only requests with a known |
18173 | | size are considered as valid and sent to the |
18174 | | application. |
18175 | | |
18176 +-------------------+-----------------------------------------------------+
18177 | CONTENT_TYPE | Contains the type of the message-body attached to |
18178 | | the request. It may not be set. |
18179 | | |
18180 +-------------------+-----------------------------------------------------+
18181 | DOCUMENT_ROOT | Contains the document root on the remote host under |
18182 | | which the script should be executed, as defined in |
18183 | | the application's configuration. |
18184 | | |
18185 +-------------------+-----------------------------------------------------+
18186 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
18187 | | to communicate with the FastCGI application. |
18188 | | Concretely, it is set to "CGI/1.1". |
18189 | | |
18190 +-------------------+-----------------------------------------------------+
18191 | PATH_INFO | Contains the portion of the URI path hierarchy |
18192 | | following the part that identifies the script |
18193 | | itself. To be set, the directive "path-info" must |
18194 | | be defined. |
18195 | | |
18196 +-------------------+-----------------------------------------------------+
18197 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
18198 | | It is the concatenation of DOCUMENT_ROOT and |
18199 | | PATH_INFO. If PATH_INFO is not set, this parameters |
18200 | | is not set too. |
18201 | | |
18202 +-------------------+-----------------------------------------------------+
18203 | QUERY_STRING | Contains the request's query string. It may not be |
18204 | | set. |
18205 | | |
18206 +-------------------+-----------------------------------------------------+
18207 | REMOTE_ADDR | Contains the network address of the client sending |
18208 | | the request. |
18209 | | |
18210 +-------------------+-----------------------------------------------------+
18211 | REMOTE_USER | Contains the user identification string supplied by |
18212 | | client as part of user authentication. |
18213 | | |
18214 +-------------------+-----------------------------------------------------+
18215 | REQUEST_METHOD | Contains the method which should be used by the |
18216 | | script to process the request. |
18217 | | |
18218 +-------------------+-----------------------------------------------------+
18219 | REQUEST_URI | Contains the request's URI. |
18220 | | |
18221 +-------------------+-----------------------------------------------------+
18222 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
18223 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
18224 | | |
18225 +-------------------+-----------------------------------------------------+
18226 | SCRIPT_NAME | Contains the name of the script. If the directive |
18227 | | "path-info" is defined, it is the first part of the |
18228 | | URI path hierarchy, ending with the script name. |
18229 | | Otherwise, it is the entire URI path. |
18230 | | |
18231 +-------------------+-----------------------------------------------------+
18232 | SERVER_NAME | Contains the name of the server host to which the |
18233 | | client request is directed. It is the value of the |
18234 | | header "Host", if defined. Otherwise, the |
18235 | | destination address of the connection on the client |
18236 | | side. |
18237 | | |
18238 +-------------------+-----------------------------------------------------+
18239 | SERVER_PORT | Contains the destination TCP port of the connection |
18240 | | on the client side, which is the port the client |
18241 | | connected to. |
18242 | | |
18243 +-------------------+-----------------------------------------------------+
18244 | SERVER_PROTOCOL | Contains the request's protocol. |
18245 | | |
18246 +-------------------+-----------------------------------------------------+
18247 | HTTPS | Set to a non-empty value ("on") if the script was |
18248 | | queried through the HTTPS protocol. |
18249 | | |
18250 +-------------------+-----------------------------------------------------+
18251
18252
1825310.3. Limitations
18254------------------
18255
18256The current implementation have some limitations. The first one is about the
18257way some request headers are hidden to the FastCGI applications. This happens
18258during the headers analysis, on the backend side, before the connection
18259establishment. At this stage, HAProxy know the backend is using a FastCGI
18260application but it don't know if the request will be routed to a FastCGI server
18261or not. But to hide request headers, it simply removes them from the HTX
18262message. So, if the request is finally routed to an HTTP server, it never see
18263these headers. For this reason, it is not recommended to mix FastCGI servers
18264and HTTP servers under the same backend.
18265
18266Similarly, the rules "set-param" and "pass-header" are evaluated during the
18267request headers analysis. So the evaluation is always performed, even if the
18268requests is finally forwarded to an HTTP server.
18269
18270About the rules "set-param", when a rule is applied, a pseudo header is added
18271into the HTX message. So, the same way than for HTTP header rewrites, it may
18272fail if the buffer is full. The rules "set-param" will compete with
18273"http-request" ones.
18274
18275Finally, all FastCGI params and HTTP headers are sent into a unique record
18276FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
18277processing error is returned. It means the record FCGI_PARAM, once encoded,
18278must not exceeds the size of a buffer. However, there is no reserve to respect
18279here.
William Lallemand86d0df02017-11-24 21:36:45 +010018280
Willy Tarreau0ba27502007-12-24 16:55:16 +010018281/*
18282 * Local variables:
18283 * fill-column: 79
18284 * End:
18285 */