blob: e1257accf553bb26620f89bc5be03b4d1e303e5f [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 Tarreauc70df532019-10-25 15:48:53 +02007 2019/10/25
Willy Tarreau6a06a402007-07-15 20:15:28 +02008
9
10This document covers the configuration language as implemented in the version
Davor Ocelice9ed2812017-12-25 17:49:28 +010011specified above. It does not provide any hints, examples, or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010012documentation, please refer to the Reference Manual or the Architecture Manual.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013The summary below is meant to help you find sections by name and navigate
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014through the document.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016Note to documentation contributors :
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017 This document is formatted with 80 columns per line, with even number of
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018 spaces for indentation and without tabs. Please follow these rules strictly
19 so that it remains easily printable everywhere. If a line needs to be
20 printed verbatim and does not fit, please end each line with a backslash
Willy Tarreau62a36c42010-08-17 15:53:10 +020021 ('\') and continue on next line, indented by two characters. It is also
Davor Ocelice9ed2812017-12-25 17:49:28 +010022 sometimes useful to prefix all output lines (logs, console outputs) with 3
23 closing angle brackets ('>>>') in order to emphasize the difference between
24 inputs and outputs when they may be ambiguous. If you add sections,
Willy Tarreau62a36c42010-08-17 15:53:10 +020025 please update the summary below for easier searching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026
27
28Summary
29-------
30
311. Quick reminder about HTTP
321.1. The HTTP transaction model
331.2. HTTP request
Davor Ocelice9ed2812017-12-25 17:49:28 +0100341.2.1. The request line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200351.2.2. The request headers
361.3. HTTP response
Davor Ocelice9ed2812017-12-25 17:49:28 +0100371.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381.3.2. The response headers
39
402. Configuring HAProxy
412.1. Configuration file format
William Lallemandf9873ba2015-05-05 17:37:14 +0200422.2. Quoting and escaping
William Lallemandb2f07452015-05-12 14:27:13 +0200432.3. Environment variables
442.4. Time format
452.5. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020046
473. Global parameters
483.1. Process management and security
493.2. Performance tuning
503.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100513.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200523.5. Peers
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200533.6. Mailers
William Lallemandc9515522019-06-12 16:32:11 +0200543.7. Programs
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
William Dauchy0fec3ab2019-10-27 20:08:11 +0100632 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100633
Willy Tarreau6a06a402007-07-15 20:15:28 +0200634 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100635 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200636 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200637 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200638 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100639 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100640 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100641 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200642 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200643 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200644 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200645 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200646 - noepoll
647 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000648 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200649 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100650 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300651 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000652 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100653 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200654 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200655 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200656 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000657 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000658 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200659 - tune.buffers.limit
660 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200661 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200662 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100663 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200664 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200665 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200666 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100667 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200668 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200669 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100670 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100671 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100672 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100673 - tune.lua.session-timeout
674 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200675 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100676 - tune.maxaccept
677 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200678 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200679 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200680 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100681 - tune.rcvbuf.client
682 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100683 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200684 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100685 - tune.sndbuf.client
686 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100687 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100688 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200689 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100690 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200691 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200692 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100693 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200694 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100695 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200696 - tune.vars.reqres-max-size
697 - tune.vars.sess-max-size
698 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100699 - tune.zlib.memlevel
700 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100701
Willy Tarreau6a06a402007-07-15 20:15:28 +0200702 * Debugging
703 - debug
704 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200705
706
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007073.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200708------------------------------------
709
Emeric Brunc8e8d122012-10-02 18:42:10 +0200710ca-base <dir>
711 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200712 relative path is used with "ca-file" or "crl-file" directives. Absolute
713 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200714
Willy Tarreau6a06a402007-07-15 20:15:28 +0200715chroot <jail dir>
716 Changes current directory to <jail dir> and performs a chroot() there before
717 dropping privileges. This increases the security level in case an unknown
718 vulnerability would be exploited, since it would make it very hard for the
719 attacker to exploit the system. This only works when the process is started
720 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100721 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100722
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100723cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
724 On Linux 2.6 and above, it is possible to bind a process or a thread to a
725 specific CPU set. This means that the process or the thread will never run on
726 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
727 sets. The first argument is a process set, eventually followed by a thread
728 set. These sets have the format
729
730 all | odd | even | number[-[number]]
731
732 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100733 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100734 ignored. It is possible to specify a range with two such number delimited by
735 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100736 "all", only odd numbers using "odd" or even numbers using "even", just like
737 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100738 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 +0100739 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100740 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100741 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100742 specified. Each "cpu-map" directive will replace the previous ones when they
743 overlap. A thread will be bound on the intersection of its mapping and the
744 one of the process on which it is attached. If the intersection is null, no
745 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100746
Christopher Fauletff4121f2017-11-22 16:38:49 +0100747 Ranges can be partially defined. The higher bound can be omitted. In such
748 case, it is replaced by the corresponding maximum value, 32 or 64 depending
749 on the machine's word size.
750
Christopher Faulet26028f62017-11-22 15:01:51 +0100751 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100752 automatically bind a process or a thread to a CPU by incrementing
753 process/thread and CPU sets. To be valid, both sets must have the same
754 size. No matter the declaration order of the CPU sets, it will be bound from
755 the lowest to the highest bound. Having a process and a thread range with the
756 "auto:" prefix is not supported. Only one range is supported, the other one
757 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100758
759 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100760 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
761
762 cpu-map 1/all 0-3 # bind all threads of the first process on the
763 # first 4 CPUs
764
765 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
766 # or "cpu-map 1-32 0-31" depending on the machine's
767 # word size.
768
Christopher Faulet26028f62017-11-22 15:01:51 +0100769 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100770 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100771 cpu-map auto:1-4 0-3
772 cpu-map auto:1-4 0-1 2-3
773 cpu-map auto:1-4 3 2 1 0
774
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100775 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
776 # and so on.
777 cpu-map auto:1/1-4 0-3
778 cpu-map auto:1/1-4 0-1 2-3
779 cpu-map auto:1/1-4 3 2 1 0
780
Davor Ocelice9ed2812017-12-25 17:49:28 +0100781 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100782 cpu-map auto:all 0-63
783 cpu-map auto:even 0-31
784 cpu-map auto:odd 32-63
785
786 # invalid cpu-map because process and CPU sets have different sizes.
787 cpu-map auto:1-4 0 # invalid
788 cpu-map auto:1 0-3 # invalid
789
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100790 # invalid cpu-map because automatic binding is used with a process range
791 # and a thread range.
792 cpu-map auto:all/all 0 # invalid
793 cpu-map auto:all/1-4 0 # invalid
794 cpu-map auto:1-4/all 0 # invalid
795
Emeric Brunc8e8d122012-10-02 18:42:10 +0200796crt-base <dir>
797 Assigns a default directory to fetch SSL certificates from when a relative
798 path is used with "crtfile" directives. Absolute locations specified after
799 "crtfile" prevail and ignore "crt-base".
800
Willy Tarreau6a06a402007-07-15 20:15:28 +0200801daemon
802 Makes the process fork into background. This is the recommended mode of
803 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100804 disabled by the command line "-db" argument. This option is ignored in
805 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200806
David Carlier8167f302015-06-01 13:50:06 +0200807deviceatlas-json-file <path>
808 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100809 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200810
811deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100812 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200813 optional and set to 0 by default if not set.
814
815deviceatlas-separator <char>
816 Sets the character separator for the API properties results. This directive
817 is optional and set to | by default if not set.
818
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100819deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200820 Sets the client cookie's name used for the detection if the DeviceAtlas
821 Client-side component was used during the request. This directive is optional
822 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100823
Simon Horman98637e52014-06-20 12:30:16 +0900824external-check
825 Allows the use of an external agent to perform health checks.
826 This is disabled by default as a security precaution.
827 See "option external-check".
828
Willy Tarreau6a06a402007-07-15 20:15:28 +0200829gid <number>
830 Changes the process' group ID to <number>. It is recommended that the group
831 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
832 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100833 Note that if haproxy is started from a user having supplementary groups, it
834 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200835 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100836
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100837hard-stop-after <time>
838 Defines the maximum time allowed to perform a clean soft-stop.
839
840 Arguments :
841 <time> is the maximum time (by default in milliseconds) for which the
842 instance will remain alive when a soft-stop is received via the
843 SIGUSR1 signal.
844
845 This may be used to ensure that the instance will quit even if connections
846 remain opened during a soft-stop (for example with long timeouts for a proxy
847 in tcp mode). It applies both in TCP and HTTP mode.
848
849 Example:
850 global
851 hard-stop-after 30s
852
Christopher Faulet98fbe952019-07-22 16:18:24 +0200853h1-case-adjust <from> <to>
854 Defines the case adjustment to apply, when enabled, to the header name
855 <from>, to change it to <to> before sending it to HTTP/1 clients or
856 servers. <from> must be in lower case, and <from> and <to> must not differ
857 except for their case. It may be repeated if several header names need to be
858 ajusted. Duplicate entries are not allowed. If a lot of header names have to
859 be adjusted, it might be more convenient to use "h1-case-adjust-file".
860 Please note that no transformation will be applied unless "option
861 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
862 specified in a proxy.
863
864 There is no standard case for header names because, as stated in RFC7230,
865 they are case-insensitive. So applications must handle them in a case-
866 insensitive manner. But some bogus applications violate the standards and
867 erroneously rely on the cases most commonly used by browsers. This problem
868 becomes critical with HTTP/2 because all header names must be exchanged in
869 lower case, and HAProxy follows the same convention. All header names are
870 sent in lower case to clients and servers, regardless of the HTTP version.
871
872 Applications which fail to properly process requests or responses may require
873 to temporarily use such workarounds to adjust header names sent to them for
874 the time it takes the application to be fixed. Please note that an
875 application which requires such workarounds might be vulnerable to content
876 smuggling attacks and must absolutely be fixed.
877
878 Example:
879 global
880 h1-case-adjust content-length Content-Length
881
882 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
883 "option h1-case-adjust-bogus-server".
884
885h1-case-adjust-file <hdrs-file>
886 Defines a file containing a list of key/value pairs used to adjust the case
887 of some header names before sending them to HTTP/1 clients or servers. The
888 file <hdrs-file> must contain 2 header names per line. The first one must be
889 in lower case and both must not differ except for their case. Lines which
890 start with '#' are ignored, just like empty lines. Leading and trailing tabs
891 and spaces are stripped. Duplicate entries are not allowed. Please note that
892 no transformation will be applied unless "option h1-case-adjust-bogus-client"
893 or "option h1-case-adjust-bogus-server" is specified in a proxy.
894
895 If this directive is repeated, only the last one will be processed. It is an
896 alternative to the directive "h1-case-adjust" if a lot of header names need
897 to be adjusted. Please read the risks associated with using this.
898
899 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
900 "option h1-case-adjust-bogus-server".
901
Willy Tarreau6a06a402007-07-15 20:15:28 +0200902group <group name>
903 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
904 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100905
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200906log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
907 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100908 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100909 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100910 configured with "log global".
911
912 <address> can be one of:
913
Willy Tarreau2769aa02007-12-27 18:26:09 +0100914 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100915 no port is specified, 514 is used by default (the standard syslog
916 port).
917
David du Colombier24bb5f52011-03-17 10:40:23 +0100918 - An IPv6 address followed by a colon and optionally a UDP port. If
919 no port is specified, 514 is used by default (the standard syslog
920 port).
921
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100922 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100923 considerations for chroot (be sure the path is accessible inside
924 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100925 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100926
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100927 - A file descriptor number in the form "fd@<number>", which may point
928 to a pipe, terminal, or socket. In this case unbuffered logs are used
929 and one writev() call per log is performed. This is a bit expensive
930 but acceptable for most workloads. Messages sent this way will not be
931 truncated but may be dropped, in which case the DroppedLogs counter
932 will be incremented. The writev() call is atomic even on pipes for
933 messages up to PIPE_BUF size, which POSIX recommends to be at least
934 512 and which is 4096 bytes on most modern operating systems. Any
935 larger message may be interleaved with messages from other processes.
936 Exceptionally for debugging purposes the file descriptor may also be
937 directed to a file, but doing so will significantly slow haproxy down
938 as non-blocking calls will be ignored. Also there will be no way to
939 purge nor rotate this file without restarting the process. Note that
940 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100941 for use with a TCP syslog server. See also the "short" and "raw"
942 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100943
944 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
945 "fd@2", see above.
946
Willy Tarreauc046d162019-08-30 15:24:59 +0200947 - A ring buffer in the form "ring@<name>", which will correspond to an
948 in-memory ring buffer accessible over the CLI using the "show events"
949 command, which will also list existing rings and their sizes. Such
950 buffers are lost on reload or restart but when used as a complement
951 this can help troubleshooting by having the logs instantly available.
952
William Lallemandb2f07452015-05-12 14:27:13 +0200953 You may want to reference some environment variables in the address
954 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100955
Willy Tarreau18324f52014-06-27 18:10:07 +0200956 <length> is an optional maximum line length. Log lines larger than this value
957 will be truncated before being sent. The reason is that syslog
958 servers act differently on log line length. All servers support the
959 default value of 1024, but some servers simply drop larger lines
960 while others do log them. If a server supports long lines, it may
961 make sense to set this value here in order to avoid truncating long
962 lines. Similarly, if a server drops long lines, it is preferable to
963 truncate them before sending them. Accepted values are 80 to 65535
964 inclusive. The default value of 1024 is generally fine for all
965 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100966 JSON-formatted logs may require larger values. You may also need to
967 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200968
Dragan Dosen7ad31542015-09-28 17:16:47 +0200969 <format> is the log format used when generating syslog messages. It may be
970 one of the following :
971
972 rfc3164 The RFC3164 syslog message format. This is the default.
973 (https://tools.ietf.org/html/rfc3164)
974
975 rfc5424 The RFC5424 syslog message format.
976 (https://tools.ietf.org/html/rfc5424)
977
Willy Tarreaue8746a02018-11-12 08:45:00 +0100978 short A message containing only a level between angle brackets such as
979 '<3>', followed by the text. The PID, date, time, process name
980 and system name are omitted. This is designed to be used with a
981 local log server. This format is compatible with what the systemd
982 logger consumes.
983
Willy Tarreauc1b06452018-11-12 11:57:56 +0100984 raw A message containing only the text. The level, PID, date, time,
985 process name and system name are omitted. This is designed to be
986 used in containers or during development, where the severity only
987 depends on the file descriptor used (stdout/stderr).
988
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200989 <ranges> A list of comma-separated ranges to identify the logs to sample.
990 This is used to balance the load of the logs to send to the log
991 server. The limits of the ranges cannot be null. They are numbered
992 from 1. The size or period (in number of logs) of the sample must be
993 set with <sample_size> parameter.
994
995 <sample_size>
996 The size of the sample in number of logs to consider when balancing
997 their logging loads. It is used to balance the load of the logs to
998 send to the syslog server. This size must be greater or equal to the
999 maximum of the high limits of the ranges.
1000 (see also <ranges> parameter).
1001
Robert Tsai81ae1952007-12-05 10:47:29 +01001002 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001003
Willy Tarreaue8746a02018-11-12 08:45:00 +01001004 kern user mail daemon auth syslog lpr news
1005 uucp cron auth2 ftp ntp audit alert cron2
1006 local0 local1 local2 local3 local4 local5 local6 local7
1007
Willy Tarreauc1b06452018-11-12 11:57:56 +01001008 Note that the facility is ignored for the "short" and "raw"
1009 formats, but still required as a positional field. It is
1010 recommended to use "daemon" in this case to make it clear that
1011 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001012
1013 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001014 all messages are sent. If a maximum level is specified, only messages with a
1015 severity at least as important as this level will be sent. An optional minimum
1016 level can be specified. If it is set, logs emitted with a more severe level
1017 than this one will be capped to this level. This is used to avoid sending
1018 "emerg" messages on all terminals on some default syslog configurations.
1019 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001020
Cyril Bontédc4d9032012-04-08 21:57:39 +02001021 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001022
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001023log-send-hostname [<string>]
1024 Sets the hostname field in the syslog header. If optional "string" parameter
1025 is set the header is set to the string contents, otherwise uses the hostname
1026 of the system. Generally used if one is not relaying logs through an
1027 intermediate syslog server or for simply customizing the hostname printed in
1028 the logs.
1029
Kevinm48936af2010-12-22 16:08:21 +00001030log-tag <string>
1031 Sets the tag field in the syslog header to this string. It defaults to the
1032 program name as launched from the command line, which usually is "haproxy".
1033 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001034 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001035
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001036lua-load <file>
1037 This global directive loads and executes a Lua file. This directive can be
1038 used multiple times.
1039
William Lallemand4cfede82017-11-24 22:02:34 +01001040master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001041 Master-worker mode. It is equivalent to the command line "-W" argument.
1042 This mode will launch a "master" which will monitor the "workers". Using
1043 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001044 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001045 or daemon mode. It is recommended to use this mode with multiprocess and
1046 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001047 By default, if a worker exits with a bad return code, in the case of a
1048 segfault for example, all workers will be killed, and the master will leave.
1049 It is convenient to combine this behavior with Restart=on-failure in a
1050 systemd unit file in order to relaunch the whole process. If you don't want
1051 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001052
William Lallemand4cfede82017-11-24 22:02:34 +01001053 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001054
William Lallemand27edc4b2019-05-07 17:49:33 +02001055mworker-max-reloads <number>
1056 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001057 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001058 number of reloads is greater than this number, the worker will receive a
1059 SIGTERM. This option helps to keep under control the number of workers.
1060 See also "show proc" in the Management Guide.
1061
Willy Tarreau6a06a402007-07-15 20:15:28 +02001062nbproc <number>
1063 Creates <number> processes when going daemon. This requires the "daemon"
1064 mode. By default, only one process is created, which is the recommended mode
1065 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001066 process, it may be needed to fork multiple daemons. When set to a value
1067 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +01001068 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
1069 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001070
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001071nbthread <number>
1072 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001073 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1074 "nbproc" historically used to be the only way to use multiple processors, it
1075 also involved a number of shortcomings related to the lack of synchronization
1076 between processes (health-checks, peers, stick-tables, stats, ...) which do
1077 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001078 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1079 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1080 when nbproc is not used, the default "nbthread" value is automatically set to
1081 the number of CPUs the process is bound to upon startup. This means that the
1082 thread count can easily be adjusted from the calling process using commands
1083 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1084 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001085
Willy Tarreau6a06a402007-07-15 20:15:28 +02001086pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001087 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +02001088 the "-p" command line argument. The file must be accessible to the user
1089 starting the process. See also "daemon".
1090
Willy Tarreau1d549722016-02-16 12:41:57 +01001091presetenv <name> <value>
1092 Sets environment variable <name> to value <value>. If the variable exists, it
1093 is NOT overwritten. The changes immediately take effect so that the next line
1094 in the configuration file sees the new value. See also "setenv", "resetenv",
1095 and "unsetenv".
1096
1097resetenv [<name> ...]
1098 Removes all environment variables except the ones specified in argument. It
1099 allows to use a clean controlled environment before setting new values with
1100 setenv or unsetenv. Please note that some internal functions may make use of
1101 some environment variables, such as time manipulation functions, but also
1102 OpenSSL or even external checks. This must be used with extreme care and only
1103 after complete validation. The changes immediately take effect so that the
1104 next line in the configuration file sees the new environment. See also
1105 "setenv", "presetenv", and "unsetenv".
1106
Christopher Fauletff4121f2017-11-22 16:38:49 +01001107stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001108 Limits the stats socket to a certain set of processes numbers. By default the
1109 stats socket is bound to all processes, causing a warning to be emitted when
1110 nbproc is greater than 1 because there is no way to select the target process
1111 when connecting. However, by using this setting, it becomes possible to pin
1112 the stats socket to a specific set of processes, typically the first one. The
1113 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001114 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001115 word size (32 or 64). Ranges can be partially defined. The higher bound can
1116 be omitted. In such case, it is replaced by the corresponding maximum
1117 value. A better option consists in using the "process" setting of the "stats
1118 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001119
Baptiste Assmann5626f482015-08-23 10:00:10 +02001120server-state-base <directory>
1121 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001122 file names which do not start with a '/'. See also "server-state-file",
1123 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001124
1125server-state-file <file>
1126 Specifies the path to the file containing state of servers. If the path starts
1127 with a slash ('/'), it is considered absolute, otherwise it is considered
1128 relative to the directory specified using "server-state-base" (if set) or to
1129 the current directory. Before reloading HAProxy, it is possible to save the
1130 servers' current state using the stats command "show servers state". The
1131 output of this command must be written in the file pointed by <file>. When
1132 starting up, before handling traffic, HAProxy will read, load and apply state
1133 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001134 configuration. See also "server-state-base" and "show servers state",
1135 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001136
Willy Tarreau1d549722016-02-16 12:41:57 +01001137setenv <name> <value>
1138 Sets environment variable <name> to value <value>. If the variable exists, it
1139 is overwritten. The changes immediately take effect so that the next line in
1140 the configuration file sees the new value. See also "presetenv", "resetenv",
1141 and "unsetenv".
1142
Willy Tarreau636848a2019-04-15 19:38:50 +02001143set-dumpable
1144 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001145 developer's request. If it has been enabled, it may still be forcibly
1146 disabled by prefixing it with the "no" keyword. It has no impact on
1147 performance nor stability but will try hard to re-enable core dumps that were
1148 possibly disabled by file size limitations (ulimit -f), core size limitations
1149 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1150 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1151 the current directory's permissions (check what directory the file is started
1152 from), the chroot directory's permission (it may be needed to temporarily
1153 disable the chroot directive or to move it to a dedicated writable location),
1154 or any other system-specific constraint. For example, some Linux flavours are
1155 notorious for replacing the default core file with a path to an executable
1156 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1157 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1158 issue. When trying to enable this option waiting for a rare issue to
1159 re-appear, it's often a good idea to first try to obtain such a dump by
1160 issuing, for example, "kill -11" to the haproxy process and verify that it
1161 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001162
Willy Tarreau610f04b2014-02-13 11:36:41 +01001163ssl-default-bind-ciphers <ciphers>
1164 This setting is only available when support for OpenSSL was built in. It sets
1165 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001166 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001167 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001168 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1169 information and recommendations see e.g.
1170 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1171 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1172 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1173 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001174
1175ssl-default-bind-ciphersuites <ciphersuites>
1176 This setting is only available when support for OpenSSL was built in and
1177 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1178 describing the list of cipher algorithms ("cipher suite") that are negotiated
1179 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1180 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001181 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1182 cipher configuration for TLSv1.2 and earlier, please check the
1183 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001184 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001185
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001186ssl-default-bind-options [<option>]...
1187 This setting is only available when support for OpenSSL was built in. It sets
1188 default ssl-options to force on all "bind" lines. Please check the "bind"
1189 keyword to see available options.
1190
1191 Example:
1192 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001193 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001194
Willy Tarreau610f04b2014-02-13 11:36:41 +01001195ssl-default-server-ciphers <ciphers>
1196 This setting is only available when support for OpenSSL was built in. It
1197 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001198 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001199 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001200 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1201 information and recommendations see e.g.
1202 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1203 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1204 For TLSv1.3 cipher configuration, please check the
1205 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1206 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001207
1208ssl-default-server-ciphersuites <ciphersuites>
1209 This setting is only available when support for OpenSSL was built in and
1210 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1211 string describing the list of cipher algorithms that are negotiated during
1212 the TLSv1.3 handshake with the server, for all "server" lines which do not
1213 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001214 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1215 cipher configuration for TLSv1.2 and earlier, please check the
1216 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1217 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001218
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001219ssl-default-server-options [<option>]...
1220 This setting is only available when support for OpenSSL was built in. It sets
1221 default ssl-options to force on all "server" lines. Please check the "server"
1222 keyword to see available options.
1223
Remi Gacogne47783ef2015-05-29 15:53:22 +02001224ssl-dh-param-file <file>
1225 This setting is only available when support for OpenSSL was built in. It sets
1226 the default DH parameters that are used during the SSL/TLS handshake when
1227 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001228 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001229 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001230 are not specified either by using ssl-dh-param-file or by setting them
1231 directly in the certificate file, pre-generated DH parameters of the size
1232 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1233 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001234 Custom DH parameters may be generated by using the OpenSSL command
1235 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1236 parameters should not be considered secure anymore.
1237
Emeric Brun850efd52014-01-29 12:24:34 +01001238ssl-server-verify [none|required]
1239 The default behavior for SSL verify on servers side. If specified to 'none',
1240 servers certificates are not verified. The default is 'required' except if
1241 forced using cmdline option '-dV'.
1242
Willy Tarreauabb175f2012-09-24 12:43:26 +02001243stats socket [<address:port>|<path>] [param*]
1244 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1245 Connections to this socket will return various statistics outputs and even
1246 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001247 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001248 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001249
Willy Tarreauabb175f2012-09-24 12:43:26 +02001250 All parameters supported by "bind" lines are supported, for instance to
1251 restrict access to some users or their access rights. Please consult
1252 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001253
1254stats timeout <timeout, in milliseconds>
1255 The default timeout on the stats socket is set to 10 seconds. It is possible
1256 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001257 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001258
1259stats maxconn <connections>
1260 By default, the stats socket is limited to 10 concurrent connections. It is
1261 possible to change this value with "stats maxconn".
1262
Willy Tarreau6a06a402007-07-15 20:15:28 +02001263uid <number>
1264 Changes the process' user ID to <number>. It is recommended that the user ID
1265 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1266 be started with superuser privileges in order to be able to switch to another
1267 one. See also "gid" and "user".
1268
1269ulimit-n <number>
1270 Sets the maximum number of per-process file-descriptors to <number>. By
1271 default, it is automatically computed, so it is recommended not to use this
1272 option.
1273
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001274unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1275 [ group <group> ] [ gid <gid> ]
1276
1277 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1278 This is mainly used to simplify declaration of those UNIX sockets and reduce
1279 the risk of errors, since those settings are most commonly required but are
1280 also process-specific. The <prefix> setting can be used to force all socket
1281 path to be relative to that directory. This might be needed to access another
1282 component's chroot. Note that those paths are resolved before haproxy chroots
1283 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1284 all have the same meaning as their homonyms used by the "bind" statement. If
1285 both are specified, the "bind" statement has priority, meaning that the
1286 "unix-bind" settings may be seen as process-wide default settings.
1287
Willy Tarreau1d549722016-02-16 12:41:57 +01001288unsetenv [<name> ...]
1289 Removes environment variables specified in arguments. This can be useful to
1290 hide some sensitive information that are occasionally inherited from the
1291 user's environment during some operations. Variables which did not exist are
1292 silently ignored so that after the operation, it is certain that none of
1293 these variables remain. The changes immediately take effect so that the next
1294 line in the configuration file will not see these variables. See also
1295 "setenv", "presetenv", and "resetenv".
1296
Willy Tarreau6a06a402007-07-15 20:15:28 +02001297user <user name>
1298 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1299 See also "uid" and "group".
1300
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001301node <name>
1302 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1303
1304 This statement is useful in HA configurations where two or more processes or
1305 servers share the same IP address. By setting a different node-name on all
1306 nodes, it becomes easy to immediately spot what server is handling the
1307 traffic.
1308
1309description <text>
1310 Add a text that describes the instance.
1311
1312 Please note that it is required to escape certain characters (# for example)
1313 and this text is inserted into a html page so you should avoid using
1314 "<" and ">" characters.
1315
Thomas Holmesdb04f192015-05-18 13:21:39 +0100131651degrees-data-file <file path>
1317 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001318 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001319
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001320 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001321 compiled with USE_51DEGREES.
1322
Ben Shillitof25e8e52016-12-02 14:25:37 +0000132351degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001324 A list of 51Degrees property names to be load from the dataset. A full list
1325 of names is available on the 51Degrees website:
1326 https://51degrees.com/resources/property-dictionary
1327
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001328 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001329 compiled with USE_51DEGREES.
1330
Dragan Dosen93b38d92015-06-29 16:43:25 +0200133151degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001332 A char that will be appended to every property value in a response header
1333 containing 51Degrees results. If not set that will be set as ','.
1334
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001335 Please note that this option is only available when haproxy has been
1336 compiled with USE_51DEGREES.
1337
133851degrees-cache-size <number>
1339 Sets the size of the 51Degrees converter cache to <number> entries. This
1340 is an LRU cache which reminds previous device detections and their results.
1341 By default, this cache is disabled.
1342
1343 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001344 compiled with USE_51DEGREES.
1345
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001346wurfl-data-file <file path>
1347 The path of the WURFL data file to provide device detection services. The
1348 file should be accessible by HAProxy with relevant permissions.
1349
1350 Please note that this option is only available when haproxy has been compiled
1351 with USE_WURFL=1.
1352
1353wurfl-information-list [<capability>]*
1354 A space-delimited list of WURFL capabilities, virtual capabilities, property
1355 names we plan to use in injected headers. A full list of capability and
1356 virtual capability names is available on the Scientiamobile website :
1357
1358 https://www.scientiamobile.com/wurflCapability
1359
1360 Valid WURFL properties are:
1361 - wurfl_id Contains the device ID of the matched device.
1362
1363 - wurfl_root_id Contains the device root ID of the matched
1364 device.
1365
1366 - wurfl_isdevroot Tells if the matched device is a root device.
1367 Possible values are "TRUE" or "FALSE".
1368
1369 - wurfl_useragent The original useragent coming with this
1370 particular web request.
1371
1372 - wurfl_api_version Contains a string representing the currently
1373 used Libwurfl API version.
1374
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001375 - wurfl_info A string containing information on the parsed
1376 wurfl.xml and its full path.
1377
1378 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1379 WURFL has been loaded successfully.
1380
1381 - wurfl_normalized_useragent The normalized useragent.
1382
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001383 Please note that this option is only available when haproxy has been compiled
1384 with USE_WURFL=1.
1385
1386wurfl-information-list-separator <char>
1387 A char that will be used to separate values in a response header containing
1388 WURFL results. If not set that a comma (',') will be used by default.
1389
1390 Please note that this option is only available when haproxy has been compiled
1391 with USE_WURFL=1.
1392
1393wurfl-patch-file [<file path>]
1394 A list of WURFL patch file paths. Note that patches are loaded during startup
1395 thus before the chroot.
1396
1397 Please note that this option is only available when haproxy has been compiled
1398 with USE_WURFL=1.
1399
paulborilebad132c2019-04-18 11:57:04 +02001400wurfl-cache-size <size>
1401 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1402 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001403 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001404 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001405
1406 Please note that this option is only available when haproxy has been compiled
1407 with USE_WURFL=1.
1408
William Dauchy0fec3ab2019-10-27 20:08:11 +01001409strict-limits
1410 Makes process fail at startup when a setrlimit fails. Haproxy is tries to set
1411 the best setrlimit according to what has been calculated. If it fails, it
1412 will emit a warning. Use this option if you want an explicit failure of
1413 haproxy when those limits fail. This option is disabled by default. If it has
1414 been enabled, it may still be forcibly disabled by prefixing it with the "no"
1415 keyword.
1416
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014173.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001418-----------------------
1419
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001420busy-polling
1421 In some situations, especially when dealing with low latency on processors
1422 supporting a variable frequency or when running inside virtual machines, each
1423 time the process waits for an I/O using the poller, the processor goes back
1424 to sleep or is offered to another VM for a long time, and it causes
1425 excessively high latencies. This option provides a solution preventing the
1426 processor from sleeping by always using a null timeout on the pollers. This
1427 results in a significant latency reduction (30 to 100 microseconds observed)
1428 at the expense of a risk to overheat the processor. It may even be used with
1429 threads, in which case improperly bound threads may heavily conflict,
1430 resulting in a worse performance and high values for the CPU stolen fields
1431 in "show info" output, indicating which threads are misconfigured. It is
1432 important not to let the process run on the same processor as the network
1433 interrupts when this option is used. It is also better to avoid using it on
1434 multiple CPU threads sharing the same core. This option is disabled by
1435 default. If it has been enabled, it may still be forcibly disabled by
1436 prefixing it with the "no" keyword. It is ignored by the "select" and
1437 "poll" pollers.
1438
Willy Tarreau1746eec2014-04-25 10:46:47 +02001439max-spread-checks <delay in milliseconds>
1440 By default, haproxy tries to spread the start of health checks across the
1441 smallest health check interval of all the servers in a farm. The principle is
1442 to avoid hammering services running on the same server. But when using large
1443 check intervals (10 seconds or more), the last servers in the farm take some
1444 time before starting to be tested, which can be a problem. This parameter is
1445 used to enforce an upper bound on delay between the first and the last check,
1446 even if the servers' check intervals are larger. When servers run with
1447 shorter intervals, their intervals will be respected though.
1448
Willy Tarreau6a06a402007-07-15 20:15:28 +02001449maxconn <number>
1450 Sets the maximum per-process number of concurrent connections to <number>. It
1451 is equivalent to the command-line argument "-n". Proxies will stop accepting
1452 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001453 automatically adjusted according to this value. See also "ulimit-n". Note:
1454 the "select" poller cannot reliably use more than 1024 file descriptors on
1455 some platforms. If your platform only supports select and reports "select
1456 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001457 below 500 in general). If this value is not set, it will automatically be
1458 calculated based on the current file descriptors limit reported by the
1459 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1460 is enforced, based on the buffer size, memory allocated to compression, SSL
1461 cache size, and use or not of SSL and the associated maxsslconn (which can
1462 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001463
Willy Tarreau81c25d02011-09-07 15:17:21 +02001464maxconnrate <number>
1465 Sets the maximum per-process number of connections per second to <number>.
1466 Proxies will stop accepting connections when this limit is reached. It can be
1467 used to limit the global capacity regardless of each frontend capacity. It is
1468 important to note that this can only be used as a service protection measure,
1469 as there will not necessarily be a fair share between frontends when the
1470 limit is reached, so it's a good idea to also limit each frontend to some
1471 value close to its expected share. Also, lowering tune.maxaccept can improve
1472 fairness.
1473
William Lallemandd85f9172012-11-09 17:05:39 +01001474maxcomprate <number>
1475 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001476 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001477 level will be decreased during the session. If the maximum is reached at the
1478 beginning of a session, the session will not compress at all. If the maximum
1479 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001480 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001481 default value.
1482
William Lallemand072a2bf2012-11-20 17:01:01 +01001483maxcompcpuusage <number>
1484 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1485 for new requests or decreasing the compression level of current requests.
1486 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1487 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1488 case of multiple processes (nbproc > 1), each process manages its individual
1489 usage. A value of 100 disable the limit. The default value is 100. Setting
1490 a lower value will prevent the compression work from slowing the whole
1491 process down and from introducing high latencies.
1492
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001493maxpipes <number>
1494 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1495 are only used by kernel-based tcp splicing. Since a pipe contains two file
1496 descriptors, the "ulimit-n" value will be increased accordingly. The default
1497 value is maxconn/4, which seems to be more than enough for most heavy usages.
1498 The splice code dynamically allocates and releases pipes, and can fall back
1499 to standard copy, so setting this value too low may only impact performance.
1500
Willy Tarreau93e7c002013-10-07 18:51:07 +02001501maxsessrate <number>
1502 Sets the maximum per-process number of sessions per second to <number>.
1503 Proxies will stop accepting connections when this limit is reached. It can be
1504 used to limit the global capacity regardless of each frontend capacity. It is
1505 important to note that this can only be used as a service protection measure,
1506 as there will not necessarily be a fair share between frontends when the
1507 limit is reached, so it's a good idea to also limit each frontend to some
1508 value close to its expected share. Also, lowering tune.maxaccept can improve
1509 fairness.
1510
Willy Tarreau403edff2012-09-06 11:58:37 +02001511maxsslconn <number>
1512 Sets the maximum per-process number of concurrent SSL connections to
1513 <number>. By default there is no SSL-specific limit, which means that the
1514 global maxconn setting will apply to all connections. Setting this limit
1515 avoids having openssl use too much memory and crash when malloc returns NULL
1516 (since it unfortunately does not reliably check for such conditions). Note
1517 that the limit applies both to incoming and outgoing connections, so one
1518 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001519 If this value is not set, but a memory limit is enforced, this value will be
1520 automatically computed based on the memory limit, maxconn, the buffer size,
1521 memory allocated to compression, SSL cache size, and use of SSL in either
1522 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1523 when there is a memory limit, haproxy will automatically adjust these values
1524 so that 100% of the connections can be made over SSL with no risk, and will
1525 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001526
Willy Tarreaue43d5322013-10-07 20:01:52 +02001527maxsslrate <number>
1528 Sets the maximum per-process number of SSL sessions per second to <number>.
1529 SSL listeners will stop accepting connections when this limit is reached. It
1530 can be used to limit the global SSL CPU usage regardless of each frontend
1531 capacity. It is important to note that this can only be used as a service
1532 protection measure, as there will not necessarily be a fair share between
1533 frontends when the limit is reached, so it's a good idea to also limit each
1534 frontend to some value close to its expected share. It is also important to
1535 note that the sessions are accounted before they enter the SSL stack and not
1536 after, which also protects the stack against bad handshakes. Also, lowering
1537 tune.maxaccept can improve fairness.
1538
William Lallemand9d5f5482012-11-07 16:12:57 +01001539maxzlibmem <number>
1540 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1541 When the maximum amount is reached, future sessions will not compress as long
1542 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001543 The default value is 0. The value is available in bytes on the UNIX socket
1544 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1545 "ZlibMemUsage" in bytes.
1546
Willy Tarreau6a06a402007-07-15 20:15:28 +02001547noepoll
1548 Disables the use of the "epoll" event polling system on Linux. It is
1549 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001550 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001551
1552nokqueue
1553 Disables the use of the "kqueue" event polling system on BSD. It is
1554 equivalent to the command-line argument "-dk". The next polling system
1555 used will generally be "poll". See also "nopoll".
1556
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001557noevports
1558 Disables the use of the event ports event polling system on SunOS systems
1559 derived from Solaris 10 and later. It is equivalent to the command-line
1560 argument "-dv". The next polling system used will generally be "poll". See
1561 also "nopoll".
1562
Willy Tarreau6a06a402007-07-15 20:15:28 +02001563nopoll
1564 Disables the use of the "poll" event polling system. It is equivalent to the
1565 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001566 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001567 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1568 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001569
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001570nosplice
1571 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001572 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001573 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001574 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001575 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1576 be used. This option makes it easier to globally disable kernel splicing in
1577 case of doubt. See also "option splice-auto", "option splice-request" and
1578 "option splice-response".
1579
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001580nogetaddrinfo
1581 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1582 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1583
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001584noreuseport
1585 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1586 command line argument "-dR".
1587
Willy Tarreaud2d33482019-04-25 17:09:07 +02001588profiling.tasks { auto | on | off }
1589 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1590 the profiling automatically turns on a thread when it starts to suffer from
1591 an average latency of 1000 microseconds or higher as reported in the
1592 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05001593 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001594 loops so it does not vary quickly and tends to significantly smooth short
1595 spikes). It may also spontaneously trigger from time to time on overloaded
1596 systems, containers, or virtual machines, or when the system swaps (which
1597 must absolutely never happen on a load balancer).
1598
1599 CPU profiling per task can be very convenient to report where the time is
1600 spent and which requests have what effect on which other request. Enabling
1601 it will typically affect the overall's performance by less than 1%, thus it
1602 is recommended to leave it to the default 'auto' value so that it only
1603 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001604 supporting the clock_gettime(2) syscall with clock identifiers
1605 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1606 be zero. This option may be changed at run time using "set profiling" on the
1607 CLI.
1608
Willy Tarreaufe255b72007-10-14 23:09:26 +02001609spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001610 Sometimes it is desirable to avoid sending agent and health checks to
1611 servers at exact intervals, for instance when many logical servers are
1612 located on the same physical server. With the help of this parameter, it
1613 becomes possible to add some randomness in the check interval between 0
1614 and +/- 50%. A value between 2 and 5 seems to show good results. The
1615 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001616
Davor Ocelice9ed2812017-12-25 17:49:28 +01001617ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001618 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001619 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001620 multiple times, it will simply enable multiple crypto engines. Referencing an
1621 unsupported engine will prevent haproxy from starting. Note that many engines
1622 will lead to lower HTTPS performance than pure software with recent
1623 processors. The optional command "algo" sets the default algorithms an ENGINE
1624 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001625 of "ALL" uses the engine for all cryptographic operations. If no list of
1626 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001627 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1628 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1629 openssl configuration file uses:
1630 https://www.openssl.org/docs/man1.0.2/apps/config.html
1631
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001632ssl-mode-async
1633 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001634 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001635 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1636 doesn't support moving read/write buffers and is not compliant with
1637 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05001638 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001639 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001640
Willy Tarreau33cb0652014-12-23 22:52:37 +01001641tune.buffers.limit <number>
1642 Sets a hard limit on the number of buffers which may be allocated per process.
1643 The default value is zero which means unlimited. The minimum non-zero value
1644 will always be greater than "tune.buffers.reserve" and should ideally always
1645 be about twice as large. Forcing this value can be particularly useful to
1646 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001647 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001648 another one to be released by another session. Since buffers are dynamically
1649 allocated and released, the waiting time is very short and not perceptible
1650 provided that limits remain reasonable. In fact sometimes reducing the limit
1651 may even increase performance by increasing the CPU cache's efficiency. Tests
1652 have shown good results on average HTTP traffic with a limit to 1/10 of the
1653 expected global maxconn setting, which also significantly reduces memory
1654 usage. The memory savings come from the fact that a number of connections
1655 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1656 advised to do so by an haproxy core developer.
1657
Willy Tarreau1058ae72014-12-23 22:40:40 +01001658tune.buffers.reserve <number>
1659 Sets the number of buffers which are pre-allocated and reserved for use only
1660 during memory shortage conditions resulting in failed memory allocations. The
1661 minimum value is 2 and is also the default. There is no reason a user would
1662 want to change this value, it's mostly aimed at haproxy core developers.
1663
Willy Tarreau27a674e2009-08-17 07:23:33 +02001664tune.bufsize <number>
1665 Sets the buffer size to this size (in bytes). Lower values allow more
1666 sessions to coexist in the same amount of RAM, and higher values allow some
1667 applications with very large cookies to work. The default value is 16384 and
1668 can be changed at build time. It is strongly recommended not to change this
1669 from the default value, as very low values will break some services such as
1670 statistics, and values larger than default size will increase memory usage,
1671 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001672 parameter should be decreased by the same factor as this one is increased. In
1673 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1674 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001675 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001676 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1677 value set using this parameter will automatically be rounded up to the next
1678 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001679
Willy Tarreau43961d52010-10-04 20:39:20 +02001680tune.chksize <number>
1681 Sets the check buffer size to this size (in bytes). Higher values may help
1682 find string or regex patterns in very large pages, though doing so may imply
1683 more memory and CPU usage. The default value is 16384 and can be changed at
1684 build time. It is not recommended to change this value, but to use better
1685 checks whenever possible.
1686
William Lallemandf3747832012-11-09 12:33:10 +01001687tune.comp.maxlevel <number>
1688 Sets the maximum compression level. The compression level affects CPU
1689 usage during compression. This value affects CPU usage during compression.
1690 Each session using compression initializes the compression algorithm with
1691 this value. The default value is 1.
1692
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001693tune.fail-alloc
1694 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1695 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1696 success). This is useful to debug and make sure memory failures are handled
1697 gracefully.
1698
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001699tune.h2.header-table-size <number>
1700 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1701 cannot be larger than 65536 bytes. A larger value may help certain clients
1702 send more compact requests, depending on their capabilities. This amount of
1703 memory is consumed for each HTTP/2 connection. It is recommended not to
1704 change it.
1705
Willy Tarreaue6baec02017-07-27 11:45:11 +02001706tune.h2.initial-window-size <number>
1707 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001708 can upload before waiting for an acknowledgment from haproxy. This setting
1709 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001710 The default value is 65535, which roughly allows up to 5 Mbps of upload
1711 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1712 over a 1-ms local network. It can make sense to increase this value to allow
1713 faster uploads, or to reduce it to increase fairness when dealing with many
1714 clients. It doesn't affect resource usage.
1715
Willy Tarreau5242ef82017-07-27 11:47:28 +02001716tune.h2.max-concurrent-streams <number>
1717 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1718 number of outstanding requests on a single connection). The default value is
1719 100. A larger one may slightly improve page load time for complex sites when
1720 visited over high latency networks, but increases the amount of resources a
1721 single client may allocate. A value of zero disables the limit so a single
1722 client may create as many streams as allocatable by haproxy. It is highly
1723 recommended not to change this value.
1724
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001725tune.h2.max-frame-size <number>
1726 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1727 receive to its peers. The default value is the largest between 16384 and the
1728 buffer size (tune.bufsize). In any case, haproxy will not announce support
1729 for frame sizes larger than buffers. The main purpose of this setting is to
1730 allow to limit the maximum frame size setting when using large buffers. Too
1731 large frame sizes might have performance impact or cause some peers to
1732 misbehave. It is highly recommended not to change this value.
1733
Willy Tarreau193b8c62012-11-22 00:17:38 +01001734tune.http.cookielen <number>
1735 Sets the maximum length of captured cookies. This is the maximum value that
1736 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1737 will automatically be truncated to this one. It is important not to set too
1738 high a value because all cookie captures still allocate this size whatever
1739 their configured value (they share a same pool). This value is per request
1740 per response, so the memory allocated is twice this value per connection.
1741 When not specified, the limit is set to 63 characters. It is recommended not
1742 to change this value.
1743
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001744tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001745 Sets the maximum length of request URI in logs. This prevents truncating long
1746 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001747 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001748 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001749 configuration directives too.
1750 The default value is 1024.
1751
Willy Tarreauac1932d2011-10-24 19:14:41 +02001752tune.http.maxhdr <number>
1753 Sets the maximum number of headers in a request. When a request comes with a
1754 number of headers greater than this value (including the first line), it is
1755 rejected with a "400 Bad Request" status code. Similarly, too large responses
1756 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1757 for all usages, considering that the widely deployed Apache server uses the
1758 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001759 a buggy application to work by the time it gets fixed. The accepted range is
1760 1..32767. Keep in mind that each new header consumes 32bits of memory for
1761 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001762
Willy Tarreau7e312732014-02-12 16:35:14 +01001763tune.idletimer <timeout>
1764 Sets the duration after which haproxy will consider that an empty buffer is
1765 probably associated with an idle stream. This is used to optimally adjust
1766 some packet sizes while forwarding large and small data alternatively. The
1767 decision to use splice() or to send large buffers in SSL is modulated by this
1768 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1769 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001770 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05001771 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01001772 tune.ssl.maxrecord below.
1773
Willy Tarreau7ac908b2019-02-27 12:02:18 +01001774tune.listener.multi-queue { on | off }
1775 Enables ('on') or disables ('off') the listener's multi-queue accept which
1776 spreads the incoming traffic to all threads a "bind" line is allowed to run
1777 on instead of taking them for itself. This provides a smoother traffic
1778 distribution and scales much better, especially in environments where threads
1779 may be unevenly loaded due to external activity (network interrupts colliding
1780 with one thread for example). This option is enabled by default, but it may
1781 be forcefully disabled for troubleshooting or for situations where it is
1782 estimated that the operating system already provides a good enough
1783 distribution and connections are extremely short-lived.
1784
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001785tune.lua.forced-yield <number>
1786 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001787 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001788 HAProxy scheduler to process other tasks like accepting connections or
1789 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001790 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001791 lowered. If the Lua code is quite long and its result is absolutely required
1792 to process the data, the <number> can be increased.
1793
Willy Tarreau32f61e22015-03-18 17:54:59 +01001794tune.lua.maxmem
1795 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1796 default it is zero which means unlimited. It is important to set a limit to
1797 ensure that a bug in a script will not result in the system running out of
1798 memory.
1799
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001800tune.lua.session-timeout <timeout>
1801 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001802 preventing infinite loops or spending too much time in Lua. This timeout
1803 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001804 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001805
1806tune.lua.task-timeout <timeout>
1807 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1808 dedicated to the tasks. By default, this timeout isn't set because a task may
1809 remain alive during of the lifetime of HAProxy. For example, a task used to
1810 check servers.
1811
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001812tune.lua.service-timeout <timeout>
1813 This is the execution timeout for the Lua services. This is useful for
1814 preventing infinite loops or spending too much time in Lua. This timeout
1815 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001816 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001817
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001818tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001819 Sets the maximum number of consecutive connections a process may accept in a
1820 row before switching to other work. In single process mode, higher numbers
1821 give better performance at high connection rates. However in multi-process
1822 modes, keeping a bit of fairness between processes generally is better to
1823 increase performance. This value applies individually to each listener, so
1824 that the number of processes a listener is bound to is taken into account.
1825 This value defaults to 64. In multi-process mode, it is divided by twice
1826 the number of processes the listener is bound to. Setting this value to -1
1827 completely disables the limitation. It should normally not be needed to tweak
1828 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001829
1830tune.maxpollevents <number>
1831 Sets the maximum amount of events that can be processed at once in a call to
1832 the polling system. The default value is adapted to the operating system. It
1833 has been noticed that reducing it below 200 tends to slightly decrease
1834 latency at the expense of network bandwidth, and increasing it above 200
1835 tends to trade latency for slightly increased bandwidth.
1836
Willy Tarreau27a674e2009-08-17 07:23:33 +02001837tune.maxrewrite <number>
1838 Sets the reserved buffer space to this size in bytes. The reserved space is
1839 used for header rewriting or appending. The first reads on sockets will never
1840 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1841 bufsize, though that does not make much sense since there are rarely large
1842 numbers of headers to add. Setting it too high prevents processing of large
1843 requests or responses. Setting it too low prevents addition of new headers
1844 to already large requests or to POST requests. It is generally wise to set it
1845 to about 1024. It is automatically readjusted to half of bufsize if it is
1846 larger than that. This means you don't have to worry about it when changing
1847 bufsize.
1848
Willy Tarreauf3045d22015-04-29 16:24:50 +02001849tune.pattern.cache-size <number>
1850 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1851 cache which reminds previous lookups and their results. It is used by ACLs
1852 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1853 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1854 strings. It applies to pattern expressions which means that it will be able
1855 to memorize the result of a lookup among all the patterns specified on a
1856 configuration line (including all those loaded from files). It automatically
1857 invalidates entries which are updated using HTTP actions or on the CLI. The
1858 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02001859 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
1860 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02001861 risk of collision in this cache, which is in the order of the size of the
1862 cache divided by 2^64. Typically, at 10000 requests per second with the
1863 default cache size of 10000 entries, there's 1% chance that a brute force
1864 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1865 This is considered much lower than the risk of a memory corruption caused by
1866 aging components. If this is not acceptable, the cache can be disabled by
1867 setting this parameter to 0.
1868
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001869tune.pipesize <number>
1870 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1871 are the default size for the system. But sometimes when using TCP splicing,
1872 it can improve performance to increase pipe sizes, especially if it is
1873 suspected that pipes are not filled and that many calls to splice() are
1874 performed. This has an impact on the kernel's memory footprint, so this must
1875 not be changed if impacts are not understood.
1876
Olivier Houchard88698d92019-04-16 19:07:22 +02001877tune.pool-low-fd-ratio <number>
1878 This setting sets the max number of file descriptors (in percentage) used by
1879 haproxy globally against the maximum number of file descriptors haproxy can
1880 use before we stop putting connection into the idle pool for reuse. The
1881 default is 20.
1882
1883tune.pool-high-fd-ratio <number>
1884 This setting sets the max number of file descriptors (in percentage) used by
1885 haproxy globally against the maximum number of file descriptors haproxy can
1886 use before we start killing idle connections when we can't reuse a connection
1887 and we have to create a new one. The default is 25 (one quarter of the file
1888 descriptor will mean that roughly half of the maximum front connections can
1889 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05001890 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02001891
Willy Tarreaue803de22010-01-21 17:43:04 +01001892tune.rcvbuf.client <number>
1893tune.rcvbuf.server <number>
1894 Forces the kernel socket receive buffer size on the client or the server side
1895 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1896 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05001897 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001898 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001899 order to save kernel memory by preventing it from buffering too large amounts
1900 of received data. Lower values will significantly increase CPU usage though.
1901
Willy Tarreaub22fc302015-12-14 12:04:35 +01001902tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001903 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001904 socket buffers. One of them is that a read returns more than <recv_enough>
1905 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1906 may be changed by this setting to better deal with workloads involving lots
1907 of short messages such as telnet or SSH sessions.
1908
Olivier Houchard1599b802018-05-24 18:59:04 +02001909tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05001910 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02001911 tasks. The default value is 200. Increasing it may incur latency when
1912 dealing with I/Os, making it too small can incur extra overhead.
1913
Willy Tarreaue803de22010-01-21 17:43:04 +01001914tune.sndbuf.client <number>
1915tune.sndbuf.server <number>
1916 Forces the kernel socket send buffer size on the client or the server side to
1917 the specified value in bytes. This value applies to all TCP/HTTP frontends
1918 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05001919 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001920 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001921 order to save kernel memory by preventing it from buffering too large amounts
1922 of received data. Lower values will significantly increase CPU usage though.
1923 Another use case is to prevent write timeouts with extremely slow clients due
1924 to the kernel waiting for a large part of the buffer to be read before
1925 notifying haproxy again.
1926
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001927tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001928 Sets the size of the global SSL session cache, in a number of blocks. A block
1929 is large enough to contain an encoded session without peer certificate.
1930 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001931 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001932 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001933 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001934 and reassigned. Higher values reduce the occurrence of such a purge, hence
1935 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1936 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001937 and are shared between all processes if "nbproc" is greater than 1. Setting
1938 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001939
Emeric Brun8dc60392014-05-09 13:52:00 +02001940tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001941 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001942 should normally not be used since it will force many renegotiations due to
1943 clients hitting a random process. But it may be required on some operating
1944 systems where none of the SSL cache synchronization method may be used. In
1945 this case, adding a first layer of hash-based load balancing before the SSL
1946 layer might limit the impact of the lack of session sharing.
1947
Emeric Brun4f65bff2012-11-16 15:11:00 +01001948tune.ssl.lifetime <timeout>
1949 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001950 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001951 does not guarantee that sessions will last that long, because if the cache is
1952 full, the longest idle sessions will be purged despite their configured
1953 lifetime. The real usefulness of this setting is to prevent sessions from
1954 being used for too long.
1955
Willy Tarreaubfd59462013-02-21 07:46:09 +01001956tune.ssl.maxrecord <number>
1957 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1958 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1959 data only once it has received a full record. With large records, it means
1960 that clients might have to download up to 16kB of data before starting to
1961 process them. Limiting the value can improve page load times on browsers
1962 located over high latency or low bandwidth networks. It is suggested to find
1963 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1964 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1965 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1966 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001967 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001968 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001969
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001970tune.ssl.default-dh-param <number>
1971 Sets the maximum size of the Diffie-Hellman parameters used for generating
1972 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1973 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1974 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1975 this maximum value. Default value if 1024. Only 1024 or higher values are
1976 allowed. Higher values will increase the CPU load, and values greater than
1977 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001978 used if static Diffie-Hellman parameters are supplied either directly
1979 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001980
Christopher Faulet31af49d2015-06-09 17:29:50 +02001981tune.ssl.ssl-ctx-cache-size <number>
1982 Sets the size of the cache used to store generated certificates to <number>
1983 entries. This is a LRU cache. Because generating a SSL certificate
1984 dynamically is expensive, they are cached. The default cache size is set to
1985 1000 entries.
1986
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001987tune.ssl.capture-cipherlist-size <number>
1988 Sets the maximum size of the buffer used for capturing client-hello cipher
1989 list. If the value is 0 (default value) the capture is disabled, otherwise
1990 a buffer is allocated for each SSL/TLS connection.
1991
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001992tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001993tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001994tune.vars.reqres-max-size <size>
1995tune.vars.sess-max-size <size>
1996tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001997 These five tunes help to manage the maximum amount of memory used by the
1998 variables system. "global" limits the overall amount of memory available for
1999 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2000 memory for the session scope, "txn" for the transaction scope, and "reqres"
2001 limits the memory for each request or response processing.
2002 Memory accounting is hierarchical, meaning more coarse grained limits include
2003 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2004 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002005
Daniel Schneller0b547052016-03-21 20:46:57 +01002006 For example, when "tune.vars.sess-max-size" is limited to 100,
2007 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2008 100 either. If we create a variable "txn.var" that contains 100 bytes,
2009 all available space is consumed.
2010 Notice that exceeding the limits at runtime will not result in an error
2011 message, but values might be cut off or corrupted. So make sure to accurately
2012 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002013
William Lallemanda509e4c2012-11-07 16:54:34 +01002014tune.zlib.memlevel <number>
2015 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002016 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002017 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002018 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002019 between 1 and 9. The default value is 8.
2020
2021tune.zlib.windowsize <number>
2022 Sets the window size (the size of the history buffer) as a parameter of the
2023 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002024 in better compression at the expense of memory usage. Can be a value between
2025 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002026
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020273.3. Debugging
2028--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002029
2030debug
2031 Enables debug mode which dumps to stdout all exchanges, and disables forking
2032 into background. It is the equivalent of the command-line argument "-d". It
2033 should never be used in a production configuration since it may prevent full
2034 system startup.
2035
2036quiet
2037 Do not display any message during startup. It is equivalent to the command-
2038 line argument "-q".
2039
Emeric Brunf099e792010-09-27 12:05:28 +02002040
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010020413.4. Userlists
2042--------------
2043It is possible to control access to frontend/backend/listen sections or to
2044http stats by allowing only authenticated and authorized users. To do this,
2045it is required to create at least one userlist and to define users.
2046
2047userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002048 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002049 used to store authentication & authorization data for independent customers.
2050
2051group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002052 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002053 attach users to this group by using a comma separated list of names
2054 proceeded by "users" keyword.
2055
Cyril Bontéf0c60612010-02-06 14:44:47 +01002056user <username> [password|insecure-password <password>]
2057 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002058 Adds user <username> to the current userlist. Both secure (encrypted) and
2059 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002060 evaluated using the crypt(3) function, so depending on the system's
2061 capabilities, different algorithms are supported. For example, modern Glibc
2062 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2063 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002064
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002065 Attention: Be aware that using encrypted passwords might cause significantly
2066 increased CPU usage, depending on the number of requests, and the algorithm
2067 used. For any of the hashed variants, the password for each request must
2068 be processed through the chosen algorithm, before it can be compared to the
2069 value specified in the config file. Most current algorithms are deliberately
2070 designed to be expensive to compute to achieve resistance against brute
2071 force attacks. They do not simply salt/hash the clear text password once,
2072 but thousands of times. This can quickly become a major factor in haproxy's
2073 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002074
2075 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002076 userlist L1
2077 group G1 users tiger,scott
2078 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002079
Cyril Bontéf0c60612010-02-06 14:44:47 +01002080 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2081 user scott insecure-password elgato
2082 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002083
Cyril Bontéf0c60612010-02-06 14:44:47 +01002084 userlist L2
2085 group G1
2086 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002087
Cyril Bontéf0c60612010-02-06 14:44:47 +01002088 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2089 user scott insecure-password elgato groups G1,G2
2090 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002091
2092 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002093
Emeric Brunf099e792010-09-27 12:05:28 +02002094
20953.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002096----------
Emeric Brun94900952015-06-11 18:25:54 +02002097It is possible to propagate entries of any data-types in stick-tables between
2098several haproxy instances over TCP connections in a multi-master fashion. Each
2099instance pushes its local updates and insertions to remote peers. The pushed
2100values overwrite remote ones without aggregation. Interrupted exchanges are
2101automatically detected and recovered from the last known point.
2102In addition, during a soft restart, the old process connects to the new one
2103using such a TCP connection to push all its entries before the new process
2104tries to connect to other peers. That ensures very fast replication during a
2105reload, it typically takes a fraction of a second even for large tables.
2106Note that Server IDs are used to identify servers remotely, so it is important
2107that configurations look similar or at least that the same IDs are forced on
2108each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002109
2110peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002111 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002112 which is referenced by one or more stick-tables.
2113
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002114bind [<address>]:<port_range> [, ...] [param*]
2115 Defines the binding parameters of the local peer of this "peers" section.
2116 Such lines are not supported with "peer" line in the same "peers" section.
2117
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002118disabled
2119 Disables a peers section. It disables both listening and any synchronization
2120 related to this section. This is provided to disable synchronization of stick
2121 tables without having to comment out all "peers" references.
2122
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002123default-bind [param*]
2124 Defines the binding parameters for the local peer, excepted its address.
2125
2126default-server [param*]
2127 Change default options for a server in a "peers" section.
2128
2129 Arguments:
2130 <param*> is a list of parameters for this server. The "default-server"
2131 keyword accepts an important number of options and has a complete
2132 section dedicated to it. Please refer to section 5 for more
2133 details.
2134
2135
2136 See also: "server" and section 5 about server options
2137
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002138enable
2139 This re-enables a disabled peers section which was previously disabled.
2140
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002141peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002142 Defines a peer inside a peers section.
2143 If <peername> is set to the local peer name (by default hostname, or forced
2144 using "-L" command line option), haproxy will listen for incoming remote peer
2145 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
2146 to join the remote peer, and <peername> is used at the protocol level to
2147 identify and validate the remote peer on the server side.
2148
2149 During a soft restart, local peer <ip>:<port> is used by the old instance to
2150 connect the new one and initiate a complete replication (teaching process).
2151
2152 It is strongly recommended to have the exact same peers declaration on all
2153 peers and to only rely on the "-L" command line argument to change the local
2154 peer name. This makes it easier to maintain coherent configuration files
2155 across all peers.
2156
William Lallemandb2f07452015-05-12 14:27:13 +02002157 You may want to reference some environment variables in the address
2158 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002159
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002160 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2161 "server" keyword explanation below).
2162
2163server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002164 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002165 with a support for all "server" parameters found in 5.2 paragraph.
2166 If the underlying peer is local, <ip>:<port> parameters must not be present.
2167 These parameters must be provided on a "bind" line (see "bind" keyword
2168 of this "peers" section).
2169 Some of these parameters are irrelevant for "peers" sections.
2170
2171
Cyril Bontédc4d9032012-04-08 21:57:39 +02002172 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002173 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002174 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002175 peer haproxy1 192.168.0.1:1024
2176 peer haproxy2 192.168.0.2:1024
2177 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002178
2179 backend mybackend
2180 mode tcp
2181 balance roundrobin
2182 stick-table type ip size 20k peers mypeers
2183 stick on src
2184
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002185 server srv1 192.168.0.30:80
2186 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002187
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002188 Example:
2189 peers mypeers
2190 bind 127.0.0.11:10001 ssl crt mycerts/pem
2191 default-server ssl verify none
2192 server hostA 127.0.0.10:10000
2193 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002194
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002195
2196table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2197 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2198
2199 Configure a stickiness table for the current section. This line is parsed
2200 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002201 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002202 mandatory first parameter to designate the stick-table. Contrary to others
2203 sections, there may be several "table" lines in "peers" sections (see also
2204 "stick-table" keyword).
2205
2206 Also be aware of the fact that "peers" sections have their own stick-table
2207 namespaces to avoid collisions between stick-table names identical in
2208 different "peers" section. This is internally handled prepending the "peers"
2209 sections names to the name of the stick-tables followed by a '/' character.
2210 If somewhere else in the configuration file you have to refer to such
2211 stick-tables declared in "peers" sections you must use the prefixed version
2212 of the stick-table name as follows:
2213
2214 peers mypeers
2215 peer A ...
2216 peer B ...
2217 table t1 ...
2218
2219 frontend fe1
2220 tcp-request content track-sc0 src table mypeers/t1
2221
2222 This is also this prefixed version of the stick-table names which must be
2223 used to refer to stick-tables through the CLI.
2224
2225 About "peers" protocol, as only "peers" belonging to the same section may
2226 communicate with each others, there is no need to do such a distinction.
2227 Several "peers" sections may declare stick-tables with the same name.
2228 This is shorter version of the stick-table name which is sent over the network.
2229 There is only a '/' character as prefix to avoid stick-table name collisions between
2230 stick-tables declared as backends and stick-table declared in "peers" sections
2231 as follows in this weird but supported configuration:
2232
2233 peers mypeers
2234 peer A ...
2235 peer B ...
2236 table t1 type string size 10m store gpc0
2237
2238 backend t1
2239 stick-table type string size 10m store gpc0 peers mypeers
2240
2241 Here "t1" table declared in "mypeeers" section has "mypeers/t1" as global name.
2242 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2243 level the former table is named "/t1", the latter is again named "t1".
2244
Simon Horman51a1cf62015-02-03 13:00:44 +090022453.6. Mailers
2246------------
2247It is possible to send email alerts when the state of servers changes.
2248If configured email alerts are sent to each mailer that is configured
2249in a mailers section. Email is sent to mailers using SMTP.
2250
Pieter Baauw386a1272015-08-16 15:26:24 +02002251mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002252 Creates a new mailer list with the name <mailersect>. It is an
2253 independent section which is referenced by one or more proxies.
2254
2255mailer <mailername> <ip>:<port>
2256 Defines a mailer inside a mailers section.
2257
2258 Example:
2259 mailers mymailers
2260 mailer smtp1 192.168.0.1:587
2261 mailer smtp2 192.168.0.2:587
2262
2263 backend mybackend
2264 mode tcp
2265 balance roundrobin
2266
2267 email-alert mailers mymailers
2268 email-alert from test1@horms.org
2269 email-alert to test2@horms.org
2270
2271 server srv1 192.168.0.30:80
2272 server srv2 192.168.0.31:80
2273
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002274timeout mail <time>
2275 Defines the time available for a mail/connection to be made and send to
2276 the mail-server. If not defined the default value is 10 seconds. To allow
2277 for at least two SYN-ACK packets to be send during initial TCP handshake it
2278 is advised to keep this value above 4 seconds.
2279
2280 Example:
2281 mailers mymailers
2282 timeout mail 20s
2283 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002284
William Lallemandc9515522019-06-12 16:32:11 +020022853.7. Programs
2286-------------
2287In master-worker mode, it is possible to launch external binaries with the
2288master, these processes are called programs. These programs are launched and
2289managed the same way as the workers.
2290
2291During a reload of HAProxy, those processes are dealing with the same
2292sequence as a worker:
2293
2294 - the master is re-executed
2295 - the master sends a SIGUSR1 signal to the program
2296 - if "option start-on-reload" is not disabled, the master launches a new
2297 instance of the program
2298
2299During a stop, or restart, a SIGTERM is sent to the programs.
2300
2301program <name>
2302 This is a new program section, this section will create an instance <name>
2303 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2304 the management guide).
2305
2306command <command> [arguments*]
2307 Define the command to start with optional arguments. The command is looked
2308 up in the current PATH if it does not include an absolute path. This is a
2309 mandatory option of the program section. Arguments containing spaces must
2310 be enclosed in quotes or double quotes or be prefixed by a backslash.
2311
Andrew Heberle97236962019-07-12 11:50:26 +08002312user <user name>
2313 Changes the executed command user ID to the <user name> from /etc/passwd.
2314 See also "group".
2315
2316group <group name>
2317 Changes the executed command group ID to the <group name> from /etc/group.
2318 See also "user".
2319
William Lallemandc9515522019-06-12 16:32:11 +02002320option start-on-reload
2321no option start-on-reload
2322 Start (or not) a new instance of the program upon a reload of the master.
2323 The default is to start a new instance. This option may only be used in a
2324 program section.
2325
2326
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023274. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002328----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002329
Willy Tarreau6a06a402007-07-15 20:15:28 +02002330Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002331 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002332 - frontend <name>
2333 - backend <name>
2334 - listen <name>
2335
2336A "defaults" section sets default parameters for all other sections following
2337its declaration. Those default parameters are reset by the next "defaults"
2338section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002339section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002340
2341A "frontend" section describes a set of listening sockets accepting client
2342connections.
2343
2344A "backend" section describes a set of servers to which the proxy will connect
2345to forward incoming connections.
2346
2347A "listen" section defines a complete proxy with its frontend and backend
2348parts combined in one section. It is generally useful for TCP-only traffic.
2349
Willy Tarreau0ba27502007-12-24 16:55:16 +01002350All proxy names must be formed from upper and lower case letters, digits,
2351'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2352case-sensitive, which means that "www" and "WWW" are two different proxies.
2353
2354Historically, all proxy names could overlap, it just caused troubles in the
2355logs. Since the introduction of content switching, it is mandatory that two
2356proxies with overlapping capabilities (frontend/backend) have different names.
2357However, it is still permitted that a frontend and a backend share the same
2358name, as this configuration seems to be commonly encountered.
2359
2360Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2361and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002362bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002363protocol, and can interact with it by allowing, blocking, switching, adding,
2364modifying, or removing arbitrary contents in requests or responses, based on
2365arbitrary criteria.
2366
Willy Tarreau70dffda2014-01-30 03:07:23 +01002367In HTTP mode, the processing applied to requests and responses flowing over
2368a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02002369the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002370
2371 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2372 requests and responses are processed, and connections remain open but idle
2373 between responses and new requests.
2374
Willy Tarreau70dffda2014-01-30 03:07:23 +01002375 - SCL: server close ("option http-server-close") : the server-facing
2376 connection is closed after the end of the response is received, but the
2377 client-facing connection remains open.
2378
Christopher Faulet315b39c2018-09-21 16:26:19 +02002379 - CLO: close ("option httpclose"): the connection is closed after the end of
2380 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002381
2382The effective mode that will be applied to a connection passing through a
2383frontend and a backend can be determined by both proxy modes according to the
2384following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002385weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002386
Christopher Faulet315b39c2018-09-21 16:26:19 +02002387 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002388
Christopher Faulet315b39c2018-09-21 16:26:19 +02002389 | KAL | SCL | CLO
2390 ----+-----+-----+----
2391 KAL | KAL | SCL | CLO
2392 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02002393 mode SCL | SCL | SCL | CLO
2394 ----+-----+-----+----
2395 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002396
Willy Tarreau0ba27502007-12-24 16:55:16 +01002397
Willy Tarreau70dffda2014-01-30 03:07:23 +01002398
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023994.1. Proxy keywords matrix
2400--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002401
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002402The following list of keywords is supported. Most of them may only be used in a
2403limited set of section types. Some of them are marked as "deprecated" because
2404they are inherited from an old syntax which may be confusing or functionally
2405limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002406marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002407option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002408and must be disabled for a specific instance. Such options may also be prefixed
2409with "default" in order to restore default settings regardless of what has been
2410specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002411
Willy Tarreau6a06a402007-07-15 20:15:28 +02002412
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002413 keyword defaults frontend listen backend
2414------------------------------------+----------+----------+---------+---------
2415acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002416backlog X X X -
2417balance X - X X
2418bind - X X -
2419bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002420capture cookie - X X -
2421capture request header - X X -
2422capture response header - X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002423compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002424cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002425declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002426default-server X - X X
2427default_backend X X X -
2428description - X X X
2429disabled X X X X
2430dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002431email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002432email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002433email-alert mailers X X X X
2434email-alert myhostname X X X X
2435email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002436enabled X X X X
2437errorfile X X X X
2438errorloc X X X X
2439errorloc302 X X X X
2440-- keyword -------------------------- defaults - frontend - listen -- backend -
2441errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002442force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002443filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002444fullconn X - X X
2445grace X X X X
2446hash-type X - X X
2447http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002448http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002449http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002450http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002451http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002452http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002453http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002454id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002455ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002456load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002457log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002458log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002459log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002460log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002461max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002462maxconn X X X -
2463mode X X X X
2464monitor fail - X X -
2465monitor-net X X X -
2466monitor-uri X X X -
2467option abortonclose (*) X - X X
2468option accept-invalid-http-request (*) X X X -
2469option accept-invalid-http-response (*) X - X X
2470option allbackups (*) X - X X
2471option checkcache (*) X - X X
2472option clitcpka (*) X X X -
2473option contstats (*) X X X -
2474option dontlog-normal (*) X X X -
2475option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002476-- keyword -------------------------- defaults - frontend - listen -- backend -
2477option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02002478option h1-case-adjust-bogus-client (*) X X X -
2479option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002480option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002481option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002482option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002483option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002484option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002485option http-server-close (*) X X X X
2486option http-use-proxy-header (*) X X X -
2487option httpchk X - X X
2488option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01002489option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002490option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002491option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002492option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002493option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002494option log-health-checks (*) X - X X
2495option log-separate-errors (*) X X X -
2496option logasap (*) X X X -
2497option mysql-check X - X X
2498option nolinger (*) X X X X
2499option originalto X X X X
2500option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002501option pgsql-check X - X X
2502option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002503option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002504option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002505option smtpchk X - X X
2506option socket-stats (*) X X X -
2507option splice-auto (*) X X X X
2508option splice-request (*) X X X X
2509option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002510option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002511option srvtcpka (*) X - X X
2512option ssl-hello-chk X - X X
2513-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002514option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002515option tcp-smart-accept (*) X X X -
2516option tcp-smart-connect (*) X - X X
2517option tcpka X X X X
2518option tcplog X X X X
2519option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002520external-check command X - X X
2521external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002522persist rdp-cookie X - X X
2523rate-limit sessions X X X -
2524redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002525-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002526retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02002527retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002528server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002529server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002530server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002531source X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002532stats admin - X X X
2533stats auth X X X X
2534stats enable X X X X
2535stats hide-version X X X X
2536stats http-request - X X X
2537stats realm X X X X
2538stats refresh X X X X
2539stats scope X X X X
2540stats show-desc X X X X
2541stats show-legends X X X X
2542stats show-node X X X X
2543stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002544-- keyword -------------------------- defaults - frontend - listen -- backend -
2545stick match - - X X
2546stick on - - X X
2547stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002548stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002549stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002550tcp-check connect - - X X
2551tcp-check expect - - X X
2552tcp-check send - - X X
2553tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002554tcp-request connection - X X -
2555tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002556tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002557tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002558tcp-response content - - X X
2559tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002560timeout check X - X X
2561timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002562timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002563timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002564timeout http-keep-alive X X X X
2565timeout http-request X X X X
2566timeout queue X - X X
2567timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002568timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002569timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002570timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002571transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002572unique-id-format X X X -
2573unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002574use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02002575use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002576use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002577------------------------------------+----------+----------+---------+---------
2578 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002579
Willy Tarreau0ba27502007-12-24 16:55:16 +01002580
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025814.2. Alphabetically sorted keywords reference
2582---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002583
2584This section provides a description of each keyword and its usage.
2585
2586
2587acl <aclname> <criterion> [flags] [operator] <value> ...
2588 Declare or complete an access list.
2589 May be used in sections : defaults | frontend | listen | backend
2590 no | yes | yes | yes
2591 Example:
2592 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2593 acl invalid_src src_port 0:1023
2594 acl local_dst hdr(host) -i localhost
2595
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002596 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002597
2598
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002599backlog <conns>
2600 Give hints to the system about the approximate listen backlog desired size
2601 May be used in sections : defaults | frontend | listen | backend
2602 yes | yes | yes | no
2603 Arguments :
2604 <conns> is the number of pending connections. Depending on the operating
2605 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002606 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002607
2608 In order to protect against SYN flood attacks, one solution is to increase
2609 the system's SYN backlog size. Depending on the system, sometimes it is just
2610 tunable via a system parameter, sometimes it is not adjustable at all, and
2611 sometimes the system relies on hints given by the application at the time of
2612 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2613 to the listen() syscall. On systems which can make use of this value, it can
2614 sometimes be useful to be able to specify a different value, hence this
2615 backlog parameter.
2616
2617 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2618 used as a hint and the system accepts up to the smallest greater power of
2619 two, and never more than some limits (usually 32768).
2620
2621 See also : "maxconn" and the target operating system's tuning guide.
2622
2623
Willy Tarreau0ba27502007-12-24 16:55:16 +01002624balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002625balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002626 Define the load balancing algorithm to be used in a backend.
2627 May be used in sections : defaults | frontend | listen | backend
2628 yes | no | yes | yes
2629 Arguments :
2630 <algorithm> is the algorithm used to select a server when doing load
2631 balancing. This only applies when no persistence information
2632 is available, or when a connection is redispatched to another
2633 server. <algorithm> may be one of the following :
2634
2635 roundrobin Each server is used in turns, according to their weights.
2636 This is the smoothest and fairest algorithm when the server's
2637 processing time remains equally distributed. This algorithm
2638 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002639 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002640 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002641 large farms, when a server becomes up after having been down
2642 for a very short time, it may sometimes take a few hundreds
2643 requests for it to be re-integrated into the farm and start
2644 receiving traffic. This is normal, though very rare. It is
2645 indicated here in case you would have the chance to observe
2646 it, so that you don't worry.
2647
2648 static-rr Each server is used in turns, according to their weights.
2649 This algorithm is as similar to roundrobin except that it is
2650 static, which means that changing a server's weight on the
2651 fly will have no effect. On the other hand, it has no design
2652 limitation on the number of servers, and when a server goes
2653 up, it is always immediately reintroduced into the farm, once
2654 the full map is recomputed. It also uses slightly less CPU to
2655 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002656
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002657 leastconn The server with the lowest number of connections receives the
2658 connection. Round-robin is performed within groups of servers
2659 of the same load to ensure that all servers will be used. Use
2660 of this algorithm is recommended where very long sessions are
2661 expected, such as LDAP, SQL, TSE, etc... but is not very well
2662 suited for protocols using short sessions such as HTTP. This
2663 algorithm is dynamic, which means that server weights may be
2664 adjusted on the fly for slow starts for instance.
2665
Willy Tarreauf09c6602012-02-13 17:12:08 +01002666 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002667 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002668 identifier to the highest (see server parameter "id"), which
2669 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002670 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002671 not make sense to use this algorithm without setting maxconn.
2672 The purpose of this algorithm is to always use the smallest
2673 number of servers so that extra servers can be powered off
2674 during non-intensive hours. This algorithm ignores the server
2675 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002676 or IMAP than HTTP, though it can be useful there too. In
2677 order to use this algorithm efficiently, it is recommended
2678 that a cloud controller regularly checks server usage to turn
2679 them off when unused, and regularly checks backend queue to
2680 turn new servers on when the queue inflates. Alternatively,
2681 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002682
Willy Tarreau0ba27502007-12-24 16:55:16 +01002683 source The source IP address is hashed and divided by the total
2684 weight of the running servers to designate which server will
2685 receive the request. This ensures that the same client IP
2686 address will always reach the same server as long as no
2687 server goes down or up. If the hash result changes due to the
2688 number of running servers changing, many clients will be
2689 directed to a different server. This algorithm is generally
2690 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002691 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002692 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002693 static by default, which means that changing a server's
2694 weight on the fly will have no effect, but this can be
2695 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002696
Oskar Stolc8dc41842012-05-19 10:19:54 +01002697 uri This algorithm hashes either the left part of the URI (before
2698 the question mark) or the whole URI (if the "whole" parameter
2699 is present) and divides the hash value by the total weight of
2700 the running servers. The result designates which server will
2701 receive the request. This ensures that the same URI will
2702 always be directed to the same server as long as no server
2703 goes up or down. This is used with proxy caches and
2704 anti-virus proxies in order to maximize the cache hit rate.
2705 Note that this algorithm may only be used in an HTTP backend.
2706 This algorithm is static by default, which means that
2707 changing a server's weight on the fly will have no effect,
2708 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002709
Oskar Stolc8dc41842012-05-19 10:19:54 +01002710 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002711 "depth", both followed by a positive integer number. These
2712 options may be helpful when it is needed to balance servers
2713 based on the beginning of the URI only. The "len" parameter
2714 indicates that the algorithm should only consider that many
2715 characters at the beginning of the URI to compute the hash.
2716 Note that having "len" set to 1 rarely makes sense since most
2717 URIs start with a leading "/".
2718
2719 The "depth" parameter indicates the maximum directory depth
2720 to be used to compute the hash. One level is counted for each
2721 slash in the request. If both parameters are specified, the
2722 evaluation stops when either is reached.
2723
Willy Tarreau0ba27502007-12-24 16:55:16 +01002724 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002725 the query string of each HTTP GET request.
2726
2727 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002728 request entity will be searched for the parameter argument,
2729 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002730 ('?') in the URL. The message body will only start to be
2731 analyzed once either the advertised amount of data has been
2732 received or the request buffer is full. In the unlikely event
2733 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002734 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002735 be randomly balanced if at all. This keyword used to support
2736 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002737
2738 If the parameter is found followed by an equal sign ('=') and
2739 a value, then the value is hashed and divided by the total
2740 weight of the running servers. The result designates which
2741 server will receive the request.
2742
2743 This is used to track user identifiers in requests and ensure
2744 that a same user ID will always be sent to the same server as
2745 long as no server goes up or down. If no value is found or if
2746 the parameter is not found, then a round robin algorithm is
2747 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002748 backend. This algorithm is static by default, which means
2749 that changing a server's weight on the fly will have no
2750 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002751
Cyril Bontédc4d9032012-04-08 21:57:39 +02002752 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2753 request. Just as with the equivalent ACL 'hdr()' function,
2754 the header name in parenthesis is not case sensitive. If the
2755 header is absent or if it does not contain any value, the
2756 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002757
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002758 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002759 reducing the hash algorithm to the main domain part with some
2760 specific headers such as 'Host'. For instance, in the Host
2761 value "haproxy.1wt.eu", only "1wt" will be considered.
2762
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002763 This algorithm is static by default, which means that
2764 changing a server's weight on the fly will have no effect,
2765 but this can be changed using "hash-type".
2766
Willy Tarreau21c741a2019-01-14 18:14:27 +01002767 random
2768 random(<draws>)
2769 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02002770 hashing function. This means that the servers' weights are
2771 respected, dynamic weight changes immediately take effect, as
2772 well as new server additions. Random load balancing can be
2773 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01002774 or removed as it may avoid the hammering effect that could
2775 result from roundrobin or leastconn in this situation. The
2776 hash-balance-factor directive can be used to further improve
2777 fairness of the load balancing, especially in situations
2778 where servers show highly variable response times. When an
2779 argument <draws> is present, it must be an integer value one
2780 or greater, indicating the number of draws before selecting
2781 the least loaded of these servers. It was indeed demonstrated
2782 that picking the least loaded of two servers is enough to
2783 significantly improve the fairness of the algorithm, by
2784 always avoiding to pick the most loaded server within a farm
2785 and getting rid of any bias that could be induced by the
2786 unfair distribution of the consistent list. Higher values N
2787 will take away N-1 of the highest loaded servers at the
2788 expense of performance. With very high values, the algorithm
2789 will converge towards the leastconn's result but much slower.
2790 The default value is 2, which generally shows very good
2791 distribution and performance. This algorithm is also known as
2792 the Power of Two Random Choices and is described here :
2793 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02002794
Emeric Brun736aa232009-06-30 17:56:00 +02002795 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002796 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002797 The RDP cookie <name> (or "mstshash" if omitted) will be
2798 looked up and hashed for each incoming TCP request. Just as
2799 with the equivalent ACL 'req_rdp_cookie()' function, the name
2800 is not case-sensitive. This mechanism is useful as a degraded
2801 persistence mode, as it makes it possible to always send the
2802 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002803 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002804 used instead.
2805
2806 Note that for this to work, the frontend must ensure that an
2807 RDP cookie is already present in the request buffer. For this
2808 you must use 'tcp-request content accept' rule combined with
2809 a 'req_rdp_cookie_cnt' ACL.
2810
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002811 This algorithm is static by default, which means that
2812 changing a server's weight on the fly will have no effect,
2813 but this can be changed using "hash-type".
2814
Cyril Bontédc4d9032012-04-08 21:57:39 +02002815 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002816
Willy Tarreau0ba27502007-12-24 16:55:16 +01002817 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002818 algorithms. Right now, only "url_param" and "uri" support an
2819 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002820
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002821 The load balancing algorithm of a backend is set to roundrobin when no other
2822 algorithm, mode nor option have been set. The algorithm may only be set once
2823 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002824
Lukas Tribus80512b12018-10-27 20:07:40 +02002825 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05002826 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02002827 to be routed to different backend servers, breaking the invalid assumptions
2828 NTLM relies on.
2829
Willy Tarreau0ba27502007-12-24 16:55:16 +01002830 Examples :
2831 balance roundrobin
2832 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002833 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002834 balance hdr(User-Agent)
2835 balance hdr(host)
2836 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002837
2838 Note: the following caveats and limitations on using the "check_post"
2839 extension with "url_param" must be considered :
2840
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002841 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002842 to determine if the parameters will be found in the body or entity which
2843 may contain binary data. Therefore another method may be required to
2844 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02002845 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002846
2847 - using a <max_wait> value larger than the request buffer size does not
2848 make sense and is useless. The buffer size is set at build time, and
2849 defaults to 16 kB.
2850
2851 - Content-Encoding is not supported, the parameter search will probably
2852 fail; and load balancing will fall back to Round Robin.
2853
2854 - Expect: 100-continue is not supported, load balancing will fall back to
2855 Round Robin.
2856
Lukas Tribus23953682017-04-28 13:24:30 +00002857 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002858 If the entire parameter value is not present in the first chunk, the
2859 selection of server is undefined (actually, defined by how little
2860 actually appeared in the first chunk).
2861
2862 - This feature does not support generation of a 100, 411 or 501 response.
2863
2864 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002865 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002866 white space or control characters are found, indicating the end of what
2867 might be a URL parameter list. This is probably not a concern with SGML
2868 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002869
Willy Tarreau294d0f02015-08-10 19:40:12 +02002870 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002871
2872
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002873bind [<address>]:<port_range> [, ...] [param*]
2874bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002875 Define one or several listening addresses and/or ports in a frontend.
2876 May be used in sections : defaults | frontend | listen | backend
2877 no | yes | yes | no
2878 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002879 <address> is optional and can be a host name, an IPv4 address, an IPv6
2880 address, or '*'. It designates the address the frontend will
2881 listen on. If unset, all IPv4 addresses of the system will be
2882 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002883 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002884 Optionally, an address family prefix may be used before the
2885 address to force the family regardless of the address format,
2886 which can be useful to specify a path to a unix socket with
2887 no slash ('/'). Currently supported prefixes are :
2888 - 'ipv4@' -> address is always IPv4
2889 - 'ipv6@' -> address is always IPv6
2890 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002891 - 'abns@' -> address is in abstract namespace (Linux only).
2892 Note: since abstract sockets are not "rebindable", they
2893 do not cope well with multi-process mode during
2894 soft-restart, so it is better to avoid them if
2895 nbproc is greater than 1. The effect is that if the
2896 new process fails to start, only one of the old ones
2897 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002898 - 'fd@<n>' -> use file descriptor <n> inherited from the
2899 parent. The fd must be bound and may or may not already
2900 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002901 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2902 connected unix socket or of a socketpair. The bind waits
2903 to receive a FD over the unix socket and uses it as if it
2904 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002905 You may want to reference some environment variables in the
2906 address parameter, see section 2.3 about environment
2907 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002908
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002909 <port_range> is either a unique TCP port, or a port range for which the
2910 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002911 above. The port is mandatory for TCP listeners. Note that in
2912 the case of an IPv6 address, the port is always the number
2913 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002914 - a numerical port (ex: '80')
2915 - a dash-delimited ports range explicitly stating the lower
2916 and upper bounds (ex: '2000-2100') which are included in
2917 the range.
2918
2919 Particular care must be taken against port ranges, because
2920 every <address:port> couple consumes one socket (= a file
2921 descriptor), so it's easy to consume lots of descriptors
2922 with a simple range, and to run out of sockets. Also, each
2923 <address:port> couple must be used only once among all
2924 instances running on a same system. Please note that binding
2925 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002926 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002927 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002928
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002929 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002930 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002931 receive UNIX connections on the socket located at this place.
2932 The path must begin with a slash and by default is absolute.
2933 It can be relative to the prefix defined by "unix-bind" in
2934 the global section. Note that the total length of the prefix
2935 followed by the socket path cannot exceed some system limits
2936 for UNIX sockets, which commonly are set to 107 characters.
2937
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002938 <param*> is a list of parameters common to all sockets declared on the
2939 same line. These numerous parameters depend on OS and build
2940 options and have a complete section dedicated to them. Please
2941 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002942
Willy Tarreau0ba27502007-12-24 16:55:16 +01002943 It is possible to specify a list of address:port combinations delimited by
2944 commas. The frontend will then listen on all of these addresses. There is no
2945 fixed limit to the number of addresses and ports which can be listened on in
2946 a frontend, as well as there is no limit to the number of "bind" statements
2947 in a frontend.
2948
2949 Example :
2950 listen http_proxy
2951 bind :80,:443
2952 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002953 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002954
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002955 listen http_https_proxy
2956 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002957 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002958
Willy Tarreau24709282013-03-10 21:32:12 +01002959 listen http_https_proxy_explicit
2960 bind ipv6@:80
2961 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2962 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2963
Willy Tarreaudad36a32013-03-11 01:20:04 +01002964 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002965 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002966
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002967 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2968 sun_path length is used for the address length. Some other programs
2969 such as socat use the string length only by default. Pass the option
2970 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2971 make it compatible with HAProxy's.
2972
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002973 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002974 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002975
2976
Christopher Fauletff4121f2017-11-22 16:38:49 +01002977bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002978 Limit visibility of an instance to a certain set of processes numbers.
2979 May be used in sections : defaults | frontend | listen | backend
2980 yes | yes | yes | yes
2981 Arguments :
2982 all All process will see this instance. This is the default. It
2983 may be used to override a default value.
2984
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002985 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002986 option may be combined with other numbers.
2987
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002988 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002989 option may be combined with other numbers. Do not use it
2990 with less than 2 processes otherwise some instances might be
2991 missing from all processes.
2992
Christopher Fauletff4121f2017-11-22 16:38:49 +01002993 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002994 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002995 the machine's word size. Ranges can be partially defined. The
2996 higher bound can be omitted. In such case, it is replaced by
2997 the corresponding maximum value. If a proxy is bound to
2998 process numbers greater than the configured global.nbproc, it
2999 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003000 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003001
3002 This keyword limits binding of certain instances to certain processes. This
3003 is useful in order not to have too many processes listening to the same
3004 ports. For instance, on a dual-core machine, it might make sense to set
3005 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3006 and 'even' instances.
3007
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003008 At the moment, it is not possible to reference more than 32 or 64 processes
3009 using this keyword, but this should be more than enough for most setups.
3010 Please note that 'all' really means all processes regardless of the machine's
3011 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003012
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003013 Each "bind" line may further be limited to a subset of the proxy's processes,
3014 please consult the "process" bind keyword in section 5.1.
3015
Willy Tarreaub369a042014-09-16 13:21:03 +02003016 When a frontend has no explicit "bind-process" line, it tries to bind to all
3017 the processes referenced by its "bind" lines. That means that frontends can
3018 easily adapt to their listeners' processes.
3019
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003020 If some backends are referenced by frontends bound to other processes, the
3021 backend automatically inherits the frontend's processes.
3022
3023 Example :
3024 listen app_ip1
3025 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003026 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003027
3028 listen app_ip2
3029 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003030 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003031
3032 listen management
3033 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003034 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003035
Willy Tarreau110ecc12012-11-15 17:50:01 +01003036 listen management
3037 bind 10.0.0.4:80
3038 bind-process 1-4
3039
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003040 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003041
3042
Willy Tarreau0ba27502007-12-24 16:55:16 +01003043capture cookie <name> len <length>
3044 Capture and log a cookie in the request and in the response.
3045 May be used in sections : defaults | frontend | listen | backend
3046 no | yes | yes | no
3047 Arguments :
3048 <name> is the beginning of the name of the cookie to capture. In order
3049 to match the exact name, simply suffix the name with an equal
3050 sign ('='). The full name will appear in the logs, which is
3051 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003052 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003053
3054 <length> is the maximum number of characters to report in the logs, which
3055 include the cookie name, the equal sign and the value, all in the
3056 standard "name=value" form. The string will be truncated on the
3057 right if it exceeds <length>.
3058
3059 Only the first cookie is captured. Both the "cookie" request headers and the
3060 "set-cookie" response headers are monitored. This is particularly useful to
3061 check for application bugs causing session crossing or stealing between
3062 users, because generally the user's cookies can only change on a login page.
3063
3064 When the cookie was not presented by the client, the associated log column
3065 will report "-". When a request does not cause a cookie to be assigned by the
3066 server, a "-" is reported in the response column.
3067
3068 The capture is performed in the frontend only because it is necessary that
3069 the log format does not change for a given frontend depending on the
3070 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003071 "capture cookie" statement in a frontend. The maximum capture length is set
3072 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3073 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003074
3075 Example:
3076 capture cookie ASPSESSION len 32
3077
3078 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003079 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003080
3081
3082capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003083 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003084 May be used in sections : defaults | frontend | listen | backend
3085 no | yes | yes | no
3086 Arguments :
3087 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003088 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003089 appear in the requests, with the first letter of each word in
3090 upper case. The header name will not appear in the logs, only the
3091 value is reported, but the position in the logs is respected.
3092
3093 <length> is the maximum number of characters to extract from the value and
3094 report in the logs. The string will be truncated on the right if
3095 it exceeds <length>.
3096
Willy Tarreau4460d032012-11-21 23:37:37 +01003097 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003098 value will be added to the logs between braces ('{}'). If multiple headers
3099 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003100 in the same order they were declared in the configuration. Non-existent
3101 headers will be logged just as an empty string. Common uses for request
3102 header captures include the "Host" field in virtual hosting environments, the
3103 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003104 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003105 environments to find where the request came from.
3106
3107 Note that when capturing headers such as "User-agent", some spaces may be
3108 logged, making the log analysis more difficult. Thus be careful about what
3109 you log if you know your log parser is not smart enough to rely on the
3110 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003111
Willy Tarreau0900abb2012-11-22 00:21:46 +01003112 There is no limit to the number of captured request headers nor to their
3113 length, though it is wise to keep them low to limit memory usage per session.
3114 In order to keep log format consistent for a same frontend, header captures
3115 can only be declared in a frontend. It is not possible to specify a capture
3116 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003117
3118 Example:
3119 capture request header Host len 15
3120 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003121 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003122
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003123 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003124 about logging.
3125
3126
3127capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003128 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003129 May be used in sections : defaults | frontend | listen | backend
3130 no | yes | yes | no
3131 Arguments :
3132 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003133 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003134 appear in the response, with the first letter of each word in
3135 upper case. The header name will not appear in the logs, only the
3136 value is reported, but the position in the logs is respected.
3137
3138 <length> is the maximum number of characters to extract from the value and
3139 report in the logs. The string will be truncated on the right if
3140 it exceeds <length>.
3141
Willy Tarreau4460d032012-11-21 23:37:37 +01003142 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003143 result will be added to the logs between braces ('{}') after the captured
3144 request headers. If multiple headers are captured, they will be delimited by
3145 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003146 the configuration. Non-existent headers will be logged just as an empty
3147 string. Common uses for response header captures include the "Content-length"
3148 header which indicates how many bytes are expected to be returned, the
3149 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003150
Willy Tarreau0900abb2012-11-22 00:21:46 +01003151 There is no limit to the number of captured response headers nor to their
3152 length, though it is wise to keep them low to limit memory usage per session.
3153 In order to keep log format consistent for a same frontend, header captures
3154 can only be declared in a frontend. It is not possible to specify a capture
3155 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003156
3157 Example:
3158 capture response header Content-length len 9
3159 capture response header Location len 15
3160
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003161 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003162 about logging.
3163
3164
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003165compression algo <algorithm> ...
3166compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003167compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003168 Enable HTTP compression.
3169 May be used in sections : defaults | frontend | listen | backend
3170 yes | yes | yes | yes
3171 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003172 algo is followed by the list of supported compression algorithms.
3173 type is followed by the list of MIME types that will be compressed.
3174 offload makes haproxy work as a compression offloader only (see notes).
3175
3176 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003177 identity this is mostly for debugging, and it was useful for developing
3178 the compression feature. Identity does not apply any change on
3179 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003180
Willy Tarreauc91840a2015-03-28 17:00:39 +01003181 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003182 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003183
3184 deflate same as "gzip", but with deflate algorithm and zlib format.
3185 Note that this algorithm has ambiguous support on many
3186 browsers and no support at all from recent ones. It is
3187 strongly recommended not to use it for anything else than
3188 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003189 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003190
Willy Tarreauc91840a2015-03-28 17:00:39 +01003191 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3192 alternative when the browser wants "deflate". All major
3193 browsers understand it and despite violating the standards,
3194 it is known to work better than "deflate", at least on MSIE
3195 and some versions of Safari. Do not use it in conjunction
3196 with "deflate", use either one or the other since both react
3197 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003198 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003199
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003200 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003201 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003202 If backend servers support HTTP compression, these directives
3203 will be no-op: haproxy will see the compressed response and will not
3204 compress again. If backend servers do not support HTTP compression and
3205 there is Accept-Encoding header in request, haproxy will compress the
3206 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003207
3208 The "offload" setting makes haproxy remove the Accept-Encoding header to
3209 prevent backend servers from compressing responses. It is strongly
3210 recommended not to do this because this means that all the compression work
3211 will be done on the single point where haproxy is located. However in some
3212 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003213 with broken HTTP compression implementation which can't be turned off.
3214 In that case haproxy can be used to prevent that gateway from emitting
3215 invalid payloads. In this case, simply removing the header in the
3216 configuration does not work because it applies before the header is parsed,
3217 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003218 then be used for such scenarios. Note: for now, the "offload" setting is
3219 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003220
William Lallemand05097442012-11-20 12:14:28 +01003221 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003222 * the request does not advertise a supported compression algorithm in the
3223 "Accept-Encoding" header
3224 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003225 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003226 * response contain neither a "Content-Length" header nor a
3227 "Transfer-Encoding" whose last value is "chunked"
3228 * response contains a "Content-Type" header whose first value starts with
3229 "multipart"
3230 * the response contains the "no-transform" value in the "Cache-control"
3231 header
3232 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3233 and later
3234 * The response contains a "Content-Encoding" header, indicating that the
3235 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003236 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003237
Tim Duesterhusb229f012019-01-29 16:38:56 +01003238 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003239
William Lallemand82fe75c2012-10-23 10:25:10 +02003240 Examples :
3241 compression algo gzip
3242 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003243
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003244
Willy Tarreau55165fe2009-05-10 12:02:55 +02003245cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003246 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3247 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003248 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003249 Enable cookie-based persistence in a backend.
3250 May be used in sections : defaults | frontend | listen | backend
3251 yes | no | yes | yes
3252 Arguments :
3253 <name> is the name of the cookie which will be monitored, modified or
3254 inserted in order to bring persistence. This cookie is sent to
3255 the client via a "Set-Cookie" header in the response, and is
3256 brought back by the client in a "Cookie" header in all requests.
3257 Special care should be taken to choose a name which does not
3258 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003259 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003260 HTTP/HTTPS), care should be taken to use different cookie names
3261 between all backends if persistence between them is not desired.
3262
3263 rewrite This keyword indicates that the cookie will be provided by the
3264 server and that haproxy will have to modify its value to set the
3265 server's identifier in it. This mode is handy when the management
3266 of complex combinations of "Set-cookie" and "Cache-control"
3267 headers is left to the application. The application can then
3268 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003269 cookie. Since all responses should be monitored, this mode
3270 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003271 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003272 start with this mode for new deployments. This keyword is
3273 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003274
3275 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003276 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003277
Willy Tarreaua79094d2010-08-31 22:54:15 +02003278 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003279 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003280 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003281 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003282 existing configurations running in the "rewrite" mode. The cookie
3283 will only be a session cookie and will not be stored on the
3284 client's disk. By default, unless the "indirect" option is added,
3285 the server will see the cookies emitted by the client. Due to
3286 caching effects, it is generally wise to add the "nocache" or
3287 "postonly" keywords (see below). The "insert" keyword is not
3288 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003289
3290 prefix This keyword indicates that instead of relying on a dedicated
3291 cookie for the persistence, an existing one will be completed.
3292 This may be needed in some specific environments where the client
3293 does not support more than one single cookie and the application
3294 already needs it. In this case, whenever the server sets a cookie
3295 named <name>, it will be prefixed with the server's identifier
3296 and a delimiter. The prefix will be removed from all client
3297 requests so that the server still finds the cookie it emitted.
3298 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003299 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003300 not compatible with "rewrite" and "insert". Note: it is highly
3301 recommended not to use "indirect" with "prefix", otherwise server
3302 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003303
Willy Tarreaua79094d2010-08-31 22:54:15 +02003304 indirect When this option is specified, no cookie will be emitted to a
3305 client which already has a valid one for the server which has
3306 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003307 it will be removed, unless the "preserve" option is also set. In
3308 "insert" mode, this will additionally remove cookies from the
3309 requests transmitted to the server, making the persistence
3310 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003311 Note: it is highly recommended not to use "indirect" with
3312 "prefix", otherwise server cookie updates would not be sent to
3313 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003314
3315 nocache This option is recommended in conjunction with the insert mode
3316 when there is a cache between the client and HAProxy, as it
3317 ensures that a cacheable response will be tagged non-cacheable if
3318 a cookie needs to be inserted. This is important because if all
3319 persistence cookies are added on a cacheable home page for
3320 instance, then all customers will then fetch the page from an
3321 outer cache and will all share the same persistence cookie,
3322 leading to one server receiving much more traffic than others.
3323 See also the "insert" and "postonly" options.
3324
3325 postonly This option ensures that cookie insertion will only be performed
3326 on responses to POST requests. It is an alternative to the
3327 "nocache" option, because POST responses are not cacheable, so
3328 this ensures that the persistence cookie will never get cached.
3329 Since most sites do not need any sort of persistence before the
3330 first POST which generally is a login request, this is a very
3331 efficient method to optimize caching without risking to find a
3332 persistence cookie in the cache.
3333 See also the "insert" and "nocache" options.
3334
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003335 preserve This option may only be used with "insert" and/or "indirect". It
3336 allows the server to emit the persistence cookie itself. In this
3337 case, if a cookie is found in the response, haproxy will leave it
3338 untouched. This is useful in order to end persistence after a
3339 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003340 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003341 the past. By combining this mechanism with the "disable-on-404"
3342 check option, it is possible to perform a completely graceful
3343 shutdown because users will definitely leave the server after
3344 they logout.
3345
Willy Tarreau4992dd22012-05-31 21:02:17 +02003346 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3347 when a cookie is inserted. This attribute is used so that a
3348 user agent doesn't share the cookie with non-HTTP components.
3349 Please check RFC6265 for more information on this attribute.
3350
3351 secure This option tells haproxy to add a "Secure" cookie attribute when
3352 a cookie is inserted. This attribute is used so that a user agent
3353 never emits this cookie over non-secure channels, which means
3354 that a cookie learned with this flag will be presented only over
3355 SSL/TLS connections. Please check RFC6265 for more information on
3356 this attribute.
3357
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003358 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003359 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003360 name. If the domain begins with a dot, the browser is allowed to
3361 use it for any host ending with that name. It is also possible to
3362 specify several domain names by invoking this option multiple
3363 times. Some browsers might have small limits on the number of
3364 domains, so be careful when doing that. For the record, sending
3365 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003366
Willy Tarreau996a92c2010-10-13 19:30:47 +02003367 maxidle This option allows inserted cookies to be ignored after some idle
3368 time. It only works with insert-mode cookies. When a cookie is
3369 sent to the client, the date this cookie was emitted is sent too.
3370 Upon further presentations of this cookie, if the date is older
3371 than the delay indicated by the parameter (in seconds), it will
3372 be ignored. Otherwise, it will be refreshed if needed when the
3373 response is sent to the client. This is particularly useful to
3374 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003375 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003376 this option is set and a cookie has no date, it is always
3377 accepted, but gets refreshed in the response. This maintains the
3378 ability for admins to access their sites. Cookies that have a
3379 date in the future further than 24 hours are ignored. Doing so
3380 lets admins fix timezone issues without risking kicking users off
3381 the site.
3382
3383 maxlife This option allows inserted cookies to be ignored after some life
3384 time, whether they're in use or not. It only works with insert
3385 mode cookies. When a cookie is first sent to the client, the date
3386 this cookie was emitted is sent too. Upon further presentations
3387 of this cookie, if the date is older than the delay indicated by
3388 the parameter (in seconds), it will be ignored. If the cookie in
3389 the request has no date, it is accepted and a date will be set.
3390 Cookies that have a date in the future further than 24 hours are
3391 ignored. Doing so lets admins fix timezone issues without risking
3392 kicking users off the site. Contrary to maxidle, this value is
3393 not refreshed, only the first visit date counts. Both maxidle and
3394 maxlife may be used at the time. This is particularly useful to
3395 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003396 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003397 is stronger than the maxidle method in that it forces a
3398 redispatch after some absolute delay.
3399
Olivier Houchard4e694042017-03-14 20:01:29 +01003400 dynamic Activate dynamic cookies. When used, a session cookie is
3401 dynamically created for each server, based on the IP and port
3402 of the server, and a secret key, specified in the
3403 "dynamic-cookie-key" backend directive.
3404 The cookie will be regenerated each time the IP address change,
3405 and is only generated for IPv4/IPv6.
3406
Willy Tarreau0ba27502007-12-24 16:55:16 +01003407 There can be only one persistence cookie per HTTP backend, and it can be
3408 declared in a defaults section. The value of the cookie will be the value
3409 indicated after the "cookie" keyword in a "server" statement. If no cookie
3410 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003411
Willy Tarreau0ba27502007-12-24 16:55:16 +01003412 Examples :
3413 cookie JSESSIONID prefix
3414 cookie SRV insert indirect nocache
3415 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003416 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003417
Willy Tarreau294d0f02015-08-10 19:40:12 +02003418 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003419
Willy Tarreau983e01e2010-01-11 18:42:06 +01003420
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003421declare capture [ request | response ] len <length>
3422 Declares a capture slot.
3423 May be used in sections : defaults | frontend | listen | backend
3424 no | yes | yes | no
3425 Arguments:
3426 <length> is the length allowed for the capture.
3427
3428 This declaration is only available in the frontend or listen section, but the
3429 reserved slot can be used in the backends. The "request" keyword allocates a
3430 capture slot for use in the request, and "response" allocates a capture slot
3431 for use in the response.
3432
3433 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003434 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003435 "http-request capture" and "http-response capture".
3436
3437
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003438default-server [param*]
3439 Change default options for a server in a backend
3440 May be used in sections : defaults | frontend | listen | backend
3441 yes | no | yes | yes
3442 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003443 <param*> is a list of parameters for this server. The "default-server"
3444 keyword accepts an important number of options and has a complete
3445 section dedicated to it. Please refer to section 5 for more
3446 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003447
Willy Tarreau983e01e2010-01-11 18:42:06 +01003448 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003449 default-server inter 1000 weight 13
3450
3451 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003452
Willy Tarreau983e01e2010-01-11 18:42:06 +01003453
Willy Tarreau0ba27502007-12-24 16:55:16 +01003454default_backend <backend>
3455 Specify the backend to use when no "use_backend" rule has been matched.
3456 May be used in sections : defaults | frontend | listen | backend
3457 yes | yes | yes | no
3458 Arguments :
3459 <backend> is the name of the backend to use.
3460
3461 When doing content-switching between frontend and backends using the
3462 "use_backend" keyword, it is often useful to indicate which backend will be
3463 used when no rule has matched. It generally is the dynamic backend which
3464 will catch all undetermined requests.
3465
Willy Tarreau0ba27502007-12-24 16:55:16 +01003466 Example :
3467
3468 use_backend dynamic if url_dyn
3469 use_backend static if url_css url_img extension_img
3470 default_backend dynamic
3471
Willy Tarreau98d04852015-05-26 12:18:29 +02003472 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003473
Willy Tarreau0ba27502007-12-24 16:55:16 +01003474
Baptiste Assmann27f51342013-10-09 06:51:49 +02003475description <string>
3476 Describe a listen, frontend or backend.
3477 May be used in sections : defaults | frontend | listen | backend
3478 no | yes | yes | yes
3479 Arguments : string
3480
3481 Allows to add a sentence to describe the related object in the HAProxy HTML
3482 stats page. The description will be printed on the right of the object name
3483 it describes.
3484 No need to backslash spaces in the <string> arguments.
3485
3486
Willy Tarreau0ba27502007-12-24 16:55:16 +01003487disabled
3488 Disable a proxy, frontend or backend.
3489 May be used in sections : defaults | frontend | listen | backend
3490 yes | yes | yes | yes
3491 Arguments : none
3492
3493 The "disabled" keyword is used to disable an instance, mainly in order to
3494 liberate a listening port or to temporarily disable a service. The instance
3495 will still be created and its configuration will be checked, but it will be
3496 created in the "stopped" state and will appear as such in the statistics. It
3497 will not receive any traffic nor will it send any health-checks or logs. It
3498 is possible to disable many instances at once by adding the "disabled"
3499 keyword in a "defaults" section.
3500
3501 See also : "enabled"
3502
3503
Willy Tarreau5ce94572010-06-07 14:35:41 +02003504dispatch <address>:<port>
3505 Set a default server address
3506 May be used in sections : defaults | frontend | listen | backend
3507 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003508 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003509
3510 <address> is the IPv4 address of the default server. Alternatively, a
3511 resolvable hostname is supported, but this name will be resolved
3512 during start-up.
3513
3514 <ports> is a mandatory port specification. All connections will be sent
3515 to this port, and it is not permitted to use port offsets as is
3516 possible with normal servers.
3517
Willy Tarreau787aed52011-04-15 06:45:37 +02003518 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003519 server can take the connection. In the past it was used to forward non
3520 persistent connections to an auxiliary load balancer. Due to its simple
3521 syntax, it has also been used for simple TCP relays. It is recommended not to
3522 use it for more clarity, and to use the "server" directive instead.
3523
3524 See also : "server"
3525
Olivier Houchard4e694042017-03-14 20:01:29 +01003526
3527dynamic-cookie-key <string>
3528 Set the dynamic cookie secret key for a backend.
3529 May be used in sections : defaults | frontend | listen | backend
3530 yes | no | yes | yes
3531 Arguments : The secret key to be used.
3532
3533 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003534 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003535 specified on the "server" line), using a hash of the IP address of the
3536 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003537 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003538 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003539
Willy Tarreau0ba27502007-12-24 16:55:16 +01003540enabled
3541 Enable a proxy, frontend or backend.
3542 May be used in sections : defaults | frontend | listen | backend
3543 yes | yes | yes | yes
3544 Arguments : none
3545
3546 The "enabled" keyword is used to explicitly enable an instance, when the
3547 defaults has been set to "disabled". This is very rarely used.
3548
3549 See also : "disabled"
3550
3551
3552errorfile <code> <file>
3553 Return a file contents instead of errors generated by HAProxy
3554 May be used in sections : defaults | frontend | listen | backend
3555 yes | yes | yes | yes
3556 Arguments :
3557 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003558 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3559 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003560
3561 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003562 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003563 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003564 error pages, and to use absolute paths, since files are read
3565 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003566
3567 It is important to understand that this keyword is not meant to rewrite
3568 errors returned by the server, but errors detected and returned by HAProxy.
3569 This is why the list of supported errors is limited to a small set.
3570
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003571 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3572
Willy Tarreau0ba27502007-12-24 16:55:16 +01003573 The files are returned verbatim on the TCP socket. This allows any trick such
3574 as redirections to another URL or site, as well as tricks to clean cookies,
3575 force enable or disable caching, etc... The package provides default error
3576 files returning the same contents as default errors.
3577
Willy Tarreau59140a22009-02-22 12:02:30 +01003578 The files should not exceed the configured buffer size (BUFSIZE), which
3579 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003580 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003581 loops between the client and HAProxy when all servers are down, causing an
3582 error to be returned instead of an image. For better HTTP compliance, it is
3583 recommended that all header lines end with CR-LF and not LF alone.
3584
Willy Tarreau0ba27502007-12-24 16:55:16 +01003585 The files are read at the same time as the configuration and kept in memory.
3586 For this reason, the errors continue to be returned even when the process is
3587 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003588 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003589 403 status code and interrogating a blocked URL.
3590
3591 See also : "errorloc", "errorloc302", "errorloc303"
3592
Willy Tarreau59140a22009-02-22 12:02:30 +01003593 Example :
3594 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003595 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003596 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3597 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3598
Willy Tarreau2769aa02007-12-27 18:26:09 +01003599
3600errorloc <code> <url>
3601errorloc302 <code> <url>
3602 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3603 May be used in sections : defaults | frontend | listen | backend
3604 yes | yes | yes | yes
3605 Arguments :
3606 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003607 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3608 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003609
3610 <url> it is the exact contents of the "Location" header. It may contain
3611 either a relative URI to an error page hosted on the same site,
3612 or an absolute URI designating an error page on another site.
3613 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003614 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003615
3616 It is important to understand that this keyword is not meant to rewrite
3617 errors returned by the server, but errors detected and returned by HAProxy.
3618 This is why the list of supported errors is limited to a small set.
3619
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003620 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3621
Willy Tarreau2769aa02007-12-27 18:26:09 +01003622 Note that both keyword return the HTTP 302 status code, which tells the
3623 client to fetch the designated URL using the same HTTP method. This can be
3624 quite problematic in case of non-GET methods such as POST, because the URL
3625 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003626 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003627 status code, indicating to the client that the URL must be fetched with a GET
3628 request.
3629
3630 See also : "errorfile", "errorloc303"
3631
3632
3633errorloc303 <code> <url>
3634 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3635 May be used in sections : defaults | frontend | listen | backend
3636 yes | yes | yes | yes
3637 Arguments :
3638 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003639 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3640 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003641
3642 <url> it is the exact contents of the "Location" header. It may contain
3643 either a relative URI to an error page hosted on the same site,
3644 or an absolute URI designating an error page on another site.
3645 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003646 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003647
3648 It is important to understand that this keyword is not meant to rewrite
3649 errors returned by the server, but errors detected and returned by HAProxy.
3650 This is why the list of supported errors is limited to a small set.
3651
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003652 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3653
Willy Tarreau2769aa02007-12-27 18:26:09 +01003654 Note that both keyword return the HTTP 303 status code, which tells the
3655 client to fetch the designated URL using the same HTTP GET method. This
3656 solves the usual problems associated with "errorloc" and the 302 code. It is
3657 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003658 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003659
3660 See also : "errorfile", "errorloc", "errorloc302"
3661
3662
Simon Horman51a1cf62015-02-03 13:00:44 +09003663email-alert from <emailaddr>
3664 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003665 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003666 May be used in sections: defaults | frontend | listen | backend
3667 yes | yes | yes | yes
3668
3669 Arguments :
3670
3671 <emailaddr> is the from email address to use when sending email alerts
3672
3673 Also requires "email-alert mailers" and "email-alert to" to be set
3674 and if so sending email alerts is enabled for the proxy.
3675
Simon Horman64e34162015-02-06 11:11:57 +09003676 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003677 "email-alert myhostname", "email-alert to", section 3.6 about
3678 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003679
3680
3681email-alert level <level>
3682 Declare the maximum log level of messages for which email alerts will be
3683 sent. This acts as a filter on the sending of email alerts.
3684 May be used in sections: defaults | frontend | listen | backend
3685 yes | yes | yes | yes
3686
3687 Arguments :
3688
3689 <level> One of the 8 syslog levels:
3690 emerg alert crit err warning notice info debug
3691 The above syslog levels are ordered from lowest to highest.
3692
3693 By default level is alert
3694
3695 Also requires "email-alert from", "email-alert mailers" and
3696 "email-alert to" to be set and if so sending email alerts is enabled
3697 for the proxy.
3698
Simon Horman1421e212015-04-30 13:10:35 +09003699 Alerts are sent when :
3700
3701 * An un-paused server is marked as down and <level> is alert or lower
3702 * A paused server is marked as down and <level> is notice or lower
3703 * A server is marked as up or enters the drain state and <level>
3704 is notice or lower
3705 * "option log-health-checks" is enabled, <level> is info or lower,
3706 and a health check status update occurs
3707
Simon Horman64e34162015-02-06 11:11:57 +09003708 See also : "email-alert from", "email-alert mailers",
3709 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003710 section 3.6 about mailers.
3711
3712
3713email-alert mailers <mailersect>
3714 Declare the mailers to be used when sending email alerts
3715 May be used in sections: defaults | frontend | listen | backend
3716 yes | yes | yes | yes
3717
3718 Arguments :
3719
3720 <mailersect> is the name of the mailers section to send email alerts.
3721
3722 Also requires "email-alert from" and "email-alert to" to be set
3723 and if so sending email alerts is enabled for the proxy.
3724
Simon Horman64e34162015-02-06 11:11:57 +09003725 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3726 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003727
3728
3729email-alert myhostname <hostname>
3730 Declare the to hostname address to be used when communicating with
3731 mailers.
3732 May be used in sections: defaults | frontend | listen | backend
3733 yes | yes | yes | yes
3734
3735 Arguments :
3736
Baptiste Assmann738bad92015-12-21 15:27:53 +01003737 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003738
3739 By default the systems hostname is used.
3740
3741 Also requires "email-alert from", "email-alert mailers" and
3742 "email-alert to" to be set and if so sending email alerts is enabled
3743 for the proxy.
3744
Simon Horman64e34162015-02-06 11:11:57 +09003745 See also : "email-alert from", "email-alert level", "email-alert mailers",
3746 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003747
3748
3749email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003750 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003751 header of email alerts. This is the address that email alerts are sent to.
3752 May be used in sections: defaults | frontend | listen | backend
3753 yes | yes | yes | yes
3754
3755 Arguments :
3756
3757 <emailaddr> is the to email address to use when sending email alerts
3758
3759 Also requires "email-alert mailers" and "email-alert to" to be set
3760 and if so sending email alerts is enabled for the proxy.
3761
Simon Horman64e34162015-02-06 11:11:57 +09003762 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003763 "email-alert myhostname", section 3.6 about mailers.
3764
3765
Willy Tarreau4de91492010-01-22 19:10:05 +01003766force-persist { if | unless } <condition>
3767 Declare a condition to force persistence on down servers
3768 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003769 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003770
3771 By default, requests are not dispatched to down servers. It is possible to
3772 force this using "option persist", but it is unconditional and redispatches
3773 to a valid server if "option redispatch" is set. That leaves with very little
3774 possibilities to force some requests to reach a server which is artificially
3775 marked down for maintenance operations.
3776
3777 The "force-persist" statement allows one to declare various ACL-based
3778 conditions which, when met, will cause a request to ignore the down status of
3779 a server and still try to connect to it. That makes it possible to start a
3780 server, still replying an error to the health checks, and run a specially
3781 configured browser to test the service. Among the handy methods, one could
3782 use a specific source IP address, or a specific cookie. The cookie also has
3783 the advantage that it can easily be added/removed on the browser from a test
3784 page. Once the service is validated, it is then possible to open the service
3785 to the world by returning a valid response to health checks.
3786
3787 The forced persistence is enabled when an "if" condition is met, or unless an
3788 "unless" condition is met. The final redispatch is always disabled when this
3789 is used.
3790
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003791 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003792 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003793
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003794
3795filter <name> [param*]
3796 Add the filter <name> in the filter list attached to the proxy.
3797 May be used in sections : defaults | frontend | listen | backend
3798 no | yes | yes | yes
3799 Arguments :
3800 <name> is the name of the filter. Officially supported filters are
3801 referenced in section 9.
3802
Tim Düsterhus4896c442016-11-29 02:15:19 +01003803 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003804 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003805 filter. Please refer to the documentation of the corresponding
3806 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003807
3808 Multiple occurrences of the filter line can be used for the same proxy. The
3809 same filter can be referenced many times if needed.
3810
3811 Example:
3812 listen
3813 bind *:80
3814
3815 filter trace name BEFORE-HTTP-COMP
3816 filter compression
3817 filter trace name AFTER-HTTP-COMP
3818
3819 compression algo gzip
3820 compression offload
3821
3822 server srv1 192.168.0.1:80
3823
3824 See also : section 9.
3825
Willy Tarreau4de91492010-01-22 19:10:05 +01003826
Willy Tarreau2769aa02007-12-27 18:26:09 +01003827fullconn <conns>
3828 Specify at what backend load the servers will reach their maxconn
3829 May be used in sections : defaults | frontend | listen | backend
3830 yes | no | yes | yes
3831 Arguments :
3832 <conns> is the number of connections on the backend which will make the
3833 servers use the maximal number of connections.
3834
Willy Tarreau198a7442008-01-17 12:05:32 +01003835 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003836 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003837 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003838 load. The server will then always accept at least <minconn> connections,
3839 never more than <maxconn>, and the limit will be on the ramp between both
3840 values when the backend has less than <conns> concurrent connections. This
3841 makes it possible to limit the load on the servers during normal loads, but
3842 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003843 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003844
Willy Tarreaufbb78422011-06-05 15:38:35 +02003845 Since it's hard to get this value right, haproxy automatically sets it to
3846 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003847 backend (based on "use_backend" and "default_backend" rules). That way it's
3848 safe to leave it unset. However, "use_backend" involving dynamic names are
3849 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003850
Willy Tarreau2769aa02007-12-27 18:26:09 +01003851 Example :
3852 # The servers will accept between 100 and 1000 concurrent connections each
3853 # and the maximum of 1000 will be reached when the backend reaches 10000
3854 # connections.
3855 backend dynamic
3856 fullconn 10000
3857 server srv1 dyn1:80 minconn 100 maxconn 1000
3858 server srv2 dyn2:80 minconn 100 maxconn 1000
3859
3860 See also : "maxconn", "server"
3861
3862
3863grace <time>
3864 Maintain a proxy operational for some time after a soft stop
3865 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003866 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003867 Arguments :
3868 <time> is the time (by default in milliseconds) for which the instance
3869 will remain operational with the frontend sockets still listening
3870 when a soft-stop is received via the SIGUSR1 signal.
3871
3872 This may be used to ensure that the services disappear in a certain order.
3873 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003874 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003875 needed by the equipment to detect the failure.
3876
3877 Note that currently, there is very little benefit in using this parameter,
3878 and it may in fact complicate the soft-reconfiguration process more than
3879 simplify it.
3880
Willy Tarreau0ba27502007-12-24 16:55:16 +01003881
Andrew Rodland17be45e2016-10-25 17:04:12 -04003882hash-balance-factor <factor>
3883 Specify the balancing factor for bounded-load consistent hashing
3884 May be used in sections : defaults | frontend | listen | backend
3885 yes | no | no | yes
3886 Arguments :
3887 <factor> is the control for the maximum number of concurrent requests to
3888 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01003889 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04003890
3891 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3892 enables an algorithm that prevents any one server from getting too many
3893 requests at once, even if some hash buckets receive many more requests than
3894 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3895 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3896 then no server will be allowed to have a load more than 1.5 times the average.
3897 If server weights are used, they will be respected.
3898
3899 If the first-choice server is disqualified, the algorithm will choose another
3900 server based on the request hash, until a server with additional capacity is
3901 found. A higher <factor> allows more imbalance between the servers, while a
3902 lower <factor> means that more servers will be checked on average, affecting
3903 performance. Reasonable values are from 125 to 200.
3904
Willy Tarreau760e81d2018-05-03 07:20:40 +02003905 This setting is also used by "balance random" which internally relies on the
3906 consistent hashing mechanism.
3907
Andrew Rodland17be45e2016-10-25 17:04:12 -04003908 See also : "balance" and "hash-type".
3909
3910
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003911hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003912 Specify a method to use for mapping hashes to servers
3913 May be used in sections : defaults | frontend | listen | backend
3914 yes | no | yes | yes
3915 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003916 <method> is the method used to select a server from the hash computed by
3917 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003918
Bhaskar98634f02013-10-29 23:30:51 -04003919 map-based the hash table is a static array containing all alive servers.
3920 The hashes will be very smooth, will consider weights, but
3921 will be static in that weight changes while a server is up
3922 will be ignored. This means that there will be no slow start.
3923 Also, since a server is selected by its position in the array,
3924 most mappings are changed when the server count changes. This
3925 means that when a server goes up or down, or when a server is
3926 added to a farm, most connections will be redistributed to
3927 different servers. This can be inconvenient with caches for
3928 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003929
Bhaskar98634f02013-10-29 23:30:51 -04003930 consistent the hash table is a tree filled with many occurrences of each
3931 server. The hash key is looked up in the tree and the closest
3932 server is chosen. This hash is dynamic, it supports changing
3933 weights while the servers are up, so it is compatible with the
3934 slow start feature. It has the advantage that when a server
3935 goes up or down, only its associations are moved. When a
3936 server is added to the farm, only a few part of the mappings
3937 are redistributed, making it an ideal method for caches.
3938 However, due to its principle, the distribution will never be
3939 very smooth and it may sometimes be necessary to adjust a
3940 server's weight or its ID to get a more balanced distribution.
3941 In order to get the same distribution on multiple load
3942 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003943 same IDs. Note: consistent hash uses sdbm and avalanche if no
3944 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003945
3946 <function> is the hash function to be used :
3947
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003948 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003949 reimplementation of ndbm) database library. It was found to do
3950 well in scrambling bits, causing better distribution of the keys
3951 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003952 function with good distribution, unless the total server weight
3953 is a multiple of 64, in which case applying the avalanche
3954 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003955
3956 djb2 this function was first proposed by Dan Bernstein many years ago
3957 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003958 function provides a better distribution than sdbm. It generally
3959 works well with text-based inputs though it can perform extremely
3960 poorly with numeric-only input or when the total server weight is
3961 a multiple of 33, unless the avalanche modifier is also used.
3962
Willy Tarreaua0f42712013-11-14 14:30:35 +01003963 wt6 this function was designed for haproxy while testing other
3964 functions in the past. It is not as smooth as the other ones, but
3965 is much less sensible to the input data set or to the number of
3966 servers. It can make sense as an alternative to sdbm+avalanche or
3967 djb2+avalanche for consistent hashing or when hashing on numeric
3968 data such as a source IP address or a visitor identifier in a URL
3969 parameter.
3970
Willy Tarreau324f07f2015-01-20 19:44:50 +01003971 crc32 this is the most common CRC32 implementation as used in Ethernet,
3972 gzip, PNG, etc. It is slower than the other ones but may provide
3973 a better distribution or less predictable results especially when
3974 used on strings.
3975
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003976 <modifier> indicates an optional method applied after hashing the key :
3977
3978 avalanche This directive indicates that the result from the hash
3979 function above should not be used in its raw form but that
3980 a 4-byte full avalanche hash must be applied first. The
3981 purpose of this step is to mix the resulting bits from the
3982 previous hash in order to avoid any undesired effect when
3983 the input contains some limited values or when the number of
3984 servers is a multiple of one of the hash's components (64
3985 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3986 result less predictable, but it's also not as smooth as when
3987 using the original function. Some testing might be needed
3988 with some workloads. This hash is one of the many proposed
3989 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003990
Bhaskar98634f02013-10-29 23:30:51 -04003991 The default hash type is "map-based" and is recommended for most usages. The
3992 default function is "sdbm", the selection of a function should be based on
3993 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003994
Andrew Rodland17be45e2016-10-25 17:04:12 -04003995 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003996
3997
Willy Tarreau0ba27502007-12-24 16:55:16 +01003998http-check disable-on-404
3999 Enable a maintenance mode upon HTTP/404 response to health-checks
4000 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01004001 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01004002 Arguments : none
4003
4004 When this option is set, a server which returns an HTTP code 404 will be
4005 excluded from further load-balancing, but will still receive persistent
4006 connections. This provides a very convenient method for Web administrators
4007 to perform a graceful shutdown of their servers. It is also important to note
4008 that a server which is detected as failed while it was in this mode will not
4009 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
4010 will immediately be reinserted into the farm. The status on the stats page
4011 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01004012 option only works in conjunction with the "httpchk" option. If this option
4013 is used with "http-check expect", then it has precedence over it so that 404
4014 responses will still be considered as soft-stop.
4015
4016 See also : "option httpchk", "http-check expect"
4017
4018
4019http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004020 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01004021 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02004022 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01004023 Arguments :
4024 <match> is a keyword indicating how to look for a specific pattern in the
4025 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004026 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01004027 exclamation mark ("!") to negate the match. Spaces are allowed
4028 between the exclamation mark and the keyword. See below for more
4029 details on the supported keywords.
4030
4031 <pattern> is the pattern to look for. It may be a string or a regular
4032 expression. If the pattern contains spaces, they must be escaped
4033 with the usual backslash ('\').
4034
4035 By default, "option httpchk" considers that response statuses 2xx and 3xx
4036 are valid, and that others are invalid. When "http-check expect" is used,
4037 it defines what is considered valid or invalid. Only one "http-check"
4038 statement is supported in a backend. If a server fails to respond or times
4039 out, the check obviously fails. The available matches are :
4040
4041 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004042 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004043 response's status code is exactly this string. If the
4044 "status" keyword is prefixed with "!", then the response
4045 will be considered invalid if the status code matches.
4046
4047 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004048 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004049 response's status code matches the expression. If the
4050 "rstatus" keyword is prefixed with "!", then the response
4051 will be considered invalid if the status code matches.
4052 This is mostly used to check for multiple codes.
4053
4054 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004055 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004056 response's body contains this exact string. If the
4057 "string" keyword is prefixed with "!", then the response
4058 will be considered invalid if the body contains this
4059 string. This can be used to look for a mandatory word at
4060 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004061 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01004062 trace).
4063
4064 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004065 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004066 response's body matches this expression. If the "rstring"
4067 keyword is prefixed with "!", then the response will be
4068 considered invalid if the body matches the expression.
4069 This can be used to look for a mandatory word at the end
4070 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01004071 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01004072
4073 It is important to note that the responses will be limited to a certain size
4074 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
4075 Thus, too large responses may not contain the mandatory pattern when using
4076 "string" or "rstring". If a large response is absolutely required, it is
4077 possible to change the default max size by setting the global variable.
4078 However, it is worth keeping in mind that parsing very large responses can
4079 waste some CPU cycles, especially when regular expressions are used, and that
4080 it is always better to focus the checks on smaller resources.
4081
Cyril Bonté32602d22015-01-30 00:07:07 +01004082 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
4083 will automatically append a "Connection: close" header, meaning that this
4084 header should not be present in the request provided by "option httpchk".
4085
Willy Tarreaubd741542010-03-16 18:46:54 +01004086 Last, if "http-check expect" is combined with "http-check disable-on-404",
4087 then this last one has precedence when the server responds with 404.
4088
4089 Examples :
4090 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004091 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01004092
4093 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004094 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01004095
4096 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004097 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01004098
4099 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004100 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01004101
Willy Tarreaubd741542010-03-16 18:46:54 +01004102 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004103
4104
Willy Tarreauef781042010-01-27 11:53:01 +01004105http-check send-state
4106 Enable emission of a state header with HTTP health checks
4107 May be used in sections : defaults | frontend | listen | backend
4108 yes | no | yes | yes
4109 Arguments : none
4110
4111 When this option is set, haproxy will systematically send a special header
4112 "X-Haproxy-Server-State" with a list of parameters indicating to each server
4113 how they are seen by haproxy. This can be used for instance when a server is
4114 manipulated without access to haproxy and the operator needs to know whether
4115 haproxy still sees it up or not, or if the server is the last one in a farm.
4116
4117 The header is composed of fields delimited by semi-colons, the first of which
4118 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
4119 checks on the total number before transition, just as appears in the stats
4120 interface. Next headers are in the form "<variable>=<value>", indicating in
4121 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08004122 - a variable "address", containing the address of the backend server.
4123 This corresponds to the <address> field in the server declaration. For
4124 unix domain sockets, it will read "unix".
4125
4126 - a variable "port", containing the port of the backend server. This
4127 corresponds to the <port> field in the server declaration. For unix
4128 domain sockets, it will read "unix".
4129
Willy Tarreauef781042010-01-27 11:53:01 +01004130 - a variable "name", containing the name of the backend followed by a slash
4131 ("/") then the name of the server. This can be used when a server is
4132 checked in multiple backends.
4133
4134 - a variable "node" containing the name of the haproxy node, as set in the
4135 global "node" variable, otherwise the system's hostname if unspecified.
4136
4137 - a variable "weight" indicating the weight of the server, a slash ("/")
4138 and the total weight of the farm (just counting usable servers). This
4139 helps to know if other servers are available to handle the load when this
4140 one fails.
4141
4142 - a variable "scur" indicating the current number of concurrent connections
4143 on the server, followed by a slash ("/") then the total number of
4144 connections on all servers of the same backend.
4145
4146 - a variable "qcur" indicating the current number of requests in the
4147 server's queue.
4148
4149 Example of a header received by the application server :
4150 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4151 scur=13/22; qcur=0
4152
4153 See also : "option httpchk", "http-check disable-on-404"
4154
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004155
4156http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004157 Access control for Layer 7 requests
4158
4159 May be used in sections: defaults | frontend | listen | backend
4160 no | yes | yes | yes
4161
Willy Tarreau20b0de52012-12-24 15:45:22 +01004162 The http-request statement defines a set of rules which apply to layer 7
4163 processing. The rules are evaluated in their declaration order when they are
4164 met in a frontend, listen or backend section. Any rule may optionally be
4165 followed by an ACL-based condition, in which case it will only be evaluated
4166 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004167
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004168 The first keyword is the rule's action. The supported actions are described
4169 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004170
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004171 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004172
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004173 Example:
4174 acl nagios src 192.168.129.3
4175 acl local_net src 192.168.0.0/16
4176 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004177
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004178 http-request allow if nagios
4179 http-request allow if local_net auth_ok
4180 http-request auth realm Gimme if local_net auth_ok
4181 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004182
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004183 Example:
4184 acl key req.hdr(X-Add-Acl-Key) -m found
4185 acl add path /addacl
4186 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004187
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004188 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004189
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004190 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4191 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004192
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004193 Example:
4194 acl value req.hdr(X-Value) -m found
4195 acl setmap path /setmap
4196 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004197
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004198 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004199
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004200 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4201 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004202
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004203 See also : "stats http-request", section 3.4 about userlists and section 7
4204 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004205
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004206http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004207
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004208 This is used to add a new entry into an ACL. The ACL must be loaded from a
4209 file (even a dummy empty file). The file name of the ACL to be updated is
4210 passed between parentheses. It takes one argument: <key fmt>, which follows
4211 log-format rules, to collect content of the new entry. It performs a lookup
4212 in the ACL before insertion, to avoid duplicated (or more) values. This
4213 lookup is done by a linear search and can be expensive with large lists!
4214 It is the equivalent of the "add acl" command from the stats socket, but can
4215 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004216
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004217http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004218
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004219 This appends an HTTP header field whose name is specified in <name> and
4220 whose value is defined by <fmt> which follows the log-format rules (see
4221 Custom Log Format in section 8.2.4). This is particularly useful to pass
4222 connection-specific information to the server (e.g. the client's SSL
4223 certificate), or to combine several headers into one. This rule is not
4224 final, so it is possible to add other similar rules. Note that header
4225 addition is performed immediately, so one rule might reuse the resulting
4226 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004227
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004228http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004229
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004230 This stops the evaluation of the rules and lets the request pass the check.
4231 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004232
Sasha Pachev218f0642014-06-16 12:05:59 -06004233
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004234http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004235
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004236 This stops the evaluation of the rules and immediately responds with an
4237 HTTP 401 or 407 error code to invite the user to present a valid user name
4238 and password. No further "http-request" rules are evaluated. An optional
4239 "realm" parameter is supported, it sets the authentication realm that is
4240 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004241
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004242 Example:
4243 acl auth_ok http_auth_group(L1) G1
4244 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004245
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004246http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004247
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004248 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004249
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004250http-request capture <sample> [ len <length> | id <id> ]
4251 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004252
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004253 This captures sample expression <sample> from the request buffer, and
4254 converts it to a string of at most <len> characters. The resulting string is
4255 stored into the next request "capture" slot, so it will possibly appear next
4256 to some captured HTTP headers. It will then automatically appear in the logs,
4257 and it will be possible to extract it using sample fetch rules to feed it
4258 into headers or anything. The length should be limited given that this size
4259 will be allocated for each capture during the whole session life.
4260 Please check section 7.3 (Fetching samples) and "capture request header" for
4261 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004262
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004263 If the keyword "id" is used instead of "len", the action tries to store the
4264 captured string in a previously declared capture slot. This is useful to run
4265 captures in backends. The slot id can be declared by a previous directive
4266 "http-request capture" or with the "declare capture" keyword. If the slot
4267 <id> doesn't exist, then HAProxy fails parsing the configuration to prevent
4268 unexpected behavior at run time.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004269
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004270http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004271
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004272 This is used to delete an entry from an ACL. The ACL must be loaded from a
4273 file (even a dummy empty file). The file name of the ACL to be updated is
4274 passed between parentheses. It takes one argument: <key fmt>, which follows
4275 log-format rules, to collect content of the entry to delete.
4276 It is the equivalent of the "del acl" command from the stats socket, but can
4277 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004278
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004279http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004280
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004281 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004282
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004283http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004284
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004285 This is used to delete an entry from a MAP. The MAP must be loaded from a
4286 file (even a dummy empty file). The file name of the MAP to be updated is
4287 passed between parentheses. It takes one argument: <key fmt>, which follows
4288 log-format rules, to collect content of the entry to delete.
4289 It takes one argument: "file name" It is the equivalent of the "del map"
4290 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004291
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004292http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004293
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004294 This stops the evaluation of the rules and immediately rejects the request
4295 and emits an HTTP 403 error, or optionally the status code specified as an
4296 argument to "deny_status". The list of permitted status codes is limited to
4297 those that can be overridden by the "errorfile" directive.
4298 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004299
Olivier Houchard602bf7d2019-05-10 13:59:15 +02004300http-request disable-l7-retry [ { if | unless } <condition> ]
4301 This disables any attempt to retry the request if it fails for any other
4302 reason than a connection failure. This can be useful for example to make
4303 sure POST requests aren't retried on failure.
4304
Baptiste Assmann333939c2019-01-21 08:34:50 +01004305http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
4306
4307 This action performs a DNS resolution of the output of <expr> and stores
4308 the result in the variable <var>. It uses the DNS resolvers section
4309 pointed by <resolvers>.
4310 It is possible to choose a resolution preference using the optional
4311 arguments 'ipv4' or 'ipv6'.
4312 When performing the DNS resolution, the client side connection is on
4313 pause waiting till the end of the resolution.
4314 If an IP address can be found, it is stored into <var>. If any kind of
4315 error occurs, then <var> is not set.
4316 One can use this action to discover a server IP address at run time and
4317 based on information found in the request (IE a Host header).
4318 If this action is used to find the server's IP address (using the
4319 "set-dst" action), then the server IP address in the backend must be set
4320 to 0.0.0.0.
4321
4322 Example:
4323 resolvers mydns
4324 nameserver local 127.0.0.53:53
4325 nameserver google 8.8.8.8:53
4326 timeout retry 1s
4327 hold valid 10s
4328 hold nx 3s
4329 hold other 3s
4330 hold obsolete 0s
4331 accepted_payload_size 8192
4332
4333 frontend fe
4334 bind 10.42.0.1:80
4335 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
4336 http-request capture var(txn.myip) len 40
4337
4338 # return 503 when the variable is not set,
4339 # which mean DNS resolution error
4340 use_backend b_503 unless { var(txn.myip) -m found }
4341
4342 default_backend be
4343
4344 backend b_503
4345 # dummy backend used to return 503.
4346 # one can use the errorfile directive to send a nice
4347 # 503 error page to end users
4348
4349 backend be
4350 # rule to prevent HAProxy from reconnecting to services
4351 # on the local network (forged DNS name used to scan the network)
4352 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
4353 http-request set-dst var(txn.myip)
4354 server clear 0.0.0.0:0
4355
4356 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
4357 be used to scan the network or worst won't loop over itself...
4358
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004359http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4360
4361 This is used to build an HTTP 103 Early Hints response prior to any other one.
4362 This appends an HTTP header field to this response whose name is specified in
4363 <name> and whose value is defined by <fmt> which follows the log-format rules
4364 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004365 to the client some Link headers to preload resources required to render the
4366 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004367
4368 See RFC 8297 for more information.
4369
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004370http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004371
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004372 This performs an HTTP redirection based on a redirect rule. This is exactly
4373 the same as the "redirect" statement except that it inserts a redirect rule
4374 which can be processed in the middle of other "http-request" rules and that
4375 these rules use the "log-format" strings. See the "redirect" keyword for the
4376 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004377
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004378http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004379
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004380 This stops the evaluation of the rules and immediately closes the connection
4381 without sending any response. It acts similarly to the
4382 "tcp-request content reject" rules. It can be useful to force an immediate
4383 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004384
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004385http-request replace-header <name> <match-regex> <replace-fmt>
4386 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004387
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004388 This matches the value of all occurences of header field <name> against
4389 <match-regex>. Matching is performed case-sensitively. Matching values are
4390 completely replaced by <replace-fmt>. Format characters are allowed in
4391 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
4392 Standard back-references using the backslash ('\') followed by a number are
4393 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004394
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004395 This action acts on whole header lines, regardless of the number of values
4396 they may contain. Thus it is well-suited to process headers naturally
4397 containing commas in their value, such as If-Modified-Since. Headers that
4398 contain a comma-separated list of values, such as Accept, should be processed
4399 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01004400
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004401 Example:
4402 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
4403
4404 # applied to:
4405 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4406
4407 # outputs:
4408 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4409
4410 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004411
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004412 http-request replace-header User-Agent curl foo
4413
4414 # applied to:
4415 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02004416
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004417 # outputs:
4418 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02004419
Willy Tarreau33810222019-06-12 17:44:02 +02004420http-request replace-uri <match-regex> <replace-fmt>
4421 [ { if | unless } <condition> ]
4422
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004423 This works like "replace-header" except that it works on the request's URI part
4424 instead of a header. The URI part may contain an optional scheme, authority or
4425 query string. These are considered to be part of the value that is matched
4426 against.
4427
4428 It is worth noting that regular expressions may be more expensive to evaluate
4429 than certain ACLs, so rare replacements may benefit from a condition to avoid
4430 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02004431
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004432 Example:
4433 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4434 http-request replace-uri (.*) /foo\1
Willy Tarreau33810222019-06-12 17:44:02 +02004435
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004436 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
4437 http-request replace-uri ([^?]*)(\?(.*))? \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02004438
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004439 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
4440 http-request replace-uri /foo/(.*) /\1
4441 # or more efficient if only some requests match :
4442 http-request replace-uri /foo/(.*) /\1 if { url_beg /foo/ }
Willy Tarreau33810222019-06-12 17:44:02 +02004443
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004444http-request replace-value <name> <match-regex> <replace-fmt>
4445 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004446
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004447 This works like "replace-header" except that it matches the regex against
4448 every comma-delimited value of the header field <name> instead of the
4449 entire header. This is suited for all headers which are allowed to carry
4450 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004451
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004452 Example:
4453 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004454
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004455 # applied to:
4456 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004457
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004458 # outputs:
4459 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004460
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004461http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4462http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004463
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004464 This actions increments the GPC0 or GPC1 counter according with the sticky
4465 counter designated by <sc-id>. If an error occurs, this action silently fails
4466 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004467
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004468http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004469
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004470 This action sets the GPT0 tag according to the sticky counter designated by
4471 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4472 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004473
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004474http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004475
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004476 This is used to set the destination IP address to the value of specified
4477 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4478 but provides the correct IP in a HTTP header; or you want to mask the IP for
4479 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4480 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004481
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004482 Arguments:
4483 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4484 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004485
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004486 Example:
4487 http-request set-dst hdr(x-dst)
4488 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004489
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004490 When possible, set-dst preserves the original destination port as long as the
4491 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004492
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004493http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004494
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004495 This is used to set the destination port address to the value of specified
4496 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4497 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004498
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004499 Arguments:
4500 <expr> Is a standard HAProxy expression formed by a sample-fetch
4501 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004502
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004503 Example:
4504 http-request set-dst-port hdr(x-port)
4505 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004506
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004507 When possible, set-dst-port preserves the original destination address as
4508 long as the address family supports a port, otherwise it forces the
4509 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004510
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004511http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004512
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004513 This does the same as "http-request add-header" except that the header name
4514 is first removed if it existed. This is useful when passing security
4515 information to the server, where the header must not be manipulated by
4516 external users. Note that the new value is computed before the removal so it
4517 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004518
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004519 Example:
4520 http-request set-header X-Haproxy-Current-Date %T
4521 http-request set-header X-SSL %[ssl_fc]
4522 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4523 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4524 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4525 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4526 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4527 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4528 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004529
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004530http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004531
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004532 This is used to change the log level of the current request when a certain
4533 condition is met. Valid levels are the 8 syslog levels (see the "log"
4534 keyword) plus the special level "silent" which disables logging for this
4535 request. This rule is not final so the last matching rule wins. This rule
4536 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004537
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004538http-request set-map(<file-name>) <key fmt> <value fmt>
4539 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004540
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004541 This is used to add a new entry into a MAP. The MAP must be loaded from a
4542 file (even a dummy empty file). The file name of the MAP to be updated is
4543 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4544 log-format rules, used to collect MAP key, and <value fmt>, which follows
4545 log-format rules, used to collect content for the new entry.
4546 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4547 more) values. This lookup is done by a linear search and can be expensive
4548 with large lists! It is the equivalent of the "set map" command from the
4549 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004550
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004551http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004552
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004553 This is used to set the Netfilter MARK on all packets sent to the client to
4554 the value passed in <mark> on platforms which support it. This value is an
4555 unsigned 32 bit value which can be matched by netfilter and by the routing
4556 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4557 "0x"). This can be useful to force certain packets to take a different route
4558 (for example a cheaper network path for bulk downloads). This works on Linux
4559 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004560
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004561http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004562
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004563 This rewrites the request method with the result of the evaluation of format
4564 string <fmt>. There should be very few valid reasons for having to do so as
4565 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004566
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004567http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004568
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004569 This sets the "nice" factor of the current request being processed. It only
4570 has effect against the other requests being processed at the same time.
4571 The default value is 0, unless altered by the "nice" setting on the "bind"
4572 line. The accepted range is -1024..1024. The higher the value, the nicest
4573 the request will be. Lower values will make the request more important than
4574 other ones. This can be useful to improve the speed of some requests, or
4575 lower the priority of non-important requests. Using this setting without
4576 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004577
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004578http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004579
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004580 This rewrites the request path with the result of the evaluation of format
4581 string <fmt>. The query string, if any, is left intact. If a scheme and
4582 authority is found before the path, they are left intact as well. If the
4583 request doesn't have a path ("*"), this one is replaced with the format.
4584 This can be used to prepend a directory component in front of a path for
4585 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004586
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004587 Example :
4588 # prepend the host name before the path
4589 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004590
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004591http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004592
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004593 This is used to set the queue priority class of the current request.
4594 The value must be a sample expression which converts to an integer in the
4595 range -2047..2047. Results outside this range will be truncated.
4596 The priority class determines the order in which queued requests are
4597 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004598
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004599http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004600
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004601 This is used to set the queue priority timestamp offset of the current
4602 request. The value must be a sample expression which converts to an integer
4603 in the range -524287..524287. Results outside this range will be truncated.
4604 When a request is queued, it is ordered first by the priority class, then by
4605 the current timestamp adjusted by the given offset in milliseconds. Lower
4606 values have higher priority.
4607 Note that the resulting timestamp is is only tracked with enough precision
4608 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4609 adjusted timestamp exceeds this value, it will be misidentified as highest
4610 priority. Thus it is important to set "timeout queue" to a value, where when
4611 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004612
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004613http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004614
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004615 This rewrites the request's query string which appears after the first
4616 question mark ("?") with the result of the evaluation of format string <fmt>.
4617 The part prior to the question mark is left intact. If the request doesn't
4618 contain a question mark and the new value is not empty, then one is added at
4619 the end of the URI, followed by the new value. If a question mark was
4620 present, it will never be removed even if the value is empty. This can be
4621 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004622
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004623 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004624
4625 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004626 # replace "%3D" with "=" in the query string
4627 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004628
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004629http-request set-src <expr> [ { if | unless } <condition> ]
4630 This is used to set the source IP address to the value of specified
4631 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4632 provides the correct IP in a HTTP header; or you want to mask source IP for
4633 privacy.
4634
4635 Arguments :
4636 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4637 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004638
Cyril Bonté78caf842010-03-10 22:41:43 +01004639 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004640 http-request set-src hdr(x-forwarded-for)
4641 http-request set-src src,ipmask(24)
4642
4643 When possible, set-src preserves the original source port as long as the
4644 address family allows it, otherwise the source port is set to 0.
4645
4646http-request set-src-port <expr> [ { if | unless } <condition> ]
4647
4648 This is used to set the source port address to the value of specified
4649 expression.
4650
4651 Arguments:
4652 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4653 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004654
Willy Tarreau20b0de52012-12-24 15:45:22 +01004655 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004656 http-request set-src-port hdr(x-port)
4657 http-request set-src-port int(4000)
4658
4659 When possible, set-src-port preserves the original source address as long as
4660 the address family supports a port, otherwise it forces the source address to
4661 IPv4 "0.0.0.0" before rewriting the port.
4662
4663http-request set-tos <tos> [ { if | unless } <condition> ]
4664
4665 This is used to set the TOS or DSCP field value of packets sent to the client
4666 to the value passed in <tos> on platforms which support this. This value
4667 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4668 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4669 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4670 be used to adjust some routing behavior on border routers based on some
4671 information from the request.
4672
4673 See RFC 2474, 2597, 3260 and 4594 for more information.
4674
4675http-request set-uri <fmt> [ { if | unless } <condition> ]
4676
4677 This rewrites the request URI with the result of the evaluation of format
4678 string <fmt>. The scheme, authority, path and query string are all replaced
4679 at once. This can be used to rewrite hosts in front of proxies, or to
4680 perform complex modifications to the URI such as moving parts between the
4681 path and the query string.
4682 See also "http-request set-path" and "http-request set-query".
4683
4684http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4685
4686 This is used to set the contents of a variable. The variable is declared
4687 inline.
4688
4689 Arguments:
4690 <var-name> The name of the variable starts with an indication about its
4691 scope. The scopes allowed are:
4692 "proc" : the variable is shared with the whole process
4693 "sess" : the variable is shared with the whole session
4694 "txn" : the variable is shared with the transaction
4695 (request and response)
4696 "req" : the variable is shared only during request
4697 processing
4698 "res" : the variable is shared only during response
4699 processing
4700 This prefix is followed by a name. The separator is a '.'.
4701 The name may only contain characters 'a-z', 'A-Z', '0-9'
4702 and '_'.
4703
4704 <expr> Is a standard HAProxy expression formed by a sample-fetch
4705 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004706
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004707 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004708 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004709
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004710http-request send-spoe-group <engine-name> <group-name>
4711 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004712
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004713 This action is used to trigger sending of a group of SPOE messages. To do so,
4714 the SPOE engine used to send messages must be defined, as well as the SPOE
4715 group to send. Of course, the SPOE engine must refer to an existing SPOE
4716 filter. If not engine name is provided on the SPOE filter line, the SPOE
4717 agent name must be used.
4718
4719 Arguments:
4720 <engine-name> The SPOE engine name.
4721
4722 <group-name> The SPOE group name as specified in the engine
4723 configuration.
4724
4725http-request silent-drop [ { if | unless } <condition> ]
4726
4727 This stops the evaluation of the rules and makes the client-facing connection
4728 suddenly disappear using a system-dependent way that tries to prevent the
4729 client from being notified. The effect it then that the client still sees an
4730 established connection while there's none on HAProxy. The purpose is to
4731 achieve a comparable effect to "tarpit" except that it doesn't use any local
4732 resource at all on the machine running HAProxy. It can resist much higher
4733 loads than "tarpit", and slow down stronger attackers. It is important to
4734 understand the impact of using this mechanism. All stateful equipment placed
4735 between the client and HAProxy (firewalls, proxies, load balancers) will also
4736 keep the established connection for a long time and may suffer from this
4737 action.
4738 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4739 option is used to block the emission of a TCP reset. On other systems, the
4740 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4741 router, though it's still delivered to local networks. Do not use it unless
4742 you fully understand how it works.
4743
4744http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4745
4746 This stops the evaluation of the rules and immediately blocks the request
4747 without responding for a delay specified by "timeout tarpit" or
4748 "timeout connect" if the former is not set. After that delay, if the client
4749 is still connected, an HTTP error 500 (or optionally the status code
4750 specified as an argument to "deny_status") is returned so that the client
4751 does not suspect it has been tarpitted. Logs will report the flags "PT".
4752 The goal of the tarpit rule is to slow down robots during an attack when
4753 they're limited on the number of concurrent requests. It can be very
4754 efficient against very dumb robots, and will significantly reduce the load
4755 on firewalls compared to a "deny" rule. But when facing "correctly"
4756 developed robots, it can make things worse by forcing haproxy and the front
4757 firewall to support insane number of concurrent connections.
4758 See also the "silent-drop" action.
4759
4760http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4761http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4762http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4763
4764 This enables tracking of sticky counters from current request. These rules do
4765 not stop evaluation and do not change default action. The number of counters
4766 that may be simultaneously tracked by the same connection is set in
4767 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4768 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4769 "track-sc0" rule executed enables tracking of the counters of the specified
4770 table as the first set. The first "track-sc1" rule executed enables tracking
4771 of the counters of the specified table as the second set. The first
4772 "track-sc2" rule executed enables tracking of the counters of the specified
4773 table as the third set. It is a recommended practice to use the first set of
4774 counters for the per-frontend counters and the second set for the per-backend
4775 ones. But this is just a guideline, all may be used everywhere.
4776
4777 Arguments :
4778 <key> is mandatory, and is a sample expression rule as described in
4779 section 7.3. It describes what elements of the incoming request or
4780 connection will be analyzed, extracted, combined, and used to
4781 select which table entry to update the counters.
4782
4783 <table> is an optional table to be used instead of the default one, which
4784 is the stick-table declared in the current proxy. All the counters
4785 for the matches and updates for the key will then be performed in
4786 that table until the session ends.
4787
4788 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4789 it is not found, an entry is allocated for it. Then a pointer to that entry
4790 is kept during all the session's life, and this entry's counters are updated
4791 as often as possible, every time the session's counters are updated, and also
4792 systematically when the session ends. Counters are only updated for events
4793 that happen after the tracking has been started. As an exception, connection
4794 counters and request counters are systematically updated so that they reflect
4795 useful information.
4796
4797 If the entry tracks concurrent connection counters, one connection is counted
4798 for as long as the entry is tracked, and the entry will not expire during
4799 that time. Tracking counters also provides a performance advantage over just
4800 checking the keys, because only one table lookup is performed for all ACL
4801 checks that make use of it.
4802
4803http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4804
4805 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004806
4807 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004808 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004809
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004810http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004811
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004812 This will delay the processing of the request until the SSL handshake
4813 happened. This is mostly useful to delay processing early data until we're
4814 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004815
Willy Tarreauef781042010-01-27 11:53:01 +01004816
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004817http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004818 Access control for Layer 7 responses
4819
4820 May be used in sections: defaults | frontend | listen | backend
4821 no | yes | yes | yes
4822
4823 The http-response statement defines a set of rules which apply to layer 7
4824 processing. The rules are evaluated in their declaration order when they are
4825 met in a frontend, listen or backend section. Any rule may optionally be
4826 followed by an ACL-based condition, in which case it will only be evaluated
4827 if the condition is true. Since these rules apply on responses, the backend
4828 rules are applied first, followed by the frontend's rules.
4829
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004830 The first keyword is the rule's action. The supported actions are described
4831 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004832
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004833 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004834
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004835 Example:
4836 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004837
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004838 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06004839
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004840 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4841 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06004842
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004843 Example:
4844 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06004845
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004846 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004847
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004848 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4849 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06004850
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004851 See also : "http-request", section 3.4 about userlists and section 7 about
4852 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004853
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004854http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004855
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004856 This is used to add a new entry into an ACL. The ACL must be loaded from a
4857 file (even a dummy empty file). The file name of the ACL to be updated is
4858 passed between parentheses. It takes one argument: <key fmt>, which follows
4859 log-format rules, to collect content of the new entry. It performs a lookup
4860 in the ACL before insertion, to avoid duplicated (or more) values.
4861 This lookup is done by a linear search and can be expensive with large lists!
4862 It is the equivalent of the "add acl" command from the stats socket, but can
4863 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06004864
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004865http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004866
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004867 This appends an HTTP header field whose name is specified in <name> and whose
4868 value is defined by <fmt> which follows the log-format rules (see Custom Log
4869 Format in section 8.2.4). This may be used to send a cookie to a client for
4870 example, or to pass some internal information.
4871 This rule is not final, so it is possible to add other similar rules.
4872 Note that header addition is performed immediately, so one rule might reuse
4873 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004874
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004875http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004876
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004877 This stops the evaluation of the rules and lets the response pass the check.
4878 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06004879
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004880http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004881
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004882 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06004883
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004884http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004885
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004886 This captures sample expression <sample> from the response buffer, and
4887 converts it to a string. The resulting string is stored into the next request
4888 "capture" slot, so it will possibly appear next to some captured HTTP
4889 headers. It will then automatically appear in the logs, and it will be
4890 possible to extract it using sample fetch rules to feed it into headers or
4891 anything. Please check section 7.3 (Fetching samples) and
4892 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004893
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004894 The keyword "id" is the id of the capture slot which is used for storing the
4895 string. The capture slot must be defined in an associated frontend.
4896 This is useful to run captures in backends. The slot id can be declared by a
4897 previous directive "http-response capture" or with the "declare capture"
4898 keyword.
4899 If the slot <id> doesn't exist, then HAProxy fails parsing the configuration
4900 to prevent unexpected behavior at run time.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004901
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004902http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004903
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004904 This is used to delete an entry from an ACL. The ACL must be loaded from a
4905 file (even a dummy empty file). The file name of the ACL to be updated is
4906 passed between parentheses. It takes one argument: <key fmt>, which follows
4907 log-format rules, to collect content of the entry to delete.
4908 It is the equivalent of the "del acl" command from the stats socket, but can
4909 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004910
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004911http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004912
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004913 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004914
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004915http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02004916
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004917 This is used to delete an entry from a MAP. The MAP must be loaded from a
4918 file (even a dummy empty file). The file name of the MAP to be updated is
4919 passed between parentheses. It takes one argument: <key fmt>, which follows
4920 log-format rules, to collect content of the entry to delete.
4921 It takes one argument: "file name" It is the equivalent of the "del map"
4922 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004923
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004924http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004925
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004926 This stops the evaluation of the rules and immediately rejects the response
4927 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004928
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004929http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004930
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004931 This performs an HTTP redirection based on a redirect rule.
4932 This supports a format string similarly to "http-request redirect" rules,
4933 with the exception that only the "location" type of redirect is possible on
4934 the response. See the "redirect" keyword for the rule's syntax. When a
4935 redirect rule is applied during a response, connections to the server are
4936 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004937
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004938http-response replace-header <name> <regex-match> <replace-fmt>
4939 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004940
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004941 This works like "http-request replace-header" except that it works on the
4942 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01004943
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004944 Example:
4945 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02004946
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004947 # applied to:
4948 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004949
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004950 # outputs:
4951 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 +02004952
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004953 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004954
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004955http-response replace-value <name> <regex-match> <replace-fmt>
4956 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004957
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004958 This works like "http-response replace-value" except that it works on the
4959 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004960
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004961 Example:
4962 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004963
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004964 # applied to:
4965 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01004966
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004967 # outputs:
4968 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004969
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004970http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4971http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004972
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004973 This action increments the GPC0 or GPC1 counter according with the sticky
4974 counter designated by <sc-id>. If an error occurs, this action silently fails
4975 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004976
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004977http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004978
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004979 This action sets the GPT0 tag according to the sticky counter designated by
4980 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4981 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004982
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004983http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02004984
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004985 This action is used to trigger sending of a group of SPOE messages. To do so,
4986 the SPOE engine used to send messages must be defined, as well as the SPOE
4987 group to send. Of course, the SPOE engine must refer to an existing SPOE
4988 filter. If not engine name is provided on the SPOE filter line, the SPOE
4989 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004990
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004991 Arguments:
4992 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004993
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004994 <group-name> The SPOE group name as specified in the engine
4995 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004996
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004997http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004998
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004999 This does the same as "add-header" except that the header name is first
5000 removed if it existed. This is useful when passing security information to
5001 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005002
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005003http-response set-log-level <level> [ { if | unless } <condition> ]
5004
5005 This is used to change the log level of the current request when a certain
5006 condition is met. Valid levels are the 8 syslog levels (see the "log"
5007 keyword) plus the special level "silent" which disables logging for this
5008 request. This rule is not final so the last matching rule wins. This rule can
5009 be useful to disable health checks coming from another equipment.
5010
5011http-response set-map(<file-name>) <key fmt> <value fmt>
5012
5013 This is used to add a new entry into a MAP. The MAP must be loaded from a
5014 file (even a dummy empty file). The file name of the MAP to be updated is
5015 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5016 log-format rules, used to collect MAP key, and <value fmt>, which follows
5017 log-format rules, used to collect content for the new entry. It performs a
5018 lookup in the MAP before insertion, to avoid duplicated (or more) values.
5019 This lookup is done by a linear search and can be expensive with large lists!
5020 It is the equivalent of the "set map" command from the stats socket, but can
5021 be triggered by an HTTP response.
5022
5023http-response set-mark <mark> [ { if | unless } <condition> ]
5024
5025 This is used to set the Netfilter MARK on all packets sent to the client to
5026 the value passed in <mark> on platforms which support it. This value is an
5027 unsigned 32 bit value which can be matched by netfilter and by the routing
5028 table. It can be expressed both in decimal or hexadecimal format (prefixed
5029 by "0x"). This can be useful to force certain packets to take a different
5030 route (for example a cheaper network path for bulk downloads). This works on
5031 Linux kernels 2.6.32 and above and requires admin privileges.
5032
5033http-response set-nice <nice> [ { if | unless } <condition> ]
5034
5035 This sets the "nice" factor of the current request being processed.
5036 It only has effect against the other requests being processed at the same
5037 time. The default value is 0, unless altered by the "nice" setting on the
5038 "bind" line. The accepted range is -1024..1024. The higher the value, the
5039 nicest the request will be. Lower values will make the request more important
5040 than other ones. This can be useful to improve the speed of some requests, or
5041 lower the priority of non-important requests. Using this setting without
5042 prior experimentation can cause some major slowdown.
5043
5044http-response set-status <status> [reason <str>]
5045 [ { if | unless } <condition> ]
5046
5047 This replaces the response status code with <status> which must be an integer
5048 between 100 and 999. Optionally, a custom reason text can be provided defined
5049 by <str>, or the default reason for the specified code will be used as a
5050 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005051
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005052 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005053 # return "431 Request Header Fields Too Large"
5054 http-response set-status 431
5055 # return "503 Slow Down", custom reason
5056 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005057
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005058http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005059
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005060 This is used to set the TOS or DSCP field value of packets sent to the client
5061 to the value passed in <tos> on platforms which support this.
5062 This value represents the whole 8 bits of the IP TOS field, and can be
5063 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
5064 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
5065 always 0. This can be used to adjust some routing behavior on border routers
5066 based on some information from the request.
5067
5068 See RFC 2474, 2597, 3260 and 4594 for more information.
5069
5070http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5071
5072 This is used to set the contents of a variable. The variable is declared
5073 inline.
5074
5075 Arguments:
5076 <var-name> The name of the variable starts with an indication about its
5077 scope. The scopes allowed are:
5078 "proc" : the variable is shared with the whole process
5079 "sess" : the variable is shared with the whole session
5080 "txn" : the variable is shared with the transaction
5081 (request and response)
5082 "req" : the variable is shared only during request
5083 processing
5084 "res" : the variable is shared only during response
5085 processing
5086 This prefix is followed by a name. The separator is a '.'.
5087 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5088 and '_'.
5089
5090 <expr> Is a standard HAProxy expression formed by a sample-fetch
5091 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005092
5093 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005094 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005095
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005096http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005097
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005098 This stops the evaluation of the rules and makes the client-facing connection
5099 suddenly disappear using a system-dependent way that tries to prevent the
5100 client from being notified. The effect it then that the client still sees an
5101 established connection while there's none on HAProxy. The purpose is to
5102 achieve a comparable effect to "tarpit" except that it doesn't use any local
5103 resource at all on the machine running HAProxy. It can resist much higher
5104 loads than "tarpit", and slow down stronger attackers. It is important to
5105 understand the impact of using this mechanism. All stateful equipment placed
5106 between the client and HAProxy (firewalls, proxies, load balancers) will also
5107 keep the established connection for a long time and may suffer from this
5108 action.
5109 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
5110 option is used to block the emission of a TCP reset. On other systems, the
5111 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
5112 router, though it's still delivered to local networks. Do not use it unless
5113 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005114
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005115http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
5116http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
5117http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005118
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005119 This enables tracking of sticky counters from current response. Please refer
5120 to "http-request track-sc" for a complete description. The only difference
5121 from "http-request track-sc" is the <key> sample expression can only make use
5122 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
5123 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
5124 supported, haproxy will fail and warn while parsing the config.
5125
5126http-response unset-var(<var-name>) [ { if | unless } <condition> ]
5127
5128 This is used to unset a variable. See "http-response set-var" for details
5129 about <var-name>.
5130
5131 Example:
5132 http-response unset-var(sess.last_redir)
5133
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02005134
Willy Tarreau30631952015-08-06 15:05:24 +02005135http-reuse { never | safe | aggressive | always }
5136 Declare how idle HTTP connections may be shared between requests
5137
5138 May be used in sections: defaults | frontend | listen | backend
5139 yes | no | yes | yes
5140
5141 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01005142 which is considered safe for reuse is moved back to the server's idle
5143 connections pool so that any other request can make use of it. This is the
5144 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02005145
5146 The argument indicates the desired connection reuse strategy :
5147
Olivier Houchard86006a52018-12-14 19:37:49 +01005148 - "never" : idle connections are never shared between sessions. This mode
5149 may be enforced to cancel a different strategy inherited from
5150 a defaults section or for troubleshooting. For example, if an
5151 old bogus application considers that multiple requests over
5152 the same connection come from the same client and it is not
5153 possible to fix the application, it may be desirable to
5154 disable connection sharing in a single backend. An example of
5155 such an application could be an old haproxy using cookie
5156 insertion in tunnel mode and not checking any request past the
5157 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02005158
Olivier Houchard86006a52018-12-14 19:37:49 +01005159 - "safe" : this is the default and the recommended strategy. The first
5160 request of a session is always sent over its own connection,
5161 and only subsequent requests may be dispatched over other
5162 existing connections. This ensures that in case the server
5163 closes the connection when the request is being sent, the
5164 browser can decide to silently retry it. Since it is exactly
5165 equivalent to regular keep-alive, there should be no side
5166 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02005167
5168 - "aggressive" : this mode may be useful in webservices environments where
5169 all servers are not necessarily known and where it would be
5170 appreciable to deliver most first requests over existing
5171 connections. In this case, first requests are only delivered
5172 over existing connections that have been reused at least once,
5173 proving that the server correctly supports connection reuse.
5174 It should only be used when it's sure that the client can
5175 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02005176 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02005177 downsides of rare connection failures.
5178
5179 - "always" : this mode is only recommended when the path to the server is
5180 known for never breaking existing connections quickly after
5181 releasing them. It allows the first request of a session to be
5182 sent to an existing connection. This can provide a significant
5183 performance increase over the "safe" strategy when the backend
5184 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005185 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02005186 sharing. It is recommended that the "http-keep-alive" timeout
5187 remains low in this mode so that no dead connections remain
5188 usable. In most cases, this will lead to the same performance
5189 gains as "aggressive" but with more risks. It should only be
5190 used when it improves the situation over "aggressive".
5191
5192 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005193 connection properties and compatibility. Specifically :
5194 - connections made with "usesrc" followed by a client-dependent value
5195 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005196
5197 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01005198 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005199
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02005200 - connections with certain bogus authentication schemes (relying on the
5201 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005202
5203 No connection pool is involved, once a session dies, the last idle connection
5204 it was attached to is deleted at the same time. This ensures that connections
5205 may not last after all sessions are closed.
5206
5207 Note: connection reuse improves the accuracy of the "server maxconn" setting,
5208 because almost no new connection will be established while idle connections
5209 remain available. This is particularly true with the "always" strategy.
5210
5211 See also : "option http-keep-alive", "server maxconn"
5212
5213
Mark Lamourinec2247f02012-01-04 13:02:01 -05005214http-send-name-header [<header>]
5215 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05005216 May be used in sections: defaults | frontend | listen | backend
5217 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05005218 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05005219 <header> The header string to use to send the server name
5220
Willy Tarreau81bef7e2019-10-07 14:58:02 +02005221 The "http-send-name-header" statement causes the header field named <header>
5222 to be set to the name of the target server at the moment the request is about
5223 to be sent on the wire. Any existing occurrences of this header are removed.
5224 Upon retries and redispatches, the header field is updated to always reflect
5225 the server being attempted to connect to. Given that this header is modified
5226 very late in the connection setup, it may have unexpected effects on already
5227 modified headers. For example using it with transport-level header such as
5228 connection, content-length, transfer-encoding and so on will likely result in
5229 invalid requests being sent to the server. Additionally it has been reported
5230 that this directive is currently being used as a way to overwrite the Host
5231 header field in outgoing requests; while this trick has been known to work
5232 as a side effect of the feature for some time, it is not officially supported
5233 and might possibly not work anymore in a future version depending on the
5234 technical difficulties this feature induces. A long-term solution instead
5235 consists in fixing the application which required this trick so that it binds
5236 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05005237
5238 See also : "server"
5239
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005240id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005241 Set a persistent ID to a proxy.
5242 May be used in sections : defaults | frontend | listen | backend
5243 no | yes | yes | yes
5244 Arguments : none
5245
5246 Set a persistent ID for the proxy. This ID must be unique and positive.
5247 An unused ID will automatically be assigned if unset. The first assigned
5248 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005249
5250
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005251ignore-persist { if | unless } <condition>
5252 Declare a condition to ignore persistence
5253 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005254 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005255
5256 By default, when cookie persistence is enabled, every requests containing
5257 the cookie are unconditionally persistent (assuming the target server is up
5258 and running).
5259
5260 The "ignore-persist" statement allows one to declare various ACL-based
5261 conditions which, when met, will cause a request to ignore persistence.
5262 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005263 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005264 persistence for a specific User-Agent (for example, some web crawler bots).
5265
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005266 The persistence is ignored when an "if" condition is met, or unless an
5267 "unless" condition is met.
5268
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005269 Example:
5270 acl url_static path_beg /static /images /img /css
5271 acl url_static path_end .gif .png .jpg .css .js
5272 ignore-persist if url_static
5273
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005274 See also : "force-persist", "cookie", and section 7 about ACL usage.
5275
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005276load-server-state-from-file { global | local | none }
5277 Allow seamless reload of HAProxy
5278 May be used in sections: defaults | frontend | listen | backend
5279 yes | no | yes | yes
5280
5281 This directive points HAProxy to a file where server state from previous
5282 running process has been saved. That way, when starting up, before handling
5283 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005284 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005285 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5286 loading state or load states from a file containing all backends and servers.
5287 The state file can be generated by running the command "show servers state"
5288 over the stats socket and redirect output.
5289
Davor Ocelice9ed2812017-12-25 17:49:28 +01005290 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005291 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005292 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005293
5294 Arguments:
5295 global load the content of the file pointed by the global directive
5296 named "server-state-file".
5297
5298 local load the content of the file pointed by the directive
5299 "server-state-file-name" if set. If not set, then the backend
5300 name is used as a file name.
5301
5302 none don't load any stat for this backend
5303
5304 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005305 - server's IP address is preserved across reloads by default, but the
5306 order can be changed thanks to the server's "init-addr" setting. This
5307 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005308 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005309 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005310
5311 - server's weight is applied from previous running process unless it has
5312 has changed between previous and new configuration files.
5313
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005314 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005315
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005316 global
5317 stats socket /tmp/socket
5318 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005319
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005320 defaults
5321 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005322
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005323 backend bk
5324 server s1 127.0.0.1:22 check weight 11
5325 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005326
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005327
5328 Then one can run :
5329
5330 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5331
5332 Content of the file /tmp/server_state would be like this:
5333
5334 1
5335 # <field names skipped for the doc example>
5336 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5337 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5338
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005339 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005340
5341 global
5342 stats socket /tmp/socket
5343 server-state-base /etc/haproxy/states
5344
5345 defaults
5346 load-server-state-from-file local
5347
5348 backend bk
5349 server s1 127.0.0.1:22 check weight 11
5350 server s2 127.0.0.1:22 check weight 12
5351
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005352
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005353 Then one can run :
5354
5355 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5356
5357 Content of the file /etc/haproxy/states/bk would be like this:
5358
5359 1
5360 # <field names skipped for the doc example>
5361 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5362 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5363
5364 See also: "server-state-file", "server-state-file-name", and
5365 "show servers state"
5366
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005367
Willy Tarreau2769aa02007-12-27 18:26:09 +01005368log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005369log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
5370 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005371no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005372 Enable per-instance logging of events and traffic.
5373 May be used in sections : defaults | frontend | listen | backend
5374 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005375
5376 Prefix :
5377 no should be used when the logger list must be flushed. For example,
5378 if you don't want to inherit from the default logger list. This
5379 prefix does not allow arguments.
5380
Willy Tarreau2769aa02007-12-27 18:26:09 +01005381 Arguments :
5382 global should be used when the instance's logging parameters are the
5383 same as the global ones. This is the most common usage. "global"
5384 replaces <address>, <facility> and <level> with those of the log
5385 entries found in the "global" section. Only one "log global"
5386 statement may be used per instance, and this form takes no other
5387 parameter.
5388
5389 <address> indicates where to send the logs. It takes the same format as
5390 for the "global" section's logs, and can be one of :
5391
5392 - An IPv4 address optionally followed by a colon (':') and a UDP
5393 port. If no port is specified, 514 is used by default (the
5394 standard syslog port).
5395
David du Colombier24bb5f52011-03-17 10:40:23 +01005396 - An IPv6 address followed by a colon (':') and optionally a UDP
5397 port. If no port is specified, 514 is used by default (the
5398 standard syslog port).
5399
Willy Tarreau2769aa02007-12-27 18:26:09 +01005400 - A filesystem path to a UNIX domain socket, keeping in mind
5401 considerations for chroot (be sure the path is accessible
5402 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005403 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005404
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005405 - A file descriptor number in the form "fd@<number>", which may
5406 point to a pipe, terminal, or socket. In this case unbuffered
5407 logs are used and one writev() call per log is performed. This
5408 is a bit expensive but acceptable for most workloads. Messages
5409 sent this way will not be truncated but may be dropped, in
5410 which case the DroppedLogs counter will be incremented. The
5411 writev() call is atomic even on pipes for messages up to
5412 PIPE_BUF size, which POSIX recommends to be at least 512 and
5413 which is 4096 bytes on most modern operating systems. Any
5414 larger message may be interleaved with messages from other
5415 processes. Exceptionally for debugging purposes the file
5416 descriptor may also be directed to a file, but doing so will
5417 significantly slow haproxy down as non-blocking calls will be
5418 ignored. Also there will be no way to purge nor rotate this
5419 file without restarting the process. Note that the configured
5420 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005421 with a TCP syslog server. See also the "short" and "raw"
5422 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005423
5424 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5425 and "fd@2", see above.
5426
Willy Tarreauc046d162019-08-30 15:24:59 +02005427 - A ring buffer in the form "ring@<name>", which will correspond
5428 to an in-memory ring buffer accessible over the CLI using the
5429 "show events" command, which will also list existing rings and
5430 their sizes. Such buffers are lost on reload or restart but
5431 when used as a complement this can help troubleshooting by
5432 having the logs instantly available.
5433
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005434 You may want to reference some environment variables in the
5435 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005436
Willy Tarreau18324f52014-06-27 18:10:07 +02005437 <length> is an optional maximum line length. Log lines larger than this
5438 value will be truncated before being sent. The reason is that
5439 syslog servers act differently on log line length. All servers
5440 support the default value of 1024, but some servers simply drop
5441 larger lines while others do log them. If a server supports long
5442 lines, it may make sense to set this value here in order to avoid
5443 truncating long lines. Similarly, if a server drops long lines,
5444 it is preferable to truncate them before sending them. Accepted
5445 values are 80 to 65535 inclusive. The default value of 1024 is
5446 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005447 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005448
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005449 <ranges> A list of comma-separated ranges to identify the logs to sample.
5450 This is used to balance the load of the logs to send to the log
5451 server. The limits of the ranges cannot be null. They are numbered
5452 from 1. The size or period (in number of logs) of the sample must
5453 be set with <sample_size> parameter.
5454
5455 <sample_size>
5456 The size of the sample in number of logs to consider when balancing
5457 their logging loads. It is used to balance the load of the logs to
5458 send to the syslog server. This size must be greater or equal to the
5459 maximum of the high limits of the ranges.
5460 (see also <ranges> parameter).
5461
Willy Tarreauadb345d2018-11-12 07:56:13 +01005462 <format> is the log format used when generating syslog messages. It may be
5463 one of the following :
5464
5465 rfc3164 The RFC3164 syslog message format. This is the default.
5466 (https://tools.ietf.org/html/rfc3164)
5467
5468 rfc5424 The RFC5424 syslog message format.
5469 (https://tools.ietf.org/html/rfc5424)
5470
Willy Tarreaue8746a02018-11-12 08:45:00 +01005471 short A message containing only a level between angle brackets such as
5472 '<3>', followed by the text. The PID, date, time, process name
5473 and system name are omitted. This is designed to be used with a
5474 local log server. This format is compatible with what the
5475 systemd logger consumes.
5476
Willy Tarreauc1b06452018-11-12 11:57:56 +01005477 raw A message containing only the text. The level, PID, date, time,
5478 process name and system name are omitted. This is designed to
5479 be used in containers or during development, where the severity
5480 only depends on the file descriptor used (stdout/stderr).
5481
Willy Tarreau2769aa02007-12-27 18:26:09 +01005482 <facility> must be one of the 24 standard syslog facilities :
5483
Willy Tarreaue8746a02018-11-12 08:45:00 +01005484 kern user mail daemon auth syslog lpr news
5485 uucp cron auth2 ftp ntp audit alert cron2
5486 local0 local1 local2 local3 local4 local5 local6 local7
5487
Willy Tarreauc1b06452018-11-12 11:57:56 +01005488 Note that the facility is ignored for the "short" and "raw"
5489 formats, but still required as a positional field. It is
5490 recommended to use "daemon" in this case to make it clear that
5491 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005492
5493 <level> is optional and can be specified to filter outgoing messages. By
5494 default, all messages are sent. If a level is specified, only
5495 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005496 will be sent. An optional minimum level can be specified. If it
5497 is set, logs emitted with a more severe level than this one will
5498 be capped to this level. This is used to avoid sending "emerg"
5499 messages on all terminals on some default syslog configurations.
5500 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005501
5502 emerg alert crit err warning notice info debug
5503
William Lallemand0f99e342011-10-12 17:50:54 +02005504 It is important to keep in mind that it is the frontend which decides what to
5505 log from a connection, and that in case of content switching, the log entries
5506 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005507
5508 However, backend log declaration define how and where servers status changes
5509 will be logged. Level "notice" will be used to indicate a server going up,
5510 "warning" will be used for termination signals and definitive service
5511 termination, and "alert" will be used for when a server goes down.
5512
5513 Note : According to RFC3164, messages are truncated to 1024 bytes before
5514 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005515
5516 Example :
5517 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005518 log stdout format short daemon # send log to systemd
5519 log stdout format raw daemon # send everything to stdout
5520 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005521 log 127.0.0.1:514 local0 notice # only send important events
5522 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005523 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005524
Willy Tarreau2769aa02007-12-27 18:26:09 +01005525
William Lallemand48940402012-01-30 16:47:22 +01005526log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005527 Specifies the log format string to use for traffic logs
5528 May be used in sections: defaults | frontend | listen | backend
5529 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005530
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005531 This directive specifies the log format string that will be used for all logs
5532 resulting from traffic passing through the frontend using this line. If the
5533 directive is used in a defaults section, all subsequent frontends will use
5534 the same log format. Please see section 8.2.4 which covers the log format
5535 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005536
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005537 "log-format" directive overrides previous "option tcplog", "log-format" and
5538 "option httplog" directives.
5539
Dragan Dosen7ad31542015-09-28 17:16:47 +02005540log-format-sd <string>
5541 Specifies the RFC5424 structured-data log format string
5542 May be used in sections: defaults | frontend | listen | backend
5543 yes | yes | yes | no
5544
5545 This directive specifies the RFC5424 structured-data log format string that
5546 will be used for all logs resulting from traffic passing through the frontend
5547 using this line. If the directive is used in a defaults section, all
5548 subsequent frontends will use the same log format. Please see section 8.2.4
5549 which covers the log format string in depth.
5550
5551 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5552 about the RFC5424 structured-data part.
5553
5554 Note : This log format string will be used only for loggers that have set
5555 log format to "rfc5424".
5556
5557 Example :
5558 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5559
5560
Willy Tarreau094af4e2015-01-07 15:03:42 +01005561log-tag <string>
5562 Specifies the log tag to use for all outgoing logs
5563 May be used in sections: defaults | frontend | listen | backend
5564 yes | yes | yes | yes
5565
5566 Sets the tag field in the syslog header to this string. It defaults to the
5567 log-tag set in the global section, otherwise the program name as launched
5568 from the command line, which usually is "haproxy". Sometimes it can be useful
5569 to differentiate between multiple processes running on the same host, or to
5570 differentiate customer instances running in the same process. In the backend,
5571 logs about servers up/down will use this tag. As a hint, it can be convenient
5572 to set a log-tag related to a hosted customer in a defaults section then put
5573 all the frontends and backends for that customer, then start another customer
5574 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005575
Willy Tarreauc35362a2014-04-25 13:58:37 +02005576max-keep-alive-queue <value>
5577 Set the maximum server queue size for maintaining keep-alive connections
5578 May be used in sections: defaults | frontend | listen | backend
5579 yes | no | yes | yes
5580
5581 HTTP keep-alive tries to reuse the same server connection whenever possible,
5582 but sometimes it can be counter-productive, for example if a server has a lot
5583 of connections while other ones are idle. This is especially true for static
5584 servers.
5585
5586 The purpose of this setting is to set a threshold on the number of queued
5587 connections at which haproxy stops trying to reuse the same server and prefers
5588 to find another one. The default value, -1, means there is no limit. A value
5589 of zero means that keep-alive requests will never be queued. For very close
5590 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005591 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005592 use a value of 10 or less). For remote servers suffering from a high latency,
5593 higher values might be needed to cover for the latency and/or the cost of
5594 picking a different server.
5595
5596 Note that this has no impact on responses which are maintained to the same
5597 server consecutively to a 401 response. They will still go to the same server
5598 even if they have to be queued.
5599
5600 See also : "option http-server-close", "option prefer-last-server", server
5601 "maxconn" and cookie persistence.
5602
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01005603max-session-srv-conns <nb>
5604 Set the maximum number of outgoing connections we can keep idling for a given
5605 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
5606 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02005607
Willy Tarreau2769aa02007-12-27 18:26:09 +01005608maxconn <conns>
5609 Fix the maximum number of concurrent connections on a frontend
5610 May be used in sections : defaults | frontend | listen | backend
5611 yes | yes | yes | no
5612 Arguments :
5613 <conns> is the maximum number of concurrent connections the frontend will
5614 accept to serve. Excess connections will be queued by the system
5615 in the socket's listen queue and will be served once a connection
5616 closes.
5617
5618 If the system supports it, it can be useful on big sites to raise this limit
5619 very high so that haproxy manages connection queues, instead of leaving the
5620 clients with unanswered connection attempts. This value should not exceed the
5621 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005622 of tune.bufsize (16kB by default) each, as well as some other data resulting
5623 in about 33 kB of RAM being consumed per established connection. That means
5624 that a medium system equipped with 1GB of RAM can withstand around
5625 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005626
5627 Also, when <conns> is set to large values, it is possible that the servers
5628 are not sized to accept such loads, and for this reason it is generally wise
5629 to assign them some reasonable connection limits.
5630
Willy Tarreauc8d5b952019-02-27 17:25:52 +01005631 When this value is set to zero, which is the default, the global "maxconn"
5632 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02005633
Willy Tarreau2769aa02007-12-27 18:26:09 +01005634 See also : "server", global section's "maxconn", "fullconn"
5635
5636
5637mode { tcp|http|health }
5638 Set the running mode or protocol of the instance
5639 May be used in sections : defaults | frontend | listen | backend
5640 yes | yes | yes | yes
5641 Arguments :
5642 tcp The instance will work in pure TCP mode. A full-duplex connection
5643 will be established between clients and servers, and no layer 7
5644 examination will be performed. This is the default mode. It
5645 should be used for SSL, SSH, SMTP, ...
5646
5647 http The instance will work in HTTP mode. The client request will be
5648 analyzed in depth before connecting to any server. Any request
5649 which is not RFC-compliant will be rejected. Layer 7 filtering,
5650 processing and switching will be possible. This is the mode which
5651 brings HAProxy most of its value.
5652
5653 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005654 to incoming connections and close the connection. Alternatively,
5655 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5656 instead. Nothing will be logged in either case. This mode is used
5657 to reply to external components health checks. This mode is
5658 deprecated and should not be used anymore as it is possible to do
5659 the same and even better by combining TCP or HTTP modes with the
5660 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005661
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005662 When doing content switching, it is mandatory that the frontend and the
5663 backend are in the same mode (generally HTTP), otherwise the configuration
5664 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005665
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005666 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005667 defaults http_instances
5668 mode http
5669
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005670 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005671
Willy Tarreau0ba27502007-12-24 16:55:16 +01005672
Cyril Bontéf0c60612010-02-06 14:44:47 +01005673monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005674 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005675 May be used in sections : defaults | frontend | listen | backend
5676 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005677 Arguments :
5678 if <cond> the monitor request will fail if the condition is satisfied,
5679 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005680 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005681 are met, for instance a low number of servers both in a
5682 backend and its backup.
5683
5684 unless <cond> the monitor request will succeed only if the condition is
5685 satisfied, and will fail otherwise. Such a condition may be
5686 based on a test on the presence of a minimum number of active
5687 servers in a list of backends.
5688
5689 This statement adds a condition which can force the response to a monitor
5690 request to report a failure. By default, when an external component queries
5691 the URI dedicated to monitoring, a 200 response is returned. When one of the
5692 conditions above is met, haproxy will return 503 instead of 200. This is
5693 very useful to report a site failure to an external component which may base
5694 routing advertisements between multiple sites on the availability reported by
5695 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005696 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5697 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005698
5699 Example:
5700 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005701 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005702 acl site_dead nbsrv(dynamic) lt 2
5703 acl site_dead nbsrv(static) lt 2
5704 monitor-uri /site_alive
5705 monitor fail if site_dead
5706
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005707 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005708
5709
5710monitor-net <source>
5711 Declare a source network which is limited to monitor requests
5712 May be used in sections : defaults | frontend | listen | backend
5713 yes | yes | yes | no
5714 Arguments :
5715 <source> is the source IPv4 address or network which will only be able to
5716 get monitor responses to any request. It can be either an IPv4
5717 address, a host name, or an address followed by a slash ('/')
5718 followed by a mask.
5719
5720 In TCP mode, any connection coming from a source matching <source> will cause
5721 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005722 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005723 forwarding the connection to a remote server.
5724
5725 In HTTP mode, a connection coming from a source matching <source> will be
5726 accepted, the following response will be sent without waiting for a request,
5727 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5728 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005729 running without forwarding the request to a backend server. Note that this
5730 response is sent in raw format, without any transformation. This is important
5731 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005732
Willy Tarreau82569f92012-09-27 23:48:56 +02005733 Monitor requests are processed very early, just after tcp-request connection
5734 ACLs which are the only ones able to block them. These connections are short
5735 lived and never wait for any data from the client. They cannot be logged, and
5736 it is the intended purpose. They are only used to report HAProxy's health to
5737 an upper component, nothing more. Please note that "monitor fail" rules do
5738 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005739
Willy Tarreau95cd2832010-03-04 23:36:33 +01005740 Last, please note that only one "monitor-net" statement can be specified in
5741 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005742
Willy Tarreau2769aa02007-12-27 18:26:09 +01005743 Example :
5744 # addresses .252 and .253 are just probing us.
5745 frontend www
5746 monitor-net 192.168.0.252/31
5747
5748 See also : "monitor fail", "monitor-uri"
5749
5750
5751monitor-uri <uri>
5752 Intercept a URI used by external components' monitor requests
5753 May be used in sections : defaults | frontend | listen | backend
5754 yes | yes | yes | no
5755 Arguments :
5756 <uri> is the exact URI which we want to intercept to return HAProxy's
5757 health status instead of forwarding the request.
5758
5759 When an HTTP request referencing <uri> will be received on a frontend,
5760 HAProxy will not forward it nor log it, but instead will return either
5761 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5762 conditions defined with "monitor fail". This is normally enough for any
5763 front-end HTTP probe to detect that the service is UP and running without
5764 forwarding the request to a backend server. Note that the HTTP method, the
5765 version and all headers are ignored, but the request must at least be valid
5766 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5767
Willy Tarreau721d8e02017-12-01 18:25:08 +01005768 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005769 and even before any "http-request". The only rulesets applied before are the
5770 tcp-request ones. They cannot be logged either, and it is the intended
5771 purpose. They are only used to report HAProxy's health to an upper component,
5772 nothing more. However, it is possible to add any number of conditions using
5773 "monitor fail" and ACLs so that the result can be adjusted to whatever check
5774 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005775
5776 Example :
5777 # Use /haproxy_test to report haproxy's status
5778 frontend www
5779 mode http
5780 monitor-uri /haproxy_test
5781
5782 See also : "monitor fail", "monitor-net"
5783
Willy Tarreau0ba27502007-12-24 16:55:16 +01005784
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005785option abortonclose
5786no option abortonclose
5787 Enable or disable early dropping of aborted requests pending in queues.
5788 May be used in sections : defaults | frontend | listen | backend
5789 yes | no | yes | yes
5790 Arguments : none
5791
5792 In presence of very high loads, the servers will take some time to respond.
5793 The per-instance connection queue will inflate, and the response time will
5794 increase respective to the size of the queue times the average per-session
5795 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005796 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005797 the queue, and slowing down other users, and the servers as well, because the
5798 request will eventually be served, then aborted at the first error
5799 encountered while delivering the response.
5800
5801 As there is no way to distinguish between a full STOP and a simple output
5802 close on the client side, HTTP agents should be conservative and consider
5803 that the client might only have closed its output channel while waiting for
5804 the response. However, this introduces risks of congestion when lots of users
5805 do the same, and is completely useless nowadays because probably no client at
5806 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005807 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005808 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005809 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005810 of being the single component to break rare but valid traffic is extremely
5811 low, which adds to the temptation to be able to abort a session early while
5812 still not served and not pollute the servers.
5813
Davor Ocelice9ed2812017-12-25 17:49:28 +01005814 In HAProxy, the user can choose the desired behavior using the option
5815 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005816 compliant and aborted requests will be served. But when the option is
5817 specified, a session with an incoming channel closed will be aborted while
5818 it is still possible, either pending in the queue for a connection slot, or
5819 during the connection establishment if the server has not yet acknowledged
5820 the connection request. This considerably reduces the queue size and the load
5821 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005822 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005823
5824 If this option has been enabled in a "defaults" section, it can be disabled
5825 in a specific instance by prepending the "no" keyword before it.
5826
5827 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5828
5829
Willy Tarreau4076a152009-04-02 15:18:36 +02005830option accept-invalid-http-request
5831no option accept-invalid-http-request
5832 Enable or disable relaxing of HTTP request parsing
5833 May be used in sections : defaults | frontend | listen | backend
5834 yes | yes | yes | no
5835 Arguments : none
5836
Willy Tarreau91852eb2015-05-01 13:26:00 +02005837 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005838 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005839 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005840 forbidden characters are essentially used to build attacks exploiting server
5841 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5842 server will emit invalid header names for whatever reason (configuration,
5843 implementation) and the issue will not be immediately fixed. In such a case,
5844 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005845 even if that does not make sense, by specifying this option. Similarly, the
5846 list of characters allowed to appear in a URI is well defined by RFC3986, and
5847 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5848 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005849 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005850 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005851 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5852 to pass through (no version specified) and multiple digits for both the major
5853 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005854
5855 This option should never be enabled by default as it hides application bugs
5856 and open security breaches. It should only be deployed after a problem has
5857 been confirmed.
5858
5859 When this option is enabled, erroneous header names will still be accepted in
5860 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005861 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5862 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005863 also helps confirming that the issue has been solved.
5864
5865 If this option has been enabled in a "defaults" section, it can be disabled
5866 in a specific instance by prepending the "no" keyword before it.
5867
5868 See also : "option accept-invalid-http-response" and "show errors" on the
5869 stats socket.
5870
5871
5872option accept-invalid-http-response
5873no option accept-invalid-http-response
5874 Enable or disable relaxing of HTTP response parsing
5875 May be used in sections : defaults | frontend | listen | backend
5876 yes | no | yes | yes
5877 Arguments : none
5878
Willy Tarreau91852eb2015-05-01 13:26:00 +02005879 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005880 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005881 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005882 forbidden characters are essentially used to build attacks exploiting server
5883 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5884 server will emit invalid header names for whatever reason (configuration,
5885 implementation) and the issue will not be immediately fixed. In such a case,
5886 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005887 even if that does not make sense, by specifying this option. This option also
5888 relaxes the test on the HTTP version format, it allows multiple digits for
5889 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005890
5891 This option should never be enabled by default as it hides application bugs
5892 and open security breaches. It should only be deployed after a problem has
5893 been confirmed.
5894
5895 When this option is enabled, erroneous header names will still be accepted in
5896 responses, but the complete response will be captured in order to permit
5897 later analysis using the "show errors" request on the UNIX stats socket.
5898 Doing this also helps confirming that the issue has been solved.
5899
5900 If this option has been enabled in a "defaults" section, it can be disabled
5901 in a specific instance by prepending the "no" keyword before it.
5902
5903 See also : "option accept-invalid-http-request" and "show errors" on the
5904 stats socket.
5905
5906
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005907option allbackups
5908no option allbackups
5909 Use either all backup servers at a time or only the first one
5910 May be used in sections : defaults | frontend | listen | backend
5911 yes | no | yes | yes
5912 Arguments : none
5913
5914 By default, the first operational backup server gets all traffic when normal
5915 servers are all down. Sometimes, it may be preferred to use multiple backups
5916 at once, because one will not be enough. When "option allbackups" is enabled,
5917 the load balancing will be performed among all backup servers when all normal
5918 ones are unavailable. The same load balancing algorithm will be used and the
5919 servers' weights will be respected. Thus, there will not be any priority
5920 order between the backup servers anymore.
5921
5922 This option is mostly used with static server farms dedicated to return a
5923 "sorry" page when an application is completely offline.
5924
5925 If this option has been enabled in a "defaults" section, it can be disabled
5926 in a specific instance by prepending the "no" keyword before it.
5927
5928
5929option checkcache
5930no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005931 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005932 May be used in sections : defaults | frontend | listen | backend
5933 yes | no | yes | yes
5934 Arguments : none
5935
5936 Some high-level frameworks set application cookies everywhere and do not
5937 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005938 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005939 high risk of session crossing or stealing between users traversing the same
5940 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005941 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005942
5943 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005944 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005945 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005946 response to check if there's a risk of caching a cookie on a client-side
5947 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005948 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005949 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005950 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5951 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005952 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005953 - all those that result from a request using a method other than GET, HEAD,
5954 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005955 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005956 - those with a 'Pragma: no-cache' header
5957 - those with a 'Cache-control: private' header
5958 - those with a 'Cache-control: no-store' header
5959 - those with a 'Cache-control: max-age=0' header
5960 - those with a 'Cache-control: s-maxage=0' header
5961 - those with a 'Cache-control: no-cache' header
5962 - those with a 'Cache-control: no-cache="set-cookie"' header
5963 - those with a 'Cache-control: no-cache="set-cookie,' header
5964 (allowing other fields after set-cookie)
5965
5966 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005967 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
5968 gateway". The session state shows "PH--" meaning that the proxy blocked the
5969 response during headers processing. Additionally, an alert will be sent in
5970 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005971
5972 Due to the high impact on the application, the application should be tested
5973 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005974 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005975 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005976
5977 If this option has been enabled in a "defaults" section, it can be disabled
5978 in a specific instance by prepending the "no" keyword before it.
5979
5980
5981option clitcpka
5982no option clitcpka
5983 Enable or disable the sending of TCP keepalive packets on the client side
5984 May be used in sections : defaults | frontend | listen | backend
5985 yes | yes | yes | no
5986 Arguments : none
5987
5988 When there is a firewall or any session-aware component between a client and
5989 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005990 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005991 components decides to expire a session which has remained idle for too long.
5992
5993 Enabling socket-level TCP keep-alives makes the system regularly send packets
5994 to the other end of the connection, leaving it active. The delay between
5995 keep-alive probes is controlled by the system only and depends both on the
5996 operating system and its tuning parameters.
5997
5998 It is important to understand that keep-alive packets are neither emitted nor
5999 received at the application level. It is only the network stacks which sees
6000 them. For this reason, even if one side of the proxy already uses keep-alives
6001 to maintain its connection alive, those keep-alive packets will not be
6002 forwarded to the other side of the proxy.
6003
6004 Please note that this has nothing to do with HTTP keep-alive.
6005
6006 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
6007 client side of a connection, which should help when session expirations are
6008 noticed between HAProxy and a client.
6009
6010 If this option has been enabled in a "defaults" section, it can be disabled
6011 in a specific instance by prepending the "no" keyword before it.
6012
6013 See also : "option srvtcpka", "option tcpka"
6014
6015
Willy Tarreau0ba27502007-12-24 16:55:16 +01006016option contstats
6017 Enable continuous traffic statistics updates
6018 May be used in sections : defaults | frontend | listen | backend
6019 yes | yes | yes | no
6020 Arguments : none
6021
6022 By default, counters used for statistics calculation are incremented
6023 only when a session finishes. It works quite well when serving small
6024 objects, but with big ones (for example large images or archives) or
6025 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01006026 a hedgehog. With this option enabled counters get incremented frequently
6027 along the session, typically every 5 seconds, which is often enough to
6028 produce clean graphs. Recounting touches a hotpath directly so it is not
6029 not enabled by default, as it can cause a lot of wakeups for very large
6030 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01006031
6032
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006033option dontlog-normal
6034no option dontlog-normal
6035 Enable or disable logging of normal, successful connections
6036 May be used in sections : defaults | frontend | listen | backend
6037 yes | yes | yes | no
6038 Arguments : none
6039
6040 There are large sites dealing with several thousand connections per second
6041 and for which logging is a major pain. Some of them are even forced to turn
6042 logs off and cannot debug production issues. Setting this option ensures that
6043 normal connections, those which experience no error, no timeout, no retry nor
6044 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
6045 mode, the response status code is checked and return codes 5xx will still be
6046 logged.
6047
6048 It is strongly discouraged to use this option as most of the time, the key to
6049 complex issues is in the normal logs which will not be logged here. If you
6050 need to separate logs, see the "log-separate-errors" option instead.
6051
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006052 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006053 logging.
6054
6055
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006056option dontlognull
6057no option dontlognull
6058 Enable or disable logging of null connections
6059 May be used in sections : defaults | frontend | listen | backend
6060 yes | yes | yes | no
6061 Arguments : none
6062
6063 In certain environments, there are components which will regularly connect to
6064 various systems to ensure that they are still alive. It can be the case from
6065 another load balancer as well as from monitoring systems. By default, even a
6066 simple port probe or scan will produce a log. If those connections pollute
6067 the logs too much, it is possible to enable option "dontlognull" to indicate
6068 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02006069 which typically corresponds to those probes. Note that errors will still be
6070 returned to the client and accounted for in the stats. If this is not what is
6071 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006072
6073 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01006074 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006075 would not be logged.
6076
6077 If this option has been enabled in a "defaults" section, it can be disabled
6078 in a specific instance by prepending the "no" keyword before it.
6079
Willy Tarreau0f228a02015-05-01 15:37:53 +02006080 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
6081 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006082
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006083
Willy Tarreau87cf5142011-08-19 22:57:24 +02006084option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006085 Enable insertion of the X-Forwarded-For header to requests sent to servers
6086 May be used in sections : defaults | frontend | listen | backend
6087 yes | yes | yes | yes
6088 Arguments :
6089 <network> is an optional argument used to disable this option for sources
6090 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02006091 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01006092 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006093
6094 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
6095 their client address. This is sometimes annoying when the client's IP address
6096 is expected in server logs. To solve this problem, the well-known HTTP header
6097 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
6098 This header contains a value representing the client's IP address. Since this
6099 header is always appended at the end of the existing header list, the server
6100 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02006101 the server's manual to find how to enable use of this standard header. Note
6102 that only the last occurrence of the header must be used, since it is really
6103 possible that the client has already brought one.
6104
Willy Tarreaud72758d2010-01-12 10:42:19 +01006105 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02006106 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01006107 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01006108 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006109 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02006110 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01006111
6112 Sometimes, a same HAProxy instance may be shared between a direct client
6113 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6114 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6115 header for a known source address or network by adding the "except" keyword
6116 followed by the network address. In this case, any source IP matching the
6117 network will not cause an addition of this header. Most common uses are with
6118 private networks or 127.0.0.1.
6119
Willy Tarreau87cf5142011-08-19 22:57:24 +02006120 Alternatively, the keyword "if-none" states that the header will only be
6121 added if it is not present. This should only be used in perfectly trusted
6122 environment, as this might cause a security issue if headers reaching haproxy
6123 are under the control of the end-user.
6124
Willy Tarreauc27debf2008-01-06 08:57:02 +01006125 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02006126 least one of them uses it, the header will be added. Note that the backend's
6127 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02006128 both are defined. In the case of the "if-none" argument, if at least one of
6129 the frontend or the backend does not specify it, it wants the addition to be
6130 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006131
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006132 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01006133 # Public HTTP address also used by stunnel on the same machine
6134 frontend www
6135 mode http
6136 option forwardfor except 127.0.0.1 # stunnel already adds the header
6137
Ross Westaf72a1d2008-08-03 10:51:45 +02006138 # Those servers want the IP Address in X-Client
6139 backend www
6140 mode http
6141 option forwardfor header X-Client
6142
Willy Tarreau87cf5142011-08-19 22:57:24 +02006143 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006144 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01006145
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006146
Christopher Faulet98fbe952019-07-22 16:18:24 +02006147option h1-case-adjust-bogus-client
6148no option h1-case-adjust-bogus-client
6149 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
6150 May be used in sections : defaults | frontend | listen | backend
6151 yes | yes | yes | no
6152 Arguments : none
6153
6154 There is no standard case for header names because, as stated in RFC7230,
6155 they are case-insensitive. So applications must handle them in a case-
6156 insensitive manner. But some bogus applications violate the standards and
6157 erroneously rely on the cases most commonly used by browsers. This problem
6158 becomes critical with HTTP/2 because all header names must be exchanged in
6159 lower case, and HAProxy follows the same convention. All header names are
6160 sent in lower case to clients and servers, regardless of the HTTP version.
6161
6162 When HAProxy receives an HTTP/1 response, its header names are converted to
6163 lower case and manipulated and sent this way to the clients. If a client is
6164 known to violate the HTTP standards and to fail to process a response coming
6165 from HAProxy, it is possible to transform the lower case header names to a
6166 different format when the response is formatted and sent to the client, by
6167 enabling this option and specifying the list of headers to be reformatted
6168 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6169 must only be a temporary workaround for the time it takes the client to be
6170 fixed, because clients which require such workarounds might be vulnerable to
6171 content smuggling attacks and must absolutely be fixed.
6172
6173 Please note that this option will not affect standards-compliant clients.
6174
6175 If this option has been enabled in a "defaults" section, it can be disabled
6176 in a specific instance by prepending the "no" keyword before it.
6177
6178 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
6179 "h1-case-adjust-file".
6180
6181
6182option h1-case-adjust-bogus-server
6183no option h1-case-adjust-bogus-server
6184 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
6185 May be used in sections : defaults | frontend | listen | backend
6186 yes | no | yes | yes
6187 Arguments : none
6188
6189 There is no standard case for header names because, as stated in RFC7230,
6190 they are case-insensitive. So applications must handle them in a case-
6191 insensitive manner. But some bogus applications violate the standards and
6192 erroneously rely on the cases most commonly used by browsers. This problem
6193 becomes critical with HTTP/2 because all header names must be exchanged in
6194 lower case, and HAProxy follows the same convention. All header names are
6195 sent in lower case to clients and servers, regardless of the HTTP version.
6196
6197 When HAProxy receives an HTTP/1 request, its header names are converted to
6198 lower case and manipulated and sent this way to the servers. If a server is
6199 known to violate the HTTP standards and to fail to process a request coming
6200 from HAProxy, it is possible to transform the lower case header names to a
6201 different format when the request is formatted and sent to the server, by
6202 enabling this option and specifying the list of headers to be reformatted
6203 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6204 must only be a temporary workaround for the time it takes the server to be
6205 fixed, because servers which require such workarounds might be vulnerable to
6206 content smuggling attacks and must absolutely be fixed.
6207
6208 Please note that this option will not affect standards-compliant servers.
6209
6210 If this option has been enabled in a "defaults" section, it can be disabled
6211 in a specific instance by prepending the "no" keyword before it.
6212
6213 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
6214 "h1-case-adjust-file".
6215
6216
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006217option http-buffer-request
6218no option http-buffer-request
6219 Enable or disable waiting for whole HTTP request body before proceeding
6220 May be used in sections : defaults | frontend | listen | backend
6221 yes | yes | yes | yes
6222 Arguments : none
6223
6224 It is sometimes desirable to wait for the body of an HTTP request before
6225 taking a decision. This is what is being done by "balance url_param" for
6226 example. The first use case is to buffer requests from slow clients before
6227 connecting to the server. Another use case consists in taking the routing
6228 decision based on the request body's contents. This option placed in a
6229 frontend or backend forces the HTTP processing to wait until either the whole
6230 body is received, or the request buffer is full, or the first chunk is
6231 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01006232 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006233 the frontend and the backend, so this should definitely not be used by
6234 default.
6235
Baptiste Assmanneccdf432015-10-28 13:49:01 +01006236 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006237
6238
Willy Tarreau0f228a02015-05-01 15:37:53 +02006239option http-ignore-probes
6240no option http-ignore-probes
6241 Enable or disable logging of null connections and request timeouts
6242 May be used in sections : defaults | frontend | listen | backend
6243 yes | yes | yes | no
6244 Arguments : none
6245
6246 Recently some browsers started to implement a "pre-connect" feature
6247 consisting in speculatively connecting to some recently visited web sites
6248 just in case the user would like to visit them. This results in many
6249 connections being established to web sites, which end up in 408 Request
6250 Timeout if the timeout strikes first, or 400 Bad Request when the browser
6251 decides to close them first. These ones pollute the log and feed the error
6252 counters. There was already "option dontlognull" but it's insufficient in
6253 this case. Instead, this option does the following things :
6254 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01006255 was received over a connection before it was closed;
6256 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02006257 - prevent any error counter from being incremented
6258
6259 That way the empty connection is silently ignored. Note that it is better
6260 not to use this unless it is clear that it is needed, because it will hide
6261 real problems. The most common reason for not receiving a request and seeing
6262 a 408 is due to an MTU inconsistency between the client and an intermediary
6263 element such as a VPN, which blocks too large packets. These issues are
6264 generally seen with POST requests as well as GET with large cookies. The logs
6265 are often the only way to detect them.
6266
6267 If this option has been enabled in a "defaults" section, it can be disabled
6268 in a specific instance by prepending the "no" keyword before it.
6269
6270 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
6271
6272
Willy Tarreau16bfb022010-01-16 19:48:41 +01006273option http-keep-alive
6274no option http-keep-alive
6275 Enable or disable HTTP keep-alive from client to server
6276 May be used in sections : defaults | frontend | listen | backend
6277 yes | yes | yes | yes
6278 Arguments : none
6279
Willy Tarreau70dffda2014-01-30 03:07:23 +01006280 By default HAProxy operates in keep-alive mode with regards to persistent
6281 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02006282 leaves the connection idle on both sides between the end of a response and
6283 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006284 as "option http-server-close" or "option httpclose". This option allows to
6285 set back the keep-alive mode, which can be useful when another mode was used
6286 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01006287
6288 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
6289 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01006290 network) and the fastest session reuse on the server side at the expense
6291 of maintaining idle connections to the servers. In general, it is possible
6292 with this option to achieve approximately twice the request rate that the
6293 "http-server-close" option achieves on small objects. There are mainly two
6294 situations where this option may be useful :
6295
6296 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01006297 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01006298
6299 - when the cost of establishing the connection to the server is significant
6300 compared to the cost of retrieving the associated object from the server.
6301
6302 This last case can happen when the server is a fast static server of cache.
6303 In this case, the server will need to be properly tuned to support high enough
6304 connection counts because connections will last until the client sends another
6305 request.
6306
6307 If the client request has to go to another backend or another server due to
6308 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006309 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6310 available to try optimize server selection so that if the server currently
6311 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006312
Willy Tarreau16bfb022010-01-16 19:48:41 +01006313 At the moment, logs will not indicate whether requests came from the same
6314 session or not. The accept date reported in the logs corresponds to the end
6315 of the previous request, and the request time corresponds to the time spent
6316 waiting for a new request. The keep-alive request time is still bound to the
6317 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6318 not set.
6319
Christopher Faulet159e6672019-07-16 15:09:52 +02006320 This option disables and replaces any previous "option httpclose" or "option
6321 http-server-close". When backend and frontend options differ, all of these 4
6322 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006323
Christopher Faulet315b39c2018-09-21 16:26:19 +02006324 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006325 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01006326 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006327
6328
Willy Tarreau96e31212011-05-30 18:10:30 +02006329option http-no-delay
6330no option http-no-delay
6331 Instruct the system to favor low interactive delays over performance in HTTP
6332 May be used in sections : defaults | frontend | listen | backend
6333 yes | yes | yes | yes
6334 Arguments : none
6335
6336 In HTTP, each payload is unidirectional and has no notion of interactivity.
6337 Any agent is expected to queue data somewhat for a reasonably low delay.
6338 There are some very rare server-to-server applications that abuse the HTTP
6339 protocol and expect the payload phase to be highly interactive, with many
6340 interleaved data chunks in both directions within a single request. This is
6341 absolutely not supported by the HTTP specification and will not work across
6342 most proxies or servers. When such applications attempt to do this through
6343 haproxy, it works but they will experience high delays due to the network
6344 optimizations which favor performance by instructing the system to wait for
6345 enough data to be available in order to only send full packets. Typical
6346 delays are around 200 ms per round trip. Note that this only happens with
6347 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6348 affected.
6349
6350 When "option http-no-delay" is present in either the frontend or the backend
6351 used by a connection, all such optimizations will be disabled in order to
6352 make the exchanges as fast as possible. Of course this offers no guarantee on
6353 the functionality, as it may break at any other place. But if it works via
6354 HAProxy, it will work as fast as possible. This option should never be used
6355 by default, and should never be used at all unless such a buggy application
6356 is discovered. The impact of using this option is an increase of bandwidth
6357 usage and CPU usage, which may significantly lower performance in high
6358 latency environments.
6359
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006360 See also : "option http-buffer-request"
6361
Willy Tarreau96e31212011-05-30 18:10:30 +02006362
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006363option http-pretend-keepalive
6364no option http-pretend-keepalive
6365 Define whether haproxy will announce keepalive to the server or not
6366 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006367 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006368 Arguments : none
6369
Christopher Faulet315b39c2018-09-21 16:26:19 +02006370 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006371 adds a "Connection: close" header to the request forwarded to the server.
6372 Unfortunately, when some servers see this header, they automatically refrain
6373 from using the chunked encoding for responses of unknown length, while this
6374 is totally unrelated. The immediate effect is that this prevents haproxy from
6375 maintaining the client connection alive. A second effect is that a client or
6376 a cache could receive an incomplete response without being aware of it, and
6377 consider the response complete.
6378
6379 By setting "option http-pretend-keepalive", haproxy will make the server
6380 believe it will keep the connection alive. The server will then not fall back
6381 to the abnormal undesired above. When haproxy gets the whole response, it
6382 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006383 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006384 connection is correctly closed on the server side.
6385
6386 It is recommended not to enable this option by default, because most servers
6387 will more efficiently close the connection themselves after the last packet,
6388 and release its buffers slightly earlier. Also, the added packet on the
6389 network could slightly reduce the overall peak performance. However it is
6390 worth noting that when this option is enabled, haproxy will have slightly
6391 less work to do. So if haproxy is the bottleneck on the whole architecture,
6392 enabling this option might save a few CPU cycles.
6393
Christopher Faulet98db9762018-09-21 10:25:19 +02006394 This option may be set in backend and listen sections. Using it in a frontend
6395 section will be ignored and a warning will be reported during startup. It is
6396 a backend related option, so there is no real reason to set it on a
6397 frontend. This option may be combined with "option httpclose", which will
6398 cause keepalive to be announced to the server and close to be announced to
6399 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006400
6401 If this option has been enabled in a "defaults" section, it can be disabled
6402 in a specific instance by prepending the "no" keyword before it.
6403
Christopher Faulet315b39c2018-09-21 16:26:19 +02006404 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006405 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006406
Willy Tarreauc27debf2008-01-06 08:57:02 +01006407
Willy Tarreaub608feb2010-01-02 22:47:18 +01006408option http-server-close
6409no option http-server-close
6410 Enable or disable HTTP connection closing on the server side
6411 May be used in sections : defaults | frontend | listen | backend
6412 yes | yes | yes | yes
6413 Arguments : none
6414
Willy Tarreau70dffda2014-01-30 03:07:23 +01006415 By default HAProxy operates in keep-alive mode with regards to persistent
6416 connections: for each connection it processes each request and response, and
6417 leaves the connection idle on both sides between the end of a response and
6418 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006419 as "option http-server-close" or "option httpclose". Setting "option
6420 http-server-close" enables HTTP connection-close mode on the server side
6421 while keeping the ability to support HTTP keep-alive and pipelining on the
6422 client side. This provides the lowest latency on the client side (slow
6423 network) and the fastest session reuse on the server side to save server
6424 resources, similarly to "option httpclose". It also permits non-keepalive
6425 capable servers to be served in keep-alive mode to the clients if they
6426 conform to the requirements of RFC7230. Please note that some servers do not
6427 always conform to those requirements when they see "Connection: close" in the
6428 request. The effect will be that keep-alive will never be used. A workaround
6429 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006430
6431 At the moment, logs will not indicate whether requests came from the same
6432 session or not. The accept date reported in the logs corresponds to the end
6433 of the previous request, and the request time corresponds to the time spent
6434 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006435 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6436 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006437
6438 This option may be set both in a frontend and in a backend. It is enabled if
6439 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02006440 It disables and replaces any previous "option httpclose" or "option
6441 http-keep-alive". Please check section 4 ("Proxies") to see how this option
6442 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006443
6444 If this option has been enabled in a "defaults" section, it can be disabled
6445 in a specific instance by prepending the "no" keyword before it.
6446
Christopher Faulet315b39c2018-09-21 16:26:19 +02006447 See also : "option httpclose", "option http-pretend-keepalive",
6448 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006449
Willy Tarreau88d349d2010-01-25 12:15:43 +01006450option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006451no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006452 Make use of non-standard Proxy-Connection header instead of Connection
6453 May be used in sections : defaults | frontend | listen | backend
6454 yes | yes | yes | no
6455 Arguments : none
6456
Lukas Tribus23953682017-04-28 13:24:30 +00006457 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006458 Connection header to indicate their wish of persistent or non-persistent
6459 connections, both browsers and proxies ignore this header for proxied
6460 connections and make use of the undocumented, non-standard Proxy-Connection
6461 header instead. The issue begins when trying to put a load balancer between
6462 browsers and such proxies, because there will be a difference between what
6463 haproxy understands and what the client and the proxy agree on.
6464
6465 By setting this option in a frontend, haproxy can automatically switch to use
6466 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006467 defined here as one where the URI begins with neither a '/' nor a '*'. This
6468 is incompatible with the HTTP tunnel mode. Note that this option can only be
6469 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006470
Willy Tarreau844a7e72010-01-31 21:46:18 +01006471 Also, when this option is set, a request which requires authentication will
6472 automatically switch to use proxy authentication headers if it is itself a
6473 proxied request. That makes it possible to check or enforce authentication in
6474 front of an existing proxy.
6475
Willy Tarreau88d349d2010-01-25 12:15:43 +01006476 This option should normally never be used, except in front of a proxy.
6477
Christopher Faulet315b39c2018-09-21 16:26:19 +02006478 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006479
Willy Tarreaud63335a2010-02-26 12:56:52 +01006480option httpchk
6481option httpchk <uri>
6482option httpchk <method> <uri>
6483option httpchk <method> <uri> <version>
6484 Enable HTTP protocol to check on the servers health
6485 May be used in sections : defaults | frontend | listen | backend
6486 yes | no | yes | yes
6487 Arguments :
6488 <method> is the optional HTTP method used with the requests. When not set,
6489 the "OPTIONS" method is used, as it generally requires low server
6490 processing and is easy to filter out from the logs. Any method
6491 may be used, though it is not recommended to invent non-standard
6492 ones.
6493
6494 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6495 which is accessible by default on almost any server, but may be
6496 changed to any other URI. Query strings are permitted.
6497
6498 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6499 but some servers might behave incorrectly in HTTP 1.0, so turning
6500 it to HTTP/1.1 may sometimes help. Note that the Host field is
6501 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6502 after "\r\n" following the version string.
6503
6504 By default, server health checks only consist in trying to establish a TCP
6505 connection. When "option httpchk" is specified, a complete HTTP request is
6506 sent once the TCP connection is established, and responses 2xx and 3xx are
6507 considered valid, while all other ones indicate a server failure, including
6508 the lack of any response.
6509
6510 The port and interval are specified in the server configuration.
6511
6512 This option does not necessarily require an HTTP backend, it also works with
6513 plain TCP backends. This is particularly useful to check simple scripts bound
6514 to some dedicated ports using the inetd daemon.
6515
6516 Examples :
6517 # Relay HTTPS traffic to Apache instance and check service availability
6518 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6519 backend https_relay
6520 mode tcp
6521 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6522 server apache1 192.168.1.1:443 check port 80
6523
Simon Hormanafc47ee2013-11-25 10:46:35 +09006524 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6525 "option pgsql-check", "http-check" and the "check", "port" and
6526 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006527
6528
Willy Tarreauc27debf2008-01-06 08:57:02 +01006529option httpclose
6530no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006531 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006532 May be used in sections : defaults | frontend | listen | backend
6533 yes | yes | yes | yes
6534 Arguments : none
6535
Willy Tarreau70dffda2014-01-30 03:07:23 +01006536 By default HAProxy operates in keep-alive mode with regards to persistent
6537 connections: for each connection it processes each request and response, and
6538 leaves the connection idle on both sides between the end of a response and
6539 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006540 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006541
Christopher Faulet315b39c2018-09-21 16:26:19 +02006542 If "option httpclose" is set, HAProxy will close connections with the server
6543 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05006544 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02006545 and will add one if missing. Any "Connection" header different from "close"
6546 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006547
Christopher Faulet315b39c2018-09-21 16:26:19 +02006548 This option may also be combined with "option http-pretend-keepalive", which
6549 will disable sending of the "Connection: close" header, but will still cause
6550 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006551
6552 This option may be set both in a frontend and in a backend. It is enabled if
6553 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02006554 It disables and replaces any previous "option http-server-close" or "option
6555 http-keep-alive". Please check section 4 ("Proxies") to see how this option
6556 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006557
6558 If this option has been enabled in a "defaults" section, it can be disabled
6559 in a specific instance by prepending the "no" keyword before it.
6560
Christopher Faulet315b39c2018-09-21 16:26:19 +02006561 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006562
6563
Emeric Brun3a058f32009-06-30 18:26:00 +02006564option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006565 Enable logging of HTTP request, session state and timers
6566 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006567 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006568 Arguments :
6569 clf if the "clf" argument is added, then the output format will be
6570 the CLF format instead of HAProxy's default HTTP format. You can
6571 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006572 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006573 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006574
6575 By default, the log output format is very poor, as it only contains the
6576 source and destination addresses, and the instance name. By specifying
6577 "option httplog", each log line turns into a much richer format including,
6578 but not limited to, the HTTP request, the connection timers, the session
6579 status, the connections numbers, the captured headers and cookies, the
6580 frontend, backend and server name, and of course the source address and
6581 ports.
6582
PiBa-NLbd556bf2014-12-11 21:31:54 +01006583 Specifying only "option httplog" will automatically clear the 'clf' mode
6584 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006585
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006586 "option httplog" overrides any previous "log-format" directive.
6587
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006588 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006589
Willy Tarreau55165fe2009-05-10 12:02:55 +02006590
6591option http_proxy
6592no option http_proxy
6593 Enable or disable plain HTTP proxy mode
6594 May be used in sections : defaults | frontend | listen | backend
6595 yes | yes | yes | yes
6596 Arguments : none
6597
6598 It sometimes happens that people need a pure HTTP proxy which understands
6599 basic proxy requests without caching nor any fancy feature. In this case,
6600 it may be worth setting up an HAProxy instance with the "option http_proxy"
6601 set. In this mode, no server is declared, and the connection is forwarded to
6602 the IP address and port found in the URL after the "http://" scheme.
6603
6604 No host address resolution is performed, so this only works when pure IP
6605 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006606 it will probably be used only by experts who know they need exactly it. This
6607 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006608
6609 If this option has been enabled in a "defaults" section, it can be disabled
6610 in a specific instance by prepending the "no" keyword before it.
6611
6612 Example :
6613 # this backend understands HTTP proxy requests and forwards them directly.
6614 backend direct_forward
6615 option httpclose
6616 option http_proxy
6617
6618 See also : "option httpclose"
6619
Willy Tarreau211ad242009-10-03 21:45:07 +02006620
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006621option independent-streams
6622no option independent-streams
6623 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006624 May be used in sections : defaults | frontend | listen | backend
6625 yes | yes | yes | yes
6626 Arguments : none
6627
6628 By default, when data is sent over a socket, both the write timeout and the
6629 read timeout for that socket are refreshed, because we consider that there is
6630 activity on that socket, and we have no other means of guessing if we should
6631 receive data or not.
6632
Davor Ocelice9ed2812017-12-25 17:49:28 +01006633 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006634 exists a situation where it is desirable to disable it, and only refresh the
6635 read timeout if there are incoming data. This happens on sessions with large
6636 timeouts and low amounts of exchanged data such as telnet session. If the
6637 server suddenly disappears, the output data accumulates in the system's
6638 socket buffers, both timeouts are correctly refreshed, and there is no way
6639 to know the server does not receive them, so we don't timeout. However, when
6640 the underlying protocol always echoes sent data, it would be enough by itself
6641 to detect the issue using the read timeout. Note that this problem does not
6642 happen with more verbose protocols because data won't accumulate long in the
6643 socket buffers.
6644
6645 When this option is set on the frontend, it will disable read timeout updates
6646 on data sent to the client. There probably is little use of this case. When
6647 the option is set on the backend, it will disable read timeout updates on
6648 data sent to the server. Doing so will typically break large HTTP posts from
6649 slow lines, so use it with caution.
6650
Willy Tarreauce887fd2012-05-12 12:50:00 +02006651 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006652
6653
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006654option ldap-check
6655 Use LDAPv3 health checks for server testing
6656 May be used in sections : defaults | frontend | listen | backend
6657 yes | no | yes | yes
6658 Arguments : none
6659
6660 It is possible to test that the server correctly talks LDAPv3 instead of just
6661 testing that it accepts the TCP connection. When this option is set, an
6662 LDAPv3 anonymous simple bind message is sent to the server, and the response
6663 is analyzed to find an LDAPv3 bind response message.
6664
6665 The server is considered valid only when the LDAP response contains success
6666 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6667
6668 Logging of bind requests is server dependent see your documentation how to
6669 configure it.
6670
6671 Example :
6672 option ldap-check
6673
6674 See also : "option httpchk"
6675
6676
Simon Horman98637e52014-06-20 12:30:16 +09006677option external-check
6678 Use external processes for server health checks
6679 May be used in sections : defaults | frontend | listen | backend
6680 yes | no | yes | yes
6681
6682 It is possible to test the health of a server using an external command.
6683 This is achieved by running the executable set using "external-check
6684 command".
6685
6686 Requires the "external-check" global to be set.
6687
6688 See also : "external-check", "external-check command", "external-check path"
6689
6690
Willy Tarreau211ad242009-10-03 21:45:07 +02006691option log-health-checks
6692no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006693 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006694 May be used in sections : defaults | frontend | listen | backend
6695 yes | no | yes | yes
6696 Arguments : none
6697
Willy Tarreaubef1b322014-05-13 21:01:39 +02006698 By default, failed health check are logged if server is UP and successful
6699 health checks are logged if server is DOWN, so the amount of additional
6700 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006701
Willy Tarreaubef1b322014-05-13 21:01:39 +02006702 When this option is enabled, any change of the health check status or to
6703 the server's health will be logged, so that it becomes possible to know
6704 that a server was failing occasional checks before crashing, or exactly when
6705 it failed to respond a valid HTTP status, then when the port started to
6706 reject connections, then when the server stopped responding at all.
6707
Davor Ocelice9ed2812017-12-25 17:49:28 +01006708 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006709 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006710
Willy Tarreaubef1b322014-05-13 21:01:39 +02006711 See also: "option httpchk", "option ldap-check", "option mysql-check",
6712 "option pgsql-check", "option redis-check", "option smtpchk",
6713 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006714
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006715
6716option log-separate-errors
6717no option log-separate-errors
6718 Change log level for non-completely successful connections
6719 May be used in sections : defaults | frontend | listen | backend
6720 yes | yes | yes | no
6721 Arguments : none
6722
6723 Sometimes looking for errors in logs is not easy. This option makes haproxy
6724 raise the level of logs containing potentially interesting information such
6725 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6726 level changes from "info" to "err". This makes it possible to log them
6727 separately to a different file with most syslog daemons. Be careful not to
6728 remove them from the original file, otherwise you would lose ordering which
6729 provides very important information.
6730
6731 Using this option, large sites dealing with several thousand connections per
6732 second may log normal traffic to a rotating buffer and only archive smaller
6733 error logs.
6734
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006735 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006736 logging.
6737
Willy Tarreauc27debf2008-01-06 08:57:02 +01006738
6739option logasap
6740no option logasap
6741 Enable or disable early logging of HTTP requests
6742 May be used in sections : defaults | frontend | listen | backend
6743 yes | yes | yes | no
6744 Arguments : none
6745
6746 By default, HTTP requests are logged upon termination so that the total
6747 transfer time and the number of bytes appear in the logs. When large objects
6748 are being transferred, it may take a while before the request appears in the
6749 logs. Using "option logasap", the request gets logged as soon as the server
6750 sends the complete headers. The only missing information in the logs will be
6751 the total number of bytes which will indicate everything except the amount
6752 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006753 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006754 "Content-Length" response header so that the logs at least indicate how many
6755 bytes are expected to be transferred.
6756
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006757 Examples :
6758 listen http_proxy 0.0.0.0:80
6759 mode http
6760 option httplog
6761 option logasap
6762 log 192.168.2.200 local3
6763
6764 >>> Feb 6 12:14:14 localhost \
6765 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6766 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6767 "GET /image.iso HTTP/1.0"
6768
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006769 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006770 logging.
6771
6772
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006773option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006774 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006775 May be used in sections : defaults | frontend | listen | backend
6776 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006777 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006778 <username> This is the username which will be used when connecting to MySQL
6779 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006780 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006781
6782 If you specify a username, the check consists of sending two MySQL packet,
6783 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006784 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006785 Error packet. It is a basic but useful test which does not produce error nor
6786 aborted connect on the server. However, it requires adding an authorization
6787 in the MySQL table, like this :
6788
6789 USE mysql;
6790 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6791 FLUSH PRIVILEGES;
6792
6793 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006794 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006795 Error packet, we don't send anything in this mode. It was reported that it
6796 can generate lockout if check is too frequent and/or if there is not enough
6797 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6798 value as if a connection is established successfully within fewer than MySQL
6799 "max_connect_errors" attempts after a previous connection was interrupted,
6800 the error count for the host is cleared to zero. If HAProxy's server get
6801 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6802
6803 Remember that this does not check database presence nor database consistency.
6804 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006805
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006806 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006807
6808 Most often, an incoming MySQL server needs to see the client's IP address for
6809 various purposes, including IP privilege matching and connection logging.
6810 When possible, it is often wise to masquerade the client's IP address when
6811 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006812 which requires the transparent proxy feature to be compiled in, and the MySQL
6813 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006814
6815 See also: "option httpchk"
6816
6817
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006818option nolinger
6819no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006820 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006821 May be used in sections: defaults | frontend | listen | backend
6822 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006823 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006824
Davor Ocelice9ed2812017-12-25 17:49:28 +01006825 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006826 physically disconnected), the session timeouts triggers and the session is
6827 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6828 using some resources and possibly limiting the ability to establish newer
6829 connections.
6830
6831 When this happens, it is possible to activate "option nolinger" which forces
6832 the system to immediately remove any socket's pending data on close. Thus,
6833 the session is instantly purged from the system's tables. This usually has
6834 side effects such as increased number of TCP resets due to old retransmits
6835 getting immediately rejected. Some firewalls may sometimes complain about
6836 this too.
6837
6838 For this reason, it is not recommended to use this option when not absolutely
6839 needed. You know that you need it when you have thousands of FIN_WAIT1
6840 sessions on your system (TIME_WAIT ones do not count).
6841
6842 This option may be used both on frontends and backends, depending on the side
6843 where it is required. Use it on the frontend for clients, and on the backend
6844 for servers.
6845
6846 If this option has been enabled in a "defaults" section, it can be disabled
6847 in a specific instance by prepending the "no" keyword before it.
6848
6849
Willy Tarreau55165fe2009-05-10 12:02:55 +02006850option originalto [ except <network> ] [ header <name> ]
6851 Enable insertion of the X-Original-To header to requests sent to servers
6852 May be used in sections : defaults | frontend | listen | backend
6853 yes | yes | yes | yes
6854 Arguments :
6855 <network> is an optional argument used to disable this option for sources
6856 matching <network>
6857 <name> an optional argument to specify a different "X-Original-To"
6858 header name.
6859
6860 Since HAProxy can work in transparent mode, every request from a client can
6861 be redirected to the proxy and HAProxy itself can proxy every request to a
6862 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6863 be lost. This is annoying when you want access rules based on destination ip
6864 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6865 added by HAProxy to all requests sent to the server. This header contains a
6866 value representing the original destination IP address. Since this must be
6867 configured to always use the last occurrence of this header only. Note that
6868 only the last occurrence of the header must be used, since it is really
6869 possible that the client has already brought one.
6870
6871 The keyword "header" may be used to supply a different header name to replace
6872 the default "X-Original-To". This can be useful where you might already
6873 have a "X-Original-To" header from a different application, and you need
6874 preserve it. Also if your backend server doesn't use the "X-Original-To"
6875 header and requires different one.
6876
6877 Sometimes, a same HAProxy instance may be shared between a direct client
6878 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6879 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6880 header for a known source address or network by adding the "except" keyword
6881 followed by the network address. In this case, any source IP matching the
6882 network will not cause an addition of this header. Most common uses are with
6883 private networks or 127.0.0.1.
6884
6885 This option may be specified either in the frontend or in the backend. If at
6886 least one of them uses it, the header will be added. Note that the backend's
6887 setting of the header subargument takes precedence over the frontend's if
6888 both are defined.
6889
Willy Tarreau55165fe2009-05-10 12:02:55 +02006890 Examples :
6891 # Original Destination address
6892 frontend www
6893 mode http
6894 option originalto except 127.0.0.1
6895
6896 # Those servers want the IP Address in X-Client-Dst
6897 backend www
6898 mode http
6899 option originalto header X-Client-Dst
6900
Christopher Faulet315b39c2018-09-21 16:26:19 +02006901 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02006902
6903
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006904option persist
6905no option persist
6906 Enable or disable forced persistence on down servers
6907 May be used in sections: defaults | frontend | listen | backend
6908 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006909 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006910
6911 When an HTTP request reaches a backend with a cookie which references a dead
6912 server, by default it is redispatched to another server. It is possible to
6913 force the request to be sent to the dead server first using "option persist"
6914 if absolutely needed. A common use case is when servers are under extreme
6915 load and spend their time flapping. In this case, the users would still be
6916 directed to the server they opened the session on, in the hope they would be
6917 correctly served. It is recommended to use "option redispatch" in conjunction
6918 with this option so that in the event it would not be possible to connect to
6919 the server at all (server definitely dead), the client would finally be
6920 redirected to another valid server.
6921
6922 If this option has been enabled in a "defaults" section, it can be disabled
6923 in a specific instance by prepending the "no" keyword before it.
6924
Willy Tarreau4de91492010-01-22 19:10:05 +01006925 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006926
6927
Willy Tarreau0c122822013-12-15 18:49:01 +01006928option pgsql-check [ user <username> ]
6929 Use PostgreSQL health checks for server testing
6930 May be used in sections : defaults | frontend | listen | backend
6931 yes | no | yes | yes
6932 Arguments :
6933 <username> This is the username which will be used when connecting to
6934 PostgreSQL server.
6935
6936 The check sends a PostgreSQL StartupMessage and waits for either
6937 Authentication request or ErrorResponse message. It is a basic but useful
6938 test which does not produce error nor aborted connect on the server.
6939 This check is identical with the "mysql-check".
6940
6941 See also: "option httpchk"
6942
6943
Willy Tarreau9420b122013-12-15 18:58:25 +01006944option prefer-last-server
6945no option prefer-last-server
6946 Allow multiple load balanced requests to remain on the same server
6947 May be used in sections: defaults | frontend | listen | backend
6948 yes | no | yes | yes
6949 Arguments : none
6950
6951 When the load balancing algorithm in use is not deterministic, and a previous
6952 request was sent to a server to which haproxy still holds a connection, it is
6953 sometimes desirable that subsequent requests on a same session go to the same
6954 server as much as possible. Note that this is different from persistence, as
6955 we only indicate a preference which haproxy tries to apply without any form
6956 of warranty. The real use is for keep-alive connections sent to servers. When
6957 this option is used, haproxy will try to reuse the same connection that is
6958 attached to the server instead of rebalancing to another server, causing a
6959 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006960 not make much sense to use this in combination with hashing algorithms. Note,
6961 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02006962 to a proxy which sends a 407 (authentication required), when the load
6963 balancing algorithm is not deterministic. This is mandatory for use with the
6964 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01006965 troubleshooting some faulty applications. Option prefer-last-server might be
6966 desirable in these environments as well, to avoid redistributing the traffic
6967 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006968
6969 If this option has been enabled in a "defaults" section, it can be disabled
6970 in a specific instance by prepending the "no" keyword before it.
6971
6972 See also: "option http-keep-alive"
6973
6974
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006975option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006976option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006977no option redispatch
6978 Enable or disable session redistribution in case of connection failure
6979 May be used in sections: defaults | frontend | listen | backend
6980 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006981 Arguments :
6982 <interval> The optional integer value that controls how often redispatches
6983 occur when retrying connections. Positive value P indicates a
6984 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006985 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006986 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006987 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006988 positive value of 1 would indicate a redispatch on every retry,
6989 and a positive value of 3 would indicate a redispatch on every
6990 third retry. You can disable redispatches with a value of 0.
6991
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006992
6993 In HTTP mode, if a server designated by a cookie is down, clients may
6994 definitely stick to it because they cannot flush the cookie, so they will not
6995 be able to access the service anymore.
6996
Willy Tarreau59884a62019-01-02 14:48:31 +01006997 Specifying "option redispatch" will allow the proxy to break cookie or
6998 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006999
Joseph Lynch726ab712015-05-11 23:25:34 -07007000 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007001 connection failures. Of course, it requires having "retries" set to a nonzero
7002 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007003
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007004 If this option has been enabled in a "defaults" section, it can be disabled
7005 in a specific instance by prepending the "no" keyword before it.
7006
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007007 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007008
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007009
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007010option redis-check
7011 Use redis health checks for server testing
7012 May be used in sections : defaults | frontend | listen | backend
7013 yes | no | yes | yes
7014 Arguments : none
7015
7016 It is possible to test that the server correctly talks REDIS protocol instead
7017 of just testing that it accepts the TCP connection. When this option is set,
7018 a PING redis command is sent to the server, and the response is analyzed to
7019 find the "+PONG" response message.
7020
7021 Example :
7022 option redis-check
7023
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007024 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007025
7026
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007027option smtpchk
7028option smtpchk <hello> <domain>
7029 Use SMTP health checks for server testing
7030 May be used in sections : defaults | frontend | listen | backend
7031 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01007032 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007033 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02007034 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007035 values will be turned into the default command ("HELO").
7036
7037 <domain> is the domain name to present to the server. It may only be
7038 specified (and is mandatory) if the hello command has been
7039 specified. By default, "localhost" is used.
7040
7041 When "option smtpchk" is set, the health checks will consist in TCP
7042 connections followed by an SMTP command. By default, this command is
7043 "HELO localhost". The server's return code is analyzed and only return codes
7044 starting with a "2" will be considered as valid. All other responses,
7045 including a lack of response will constitute an error and will indicate a
7046 dead server.
7047
7048 This test is meant to be used with SMTP servers or relays. Depending on the
7049 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01007050 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007051 25 is often easier than adjusting the configuration.
7052
7053 Most often, an incoming SMTP server needs to see the client's IP address for
7054 various purposes, including spam filtering, anti-spoofing and logging. When
7055 possible, it is often wise to masquerade the client's IP address when
7056 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007057 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007058
7059 Example :
7060 option smtpchk HELO mydomain.org
7061
7062 See also : "option httpchk", "source"
7063
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007064
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02007065option socket-stats
7066no option socket-stats
7067
7068 Enable or disable collecting & providing separate statistics for each socket.
7069 May be used in sections : defaults | frontend | listen | backend
7070 yes | yes | yes | no
7071
7072 Arguments : none
7073
7074
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007075option splice-auto
7076no option splice-auto
7077 Enable or disable automatic kernel acceleration on sockets in both directions
7078 May be used in sections : defaults | frontend | listen | backend
7079 yes | yes | yes | yes
7080 Arguments : none
7081
7082 When this option is enabled either on a frontend or on a backend, haproxy
7083 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007084 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007085 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007086 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007087 are not much aggressive in order to limit excessive use of splicing. This
7088 option requires splicing to be enabled at compile time, and may be globally
7089 disabled with the global option "nosplice". Since splice uses pipes, using it
7090 requires that there are enough spare pipes.
7091
7092 Important note: kernel-based TCP splicing is a Linux-specific feature which
7093 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
7094 transfer data between sockets without copying these data to user-space, thus
7095 providing noticeable performance gains and CPU cycles savings. Since many
7096 early implementations are buggy, corrupt data and/or are inefficient, this
7097 feature is not enabled by default, and it should be used with extreme care.
7098 While it is not possible to detect the correctness of an implementation,
7099 2.6.29 is the first version offering a properly working implementation. In
7100 case of doubt, splicing may be globally disabled using the global "nosplice"
7101 keyword.
7102
7103 Example :
7104 option splice-auto
7105
7106 If this option has been enabled in a "defaults" section, it can be disabled
7107 in a specific instance by prepending the "no" keyword before it.
7108
7109 See also : "option splice-request", "option splice-response", and global
7110 options "nosplice" and "maxpipes"
7111
7112
7113option splice-request
7114no option splice-request
7115 Enable or disable automatic kernel acceleration on sockets for requests
7116 May be used in sections : defaults | frontend | listen | backend
7117 yes | yes | yes | yes
7118 Arguments : none
7119
7120 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007121 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007122 the client to the server. It might still use the recv/send scheme if there
7123 are no spare pipes left. This option requires splicing to be enabled at
7124 compile time, and may be globally disabled with the global option "nosplice".
7125 Since splice uses pipes, using it requires that there are enough spare pipes.
7126
7127 Important note: see "option splice-auto" for usage limitations.
7128
7129 Example :
7130 option splice-request
7131
7132 If this option has been enabled in a "defaults" section, it can be disabled
7133 in a specific instance by prepending the "no" keyword before it.
7134
7135 See also : "option splice-auto", "option splice-response", and global options
7136 "nosplice" and "maxpipes"
7137
7138
7139option splice-response
7140no option splice-response
7141 Enable or disable automatic kernel acceleration on sockets for responses
7142 May be used in sections : defaults | frontend | listen | backend
7143 yes | yes | yes | yes
7144 Arguments : none
7145
7146 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007147 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007148 the server to the client. It might still use the recv/send scheme if there
7149 are no spare pipes left. This option requires splicing to be enabled at
7150 compile time, and may be globally disabled with the global option "nosplice".
7151 Since splice uses pipes, using it requires that there are enough spare pipes.
7152
7153 Important note: see "option splice-auto" for usage limitations.
7154
7155 Example :
7156 option splice-response
7157
7158 If this option has been enabled in a "defaults" section, it can be disabled
7159 in a specific instance by prepending the "no" keyword before it.
7160
7161 See also : "option splice-auto", "option splice-request", and global options
7162 "nosplice" and "maxpipes"
7163
7164
Christopher Fauletba7bc162016-11-07 21:07:38 +01007165option spop-check
7166 Use SPOP health checks for server testing
7167 May be used in sections : defaults | frontend | listen | backend
7168 no | no | no | yes
7169 Arguments : none
7170
7171 It is possible to test that the server correctly talks SPOP protocol instead
7172 of just testing that it accepts the TCP connection. When this option is set,
7173 a HELLO handshake is performed between HAProxy and the server, and the
7174 response is analyzed to check no error is reported.
7175
7176 Example :
7177 option spop-check
7178
7179 See also : "option httpchk"
7180
7181
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007182option srvtcpka
7183no option srvtcpka
7184 Enable or disable the sending of TCP keepalive packets on the server side
7185 May be used in sections : defaults | frontend | listen | backend
7186 yes | no | yes | yes
7187 Arguments : none
7188
7189 When there is a firewall or any session-aware component between a client and
7190 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007191 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007192 components decides to expire a session which has remained idle for too long.
7193
7194 Enabling socket-level TCP keep-alives makes the system regularly send packets
7195 to the other end of the connection, leaving it active. The delay between
7196 keep-alive probes is controlled by the system only and depends both on the
7197 operating system and its tuning parameters.
7198
7199 It is important to understand that keep-alive packets are neither emitted nor
7200 received at the application level. It is only the network stacks which sees
7201 them. For this reason, even if one side of the proxy already uses keep-alives
7202 to maintain its connection alive, those keep-alive packets will not be
7203 forwarded to the other side of the proxy.
7204
7205 Please note that this has nothing to do with HTTP keep-alive.
7206
7207 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
7208 server side of a connection, which should help when session expirations are
7209 noticed between HAProxy and a server.
7210
7211 If this option has been enabled in a "defaults" section, it can be disabled
7212 in a specific instance by prepending the "no" keyword before it.
7213
7214 See also : "option clitcpka", "option tcpka"
7215
7216
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007217option ssl-hello-chk
7218 Use SSLv3 client hello health checks for server testing
7219 May be used in sections : defaults | frontend | listen | backend
7220 yes | no | yes | yes
7221 Arguments : none
7222
7223 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
7224 possible to test that the server correctly talks SSL instead of just testing
7225 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
7226 SSLv3 client hello messages are sent once the connection is established to
7227 the server, and the response is analyzed to find an SSL server hello message.
7228 The server is considered valid only when the response contains this server
7229 hello message.
7230
7231 All servers tested till there correctly reply to SSLv3 client hello messages,
7232 and most servers tested do not even log the requests containing only hello
7233 messages, which is appreciable.
7234
Willy Tarreau763a95b2012-10-04 23:15:39 +02007235 Note that this check works even when SSL support was not built into haproxy
7236 because it forges the SSL message. When SSL support is available, it is best
7237 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007238
Willy Tarreau763a95b2012-10-04 23:15:39 +02007239 See also: "option httpchk", "check-ssl"
7240
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007241
Willy Tarreaued179852013-12-16 01:07:00 +01007242option tcp-check
7243 Perform health checks using tcp-check send/expect sequences
7244 May be used in sections: defaults | frontend | listen | backend
7245 yes | no | yes | yes
7246
7247 This health check method is intended to be combined with "tcp-check" command
7248 lists in order to support send/expect types of health check sequences.
7249
7250 TCP checks currently support 4 modes of operations :
7251 - no "tcp-check" directive : the health check only consists in a connection
7252 attempt, which remains the default mode.
7253
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007254 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007255 used to send a string along with a connection opening. With some
7256 protocols, it helps sending a "QUIT" message for example that prevents
7257 the server from logging a connection error for each health check. The
7258 check result will still be based on the ability to open the connection
7259 only.
7260
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007261 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007262 The connection is opened and haproxy waits for the server to present some
7263 contents which must validate some rules. The check result will be based
7264 on the matching between the contents and the rules. This is suited for
7265 POP, IMAP, SMTP, FTP, SSH, TELNET.
7266
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007267 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007268 used to test a hello-type protocol. HAProxy sends a message, the server
7269 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007270 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007271 suited for protocols which require a binding or a request/response model.
7272 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7273 already all have their dedicated checks with a deeper understanding of
7274 the respective protocols.
7275 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007276 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007277
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007278 A fifth mode can be used to insert comments in different steps of the
7279 script.
7280
7281 For each tcp-check rule you create, you can add a "comment" directive,
7282 followed by a string. This string will be reported in the log and stderr
7283 in debug mode. It is useful to make user-friendly error reporting.
7284 The "comment" is of course optional.
7285
7286
Willy Tarreaued179852013-12-16 01:07:00 +01007287 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007288 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007289 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007290 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007291
Davor Ocelice9ed2812017-12-25 17:49:28 +01007292 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007293 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007294 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007295
7296 # look for the redis master server after ensuring it speaks well
7297 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007298 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007299 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007300 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007301 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007302 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007303 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007304 tcp-check send info\ replication\r\n
7305 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007306 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007307 tcp-check send QUIT\r\n
7308 tcp-check expect string +OK
7309
Davor Ocelice9ed2812017-12-25 17:49:28 +01007310 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007311 (send many headers before analyzing)
7312 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007313 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007314 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7315 tcp-check send Host:\ www.mydomain.com\r\n
7316 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7317 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007318 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007319
7320
7321 See also : "tcp-check expect", "tcp-check send"
7322
7323
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007324option tcp-smart-accept
7325no option tcp-smart-accept
7326 Enable or disable the saving of one ACK packet during the accept sequence
7327 May be used in sections : defaults | frontend | listen | backend
7328 yes | yes | yes | no
7329 Arguments : none
7330
7331 When an HTTP connection request comes in, the system acknowledges it on
7332 behalf of HAProxy, then the client immediately sends its request, and the
7333 system acknowledges it too while it is notifying HAProxy about the new
7334 connection. HAProxy then reads the request and responds. This means that we
7335 have one TCP ACK sent by the system for nothing, because the request could
7336 very well be acknowledged by HAProxy when it sends its response.
7337
7338 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7339 sending this useless ACK on platforms which support it (currently at least
7340 Linux). It must not cause any problem, because the system will send it anyway
7341 after 40 ms if the response takes more time than expected to come.
7342
7343 During complex network debugging sessions, it may be desirable to disable
7344 this optimization because delayed ACKs can make troubleshooting more complex
7345 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007346 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007347
7348 It is also possible to force it for non-HTTP proxies by simply specifying
7349 "option tcp-smart-accept". For instance, it can make sense with some services
7350 such as SMTP where the server speaks first.
7351
7352 It is recommended to avoid forcing this option in a defaults section. In case
7353 of doubt, consider setting it back to automatic values by prepending the
7354 "default" keyword before it, or disabling it using the "no" keyword.
7355
Willy Tarreaud88edf22009-06-14 15:48:17 +02007356 See also : "option tcp-smart-connect"
7357
7358
7359option tcp-smart-connect
7360no option tcp-smart-connect
7361 Enable or disable the saving of one ACK packet during the connect sequence
7362 May be used in sections : defaults | frontend | listen | backend
7363 yes | no | yes | yes
7364 Arguments : none
7365
7366 On certain systems (at least Linux), HAProxy can ask the kernel not to
7367 immediately send an empty ACK upon a connection request, but to directly
7368 send the buffer request instead. This saves one packet on the network and
7369 thus boosts performance. It can also be useful for some servers, because they
7370 immediately get the request along with the incoming connection.
7371
7372 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7373 It is not enabled by default because it makes network troubleshooting more
7374 complex.
7375
7376 It only makes sense to enable it with protocols where the client speaks first
7377 such as HTTP. In other situations, if there is no data to send in place of
7378 the ACK, a normal ACK is sent.
7379
7380 If this option has been enabled in a "defaults" section, it can be disabled
7381 in a specific instance by prepending the "no" keyword before it.
7382
7383 See also : "option tcp-smart-accept"
7384
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007385
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007386option tcpka
7387 Enable or disable the sending of TCP keepalive packets on both sides
7388 May be used in sections : defaults | frontend | listen | backend
7389 yes | yes | yes | yes
7390 Arguments : none
7391
7392 When there is a firewall or any session-aware component between a client and
7393 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007394 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007395 components decides to expire a session which has remained idle for too long.
7396
7397 Enabling socket-level TCP keep-alives makes the system regularly send packets
7398 to the other end of the connection, leaving it active. The delay between
7399 keep-alive probes is controlled by the system only and depends both on the
7400 operating system and its tuning parameters.
7401
7402 It is important to understand that keep-alive packets are neither emitted nor
7403 received at the application level. It is only the network stacks which sees
7404 them. For this reason, even if one side of the proxy already uses keep-alives
7405 to maintain its connection alive, those keep-alive packets will not be
7406 forwarded to the other side of the proxy.
7407
7408 Please note that this has nothing to do with HTTP keep-alive.
7409
7410 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7411 the client and server sides of a connection. Note that this is meaningful
7412 only in "defaults" or "listen" sections. If this option is used in a
7413 frontend, only the client side will get keep-alives, and if this option is
7414 used in a backend, only the server side will get keep-alives. For this
7415 reason, it is strongly recommended to explicitly use "option clitcpka" and
7416 "option srvtcpka" when the configuration is split between frontends and
7417 backends.
7418
7419 See also : "option clitcpka", "option srvtcpka"
7420
Willy Tarreau844e3c52008-01-11 16:28:18 +01007421
7422option tcplog
7423 Enable advanced logging of TCP connections with session state and timers
7424 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007425 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007426 Arguments : none
7427
7428 By default, the log output format is very poor, as it only contains the
7429 source and destination addresses, and the instance name. By specifying
7430 "option tcplog", each log line turns into a much richer format including, but
7431 not limited to, the connection timers, the session status, the connections
7432 numbers, the frontend, backend and server name, and of course the source
7433 address and ports. This option is useful for pure TCP proxies in order to
7434 find which of the client or server disconnects or times out. For normal HTTP
7435 proxies, it's better to use "option httplog" which is even more complete.
7436
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007437 "option tcplog" overrides any previous "log-format" directive.
7438
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007439 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007440
7441
Willy Tarreau844e3c52008-01-11 16:28:18 +01007442option transparent
7443no option transparent
7444 Enable client-side transparent proxying
7445 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007446 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007447 Arguments : none
7448
7449 This option was introduced in order to provide layer 7 persistence to layer 3
7450 load balancers. The idea is to use the OS's ability to redirect an incoming
7451 connection for a remote address to a local process (here HAProxy), and let
7452 this process know what address was initially requested. When this option is
7453 used, sessions without cookies will be forwarded to the original destination
7454 IP address of the incoming request (which should match that of another
7455 equipment), while requests with cookies will still be forwarded to the
7456 appropriate server.
7457
7458 Note that contrary to a common belief, this option does NOT make HAProxy
7459 present the client's IP to the server when establishing the connection.
7460
Willy Tarreaua1146052011-03-01 09:51:54 +01007461 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007462 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007463
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007464
Simon Horman98637e52014-06-20 12:30:16 +09007465external-check command <command>
7466 Executable to run when performing an external-check
7467 May be used in sections : defaults | frontend | listen | backend
7468 yes | no | yes | yes
7469
7470 Arguments :
7471 <command> is the external command to run
7472
Simon Horman98637e52014-06-20 12:30:16 +09007473 The arguments passed to the to the command are:
7474
Cyril Bonté777be862014-12-02 21:21:35 +01007475 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007476
Cyril Bonté777be862014-12-02 21:21:35 +01007477 The <proxy_address> and <proxy_port> are derived from the first listener
7478 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7479 listener the proxy_address will be the path of the socket and the
7480 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7481 possible to determine a listener, and both <proxy_address> and <proxy_port>
7482 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007483
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007484 Some values are also provided through environment variables.
7485
7486 Environment variables :
7487 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7488 applicable, for example in a "backend" section).
7489
7490 HAPROXY_PROXY_ID The backend id.
7491
7492 HAPROXY_PROXY_NAME The backend name.
7493
7494 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7495 applicable, for example in a "backend" section or
7496 for a UNIX socket).
7497
7498 HAPROXY_SERVER_ADDR The server address.
7499
7500 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7501
7502 HAPROXY_SERVER_ID The server id.
7503
7504 HAPROXY_SERVER_MAXCONN The server max connections.
7505
7506 HAPROXY_SERVER_NAME The server name.
7507
7508 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7509 socket).
7510
7511 PATH The PATH environment variable used when executing
7512 the command may be set using "external-check path".
7513
William Lallemand4d03e432019-06-14 15:35:37 +02007514 See also "2.3. Environment variables" for other variables.
7515
Simon Horman98637e52014-06-20 12:30:16 +09007516 If the command executed and exits with a zero status then the check is
7517 considered to have passed, otherwise the check is considered to have
7518 failed.
7519
7520 Example :
7521 external-check command /bin/true
7522
7523 See also : "external-check", "option external-check", "external-check path"
7524
7525
7526external-check path <path>
7527 The value of the PATH environment variable used when running an external-check
7528 May be used in sections : defaults | frontend | listen | backend
7529 yes | no | yes | yes
7530
7531 Arguments :
7532 <path> is the path used when executing external command to run
7533
7534 The default path is "".
7535
7536 Example :
7537 external-check path "/usr/bin:/bin"
7538
7539 See also : "external-check", "option external-check",
7540 "external-check command"
7541
7542
Emeric Brun647caf12009-06-30 17:57:00 +02007543persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007544persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007545 Enable RDP cookie-based persistence
7546 May be used in sections : defaults | frontend | listen | backend
7547 yes | no | yes | yes
7548 Arguments :
7549 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007550 default cookie name "msts" will be used. There currently is no
7551 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007552
7553 This statement enables persistence based on an RDP cookie. The RDP cookie
7554 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007555 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007556 and if an RDP cookie is found, it is decoded. If it matches a known server
7557 which is still UP (or if "option persist" is set), then the connection is
7558 forwarded to this server.
7559
7560 Note that this only makes sense in a TCP backend, but for this to work, the
7561 frontend must have waited long enough to ensure that an RDP cookie is present
7562 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007563 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007564 a single "listen" section.
7565
Willy Tarreau61e28f22010-05-16 22:31:05 +02007566 Also, it is important to understand that the terminal server will emit this
7567 RDP cookie only if it is configured for "token redirection mode", which means
7568 that the "IP address redirection" option is disabled.
7569
Emeric Brun647caf12009-06-30 17:57:00 +02007570 Example :
7571 listen tse-farm
7572 bind :3389
7573 # wait up to 5s for an RDP cookie in the request
7574 tcp-request inspect-delay 5s
7575 tcp-request content accept if RDP_COOKIE
7576 # apply RDP cookie persistence
7577 persist rdp-cookie
7578 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007579 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007580 balance rdp-cookie
7581 server srv1 1.1.1.1:3389
7582 server srv2 1.1.1.2:3389
7583
Simon Hormanab814e02011-06-24 14:50:20 +09007584 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7585 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007586
7587
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007588rate-limit sessions <rate>
7589 Set a limit on the number of new sessions accepted per second on a frontend
7590 May be used in sections : defaults | frontend | listen | backend
7591 yes | yes | yes | no
7592 Arguments :
7593 <rate> The <rate> parameter is an integer designating the maximum number
7594 of new sessions per second to accept on the frontend.
7595
7596 When the frontend reaches the specified number of new sessions per second, it
7597 stops accepting new connections until the rate drops below the limit again.
7598 During this time, the pending sessions will be kept in the socket's backlog
7599 (in system buffers) and haproxy will not even be aware that sessions are
7600 pending. When applying very low limit on a highly loaded service, it may make
7601 sense to increase the socket's backlog using the "backlog" keyword.
7602
7603 This feature is particularly efficient at blocking connection-based attacks
7604 or service abuse on fragile servers. Since the session rate is measured every
7605 millisecond, it is extremely accurate. Also, the limit applies immediately,
7606 no delay is needed at all to detect the threshold.
7607
7608 Example : limit the connection rate on SMTP to 10 per second max
7609 listen smtp
7610 mode tcp
7611 bind :25
7612 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007613 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007614
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007615 Note : when the maximum rate is reached, the frontend's status is not changed
7616 but its sockets appear as "WAITING" in the statistics if the
7617 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007618
7619 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7620
7621
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007622redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7623redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7624redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007625 Return an HTTP redirection if/unless a condition is matched
7626 May be used in sections : defaults | frontend | listen | backend
7627 no | yes | yes | yes
7628
7629 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007630 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007631
Willy Tarreau0140f252008-11-19 21:07:09 +01007632 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007633 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007634 the HTTP "Location" header. When used in an "http-request" rule,
7635 <loc> value follows the log-format rules and can include some
7636 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007637
7638 <pfx> With "redirect prefix", the "Location" header is built from the
7639 concatenation of <pfx> and the complete URI path, including the
7640 query string, unless the "drop-query" option is specified (see
7641 below). As a special case, if <pfx> equals exactly "/", then
7642 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007643 redirect to the same URL (for instance, to insert a cookie). When
7644 used in an "http-request" rule, <pfx> value follows the log-format
7645 rules and can include some dynamic values (see Custom Log Format
7646 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007647
7648 <sch> With "redirect scheme", then the "Location" header is built by
7649 concatenating <sch> with "://" then the first occurrence of the
7650 "Host" header, and then the URI path, including the query string
7651 unless the "drop-query" option is specified (see below). If no
7652 path is found or if the path is "*", then "/" is used instead. If
7653 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007654 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007655 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007656 HTTPS. When used in an "http-request" rule, <sch> value follows
7657 the log-format rules and can include some dynamic values (see
7658 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007659
7660 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007661 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7662 with 302 used by default if no code is specified. 301 means
7663 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007664 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007665 cache the redirection. 303 is equivalent to 302 except that the
7666 browser will fetch the location with a GET method. 307 is just
7667 like 302 but makes it clear that the same method must be reused.
7668 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007669
7670 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007671 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007672
7673 - "drop-query"
7674 When this keyword is used in a prefix-based redirection, then the
7675 location will be set without any possible query-string, which is useful
7676 for directing users to a non-secure page for instance. It has no effect
7677 with a location-type redirect.
7678
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007679 - "append-slash"
7680 This keyword may be used in conjunction with "drop-query" to redirect
7681 users who use a URL not ending with a '/' to the same one with the '/'.
7682 It can be useful to ensure that search engines will only see one URL.
7683 For this, a return code 301 is preferred.
7684
Willy Tarreau0140f252008-11-19 21:07:09 +01007685 - "set-cookie NAME[=value]"
7686 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7687 to the response. This is sometimes used to indicate that a user has
7688 been seen, for instance to protect against some types of DoS. No other
7689 cookie option is added, so the cookie will be a session cookie. Note
7690 that for a browser, a sole cookie name without an equal sign is
7691 different from a cookie with an equal sign.
7692
7693 - "clear-cookie NAME[=]"
7694 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7695 with the "Max-Age" attribute set to zero. This will tell the browser to
7696 delete this cookie. It is useful for instance on logout pages. It is
7697 important to note that clearing the cookie "NAME" will not remove a
7698 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7699 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007700
7701 Example: move the login URL only to HTTPS.
7702 acl clear dst_port 80
7703 acl secure dst_port 8080
7704 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007705 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007706 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007707 acl cookie_set hdr_sub(cookie) SEEN=1
7708
7709 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007710 redirect prefix https://mysite.com if login_page !secure
7711 redirect prefix http://mysite.com drop-query if login_page !uid_given
7712 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007713 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007714
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007715 Example: send redirects for request for articles without a '/'.
7716 acl missing_slash path_reg ^/article/[^/]*$
7717 redirect code 301 prefix / drop-query append-slash if missing_slash
7718
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007719 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007720 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007721
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007722 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007723 http-request redirect code 301 location \
7724 http://www.%[hdr(host)]%[capture.req.uri] \
7725 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007726
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007727 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007728
Willy Tarreau303c0352008-01-17 19:01:39 +01007729
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007730retries <value>
7731 Set the number of retries to perform on a server after a connection failure
7732 May be used in sections: defaults | frontend | listen | backend
7733 yes | no | yes | yes
7734 Arguments :
7735 <value> is the number of times a connection attempt should be retried on
7736 a server when a connection either is refused or times out. The
7737 default value is 3.
7738
7739 It is important to understand that this value applies to the number of
7740 connection attempts, not full requests. When a connection has effectively
7741 been established to a server, there will be no more retry.
7742
7743 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007744 a turn-around timer of min("timeout connect", one second) is applied before
7745 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007746
7747 When "option redispatch" is set, the last retry may be performed on another
7748 server even if a cookie references a different server.
7749
7750 See also : "option redispatch"
7751
7752
Olivier Houcharda254a372019-04-05 15:30:12 +02007753retry-on [list of keywords]
7754 Specify when to attempt to automatically retry a failed request
7755 May be used in sections: defaults | frontend | listen | backend
7756 yes | no | yes | yes
7757 Arguments :
7758 <keywords> is a list of keywords or HTTP status codes, each representing a
7759 type of failure event on which an attempt to retry the request
7760 is desired. Please read the notes at the bottom before changing
7761 this setting. The following keywords are supported :
7762
7763 none never retry
7764
7765 conn-failure retry when the connection or the SSL handshake failed
7766 and the request could not be sent. This is the default.
7767
7768 empty-response retry when the server connection was closed after part
7769 of the request was sent, and nothing was received from
7770 the server. This type of failure may be caused by the
7771 request timeout on the server side, poor network
7772 condition, or a server crash or restart while
7773 processing the request.
7774
Olivier Houcharde3249a92019-05-03 23:01:47 +02007775 junk-response retry when the server returned something not looking
7776 like a complete HTTP response. This includes partial
7777 responses headers as well as non-HTTP contents. It
7778 usually is a bad idea to retry on such events, which
7779 may be caused a configuration issue (wrong server port)
7780 or by the request being harmful to the server (buffer
7781 overflow attack for example).
7782
Olivier Houcharda254a372019-04-05 15:30:12 +02007783 response-timeout the server timeout stroke while waiting for the server
7784 to respond to the request. This may be caused by poor
7785 network condition, the reuse of an idle connection
7786 which has expired on the path, or by the request being
7787 extremely expensive to process. It generally is a bad
7788 idea to retry on such events on servers dealing with
7789 heavy database processing (full scans, etc) as it may
7790 amplify denial of service attacks.
7791
Olivier Houchard865d8392019-05-03 22:46:27 +02007792 0rtt-rejected retry requests which were sent over early data and were
7793 rejected by the server. These requests are generally
7794 considered to be safe to retry.
7795
Olivier Houcharda254a372019-04-05 15:30:12 +02007796 <status> any HTTP status code among "404" (Not Found), "408"
7797 (Request Timeout), "425" (Too Early), "500" (Server
7798 Error), "501" (Not Implemented), "502" (Bad Gateway),
7799 "503" (Service Unavailable), "504" (Gateway Timeout).
7800
Olivier Houchardddf0e032019-05-10 18:05:40 +02007801 all-retryable-errors
7802 retry request for any error that are considered
7803 retryable. This currently activates "conn-failure",
7804 "empty-response", "junk-response", "response-timeout",
7805 "0rtt-rejected", "500", "502", "503", and "504".
7806
Olivier Houcharda254a372019-04-05 15:30:12 +02007807 Using this directive replaces any previous settings with the new ones; it is
7808 not cumulative.
7809
7810 Please note that using anything other than "none" and "conn-failure" requires
7811 to allocate a buffer and copy the whole request into it, so it has memory and
7812 performance impacts. Requests not fitting in a single buffer will never be
7813 retried (see the global tune.bufsize setting).
7814
7815 You have to make sure the application has a replay protection mechanism built
7816 in such as a unique transaction IDs passed in requests, or that replaying the
7817 same request has no consequence, or it is very dangerous to use any retry-on
7818 value beside "conn-failure" and "none". Static file servers and caches are
7819 generally considered safe against any type of retry. Using a status code can
7820 be useful to quickly leave a server showing an abnormal behavior (out of
7821 memory, file system issues, etc), but in this case it may be a good idea to
7822 immediately redispatch the connection to another server (please see "option
7823 redispatch" for this). Last, it is important to understand that most causes
7824 of failures are the requests themselves and that retrying a request causing a
7825 server to misbehave will often make the situation even worse for this server,
7826 or for the whole service in case of redispatch.
7827
7828 Unless you know exactly how the application deals with replayed requests, you
7829 should not use this directive.
7830
7831 The default is "conn-failure".
7832
7833 See also: "retries", "option redispatch", "tune.bufsize"
7834
David du Colombier486df472011-03-17 10:40:26 +01007835server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007836 Declare a server in a backend
7837 May be used in sections : defaults | frontend | listen | backend
7838 no | no | yes | yes
7839 Arguments :
7840 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007841 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007842 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007843
David du Colombier486df472011-03-17 10:40:26 +01007844 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7845 resolvable hostname is supported, but this name will be resolved
7846 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7847 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007848 address as the one from the client connection. This is useful in
7849 transparent proxy architectures where the client's connection is
7850 intercepted and haproxy must forward to the original destination
7851 address. This is more or less what the "transparent" keyword does
7852 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007853 to report statistics. Optionally, an address family prefix may be
7854 used before the address to force the family regardless of the
7855 address format, which can be useful to specify a path to a unix
7856 socket with no slash ('/'). Currently supported prefixes are :
7857 - 'ipv4@' -> address is always IPv4
7858 - 'ipv6@' -> address is always IPv6
7859 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007860 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02007861 - 'sockpair@' -> address is the FD of a connected unix
7862 socket or of a socketpair. During a connection, the
7863 backend creates a pair of connected sockets, and passes
7864 one of them over the FD. The bind part will use the
7865 received socket as the client FD. Should be used
7866 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02007867 You may want to reference some environment variables in the
7868 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007869 variables. The "init-addr" setting can be used to modify the way
7870 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007871
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007872 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007873 be sent to this port. If unset, the same port the client
7874 connected to will be used. The port may also be prefixed by a "+"
7875 or a "-". In this case, the server's port will be determined by
7876 adding this value to the client's port.
7877
7878 <param*> is a list of parameters for this server. The "server" keywords
7879 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007880 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007881
7882 Examples :
7883 server first 10.1.1.1:1080 cookie first check inter 1000
7884 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007885 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007886 server backup "${SRV_BACKUP}:1080" backup
7887 server www1_dc1 "${LAN_DC1}.101:80"
7888 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007889
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007890 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7891 sun_path length is used for the address length. Some other programs
7892 such as socat use the string length only by default. Pass the option
7893 ",unix-tightsocklen=0" to any abstract socket definition in socat to
7894 make it compatible with HAProxy's.
7895
Mark Lamourinec2247f02012-01-04 13:02:01 -05007896 See also: "default-server", "http-send-name-header" and section 5 about
7897 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007898
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007899server-state-file-name [<file>]
7900 Set the server state file to read, load and apply to servers available in
7901 this backend. It only applies when the directive "load-server-state-from-file"
7902 is set to "local". When <file> is not provided or if this directive is not
7903 set, then backend name is used. If <file> starts with a slash '/', then it is
7904 considered as an absolute path. Otherwise, <file> is concatenated to the
7905 global directive "server-state-file-base".
7906
7907 Example: the minimal configuration below would make HAProxy look for the
7908 state server file '/etc/haproxy/states/bk':
7909
7910 global
7911 server-state-file-base /etc/haproxy/states
7912
7913 backend bk
7914 load-server-state-from-file
7915
7916 See also: "server-state-file-base", "load-server-state-from-file", and
7917 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007918
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02007919server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
7920 Set a template to initialize servers with shared parameters.
7921 The names of these servers are built from <prefix> and <num | range> parameters.
7922 May be used in sections : defaults | frontend | listen | backend
7923 no | no | yes | yes
7924
7925 Arguments:
7926 <prefix> A prefix for the server names to be built.
7927
7928 <num | range>
7929 If <num> is provided, this template initializes <num> servers
7930 with 1 up to <num> as server name suffixes. A range of numbers
7931 <num_low>-<num_high> may also be used to use <num_low> up to
7932 <num_high> as server name suffixes.
7933
7934 <fqdn> A FQDN for all the servers this template initializes.
7935
7936 <port> Same meaning as "server" <port> argument (see "server" keyword).
7937
7938 <params*>
7939 Remaining server parameters among all those supported by "server"
7940 keyword.
7941
7942 Examples:
7943 # Initializes 3 servers with srv1, srv2 and srv3 as names,
7944 # google.com as FQDN, and health-check enabled.
7945 server-template srv 1-3 google.com:80 check
7946
7947 # or
7948 server-template srv 3 google.com:80 check
7949
7950 # would be equivalent to:
7951 server srv1 google.com:80 check
7952 server srv2 google.com:80 check
7953 server srv3 google.com:80 check
7954
7955
7956
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007957source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02007958source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007959source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007960 Set the source address for outgoing connections
7961 May be used in sections : defaults | frontend | listen | backend
7962 yes | no | yes | yes
7963 Arguments :
7964 <addr> is the IPv4 address HAProxy will bind to before connecting to a
7965 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01007966
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007967 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01007968 the most appropriate address to reach its destination. Optionally
7969 an address family prefix may be used before the address to force
7970 the family regardless of the address format, which can be useful
7971 to specify a path to a unix socket with no slash ('/'). Currently
7972 supported prefixes are :
7973 - 'ipv4@' -> address is always IPv4
7974 - 'ipv6@' -> address is always IPv6
7975 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007976 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02007977 You may want to reference some environment variables in the
7978 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007979
7980 <port> is an optional port. It is normally not needed but may be useful
7981 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02007982 the system will select a free port. Note that port ranges are not
7983 supported in the backend. If you want to force port ranges, you
7984 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007985
7986 <addr2> is the IP address to present to the server when connections are
7987 forwarded in full transparent proxy mode. This is currently only
7988 supported on some patched Linux kernels. When this address is
7989 specified, clients connecting to the server will be presented
7990 with this address, while health checks will still use the address
7991 <addr>.
7992
7993 <port2> is the optional port to present to the server when connections
7994 are forwarded in full transparent proxy mode (see <addr2> above).
7995 The default value of zero means the system will select a free
7996 port.
7997
Willy Tarreaubce70882009-09-07 11:51:47 +02007998 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
7999 This is the name of a comma-separated header list which can
8000 contain multiple IP addresses. By default, the last occurrence is
8001 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008002 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008003 by previous proxy, typically Stunnel. In order to use another
8004 occurrence from the last one, please see the <occ> parameter
8005 below. When the header (or occurrence) is not found, no binding
8006 is performed so that the proxy's default IP address is used. Also
8007 keep in mind that the header name is case insensitive, as for any
8008 HTTP header.
8009
8010 <occ> is the occurrence number of a value to be used in a multi-value
8011 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008012 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008013 address. Positive values indicate a position from the first
8014 occurrence, 1 being the first one. Negative values indicate
8015 positions relative to the last one, -1 being the last one. This
8016 is helpful for situations where an X-Forwarded-For header is set
8017 at the entry point of an infrastructure and must be used several
8018 proxy layers away. When this value is not specified, -1 is
8019 assumed. Passing a zero here disables the feature.
8020
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008021 <name> is an optional interface name to which to bind to for outgoing
8022 traffic. On systems supporting this features (currently, only
8023 Linux), this allows one to bind all traffic to the server to
8024 this interface even if it is not the one the system would select
8025 based on routing tables. This should be used with extreme care.
8026 Note that using this option requires root privileges.
8027
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008028 The "source" keyword is useful in complex environments where a specific
8029 address only is allowed to connect to the servers. It may be needed when a
8030 private address must be used through a public gateway for instance, and it is
8031 known that the system cannot determine the adequate source address by itself.
8032
8033 An extension which is available on certain patched Linux kernels may be used
8034 through the "usesrc" optional keyword. It makes it possible to connect to the
8035 servers with an IP address which does not belong to the system itself. This
8036 is called "full transparent proxy mode". For this to work, the destination
8037 servers have to route their traffic back to this address through the machine
8038 running HAProxy, and IP forwarding must generally be enabled on this machine.
8039
8040 In this "full transparent proxy" mode, it is possible to force a specific IP
8041 address to be presented to the servers. This is not much used in fact. A more
8042 common use is to tell HAProxy to present the client's IP address. For this,
8043 there are two methods :
8044
8045 - present the client's IP and port addresses. This is the most transparent
8046 mode, but it can cause problems when IP connection tracking is enabled on
8047 the machine, because a same connection may be seen twice with different
8048 states. However, this solution presents the huge advantage of not
8049 limiting the system to the 64k outgoing address+port couples, because all
8050 of the client ranges may be used.
8051
8052 - present only the client's IP address and select a spare port. This
8053 solution is still quite elegant but slightly less transparent (downstream
8054 firewalls logs will not match upstream's). It also presents the downside
8055 of limiting the number of concurrent connections to the usual 64k ports.
8056 However, since the upstream and downstream ports are different, local IP
8057 connection tracking on the machine will not be upset by the reuse of the
8058 same session.
8059
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008060 This option sets the default source for all servers in the backend. It may
8061 also be specified in a "defaults" section. Finer source address specification
8062 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008063 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008064
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008065 In order to work, "usesrc" requires root privileges.
8066
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008067 Examples :
8068 backend private
8069 # Connect to the servers using our 192.168.1.200 source address
8070 source 192.168.1.200
8071
8072 backend transparent_ssl1
8073 # Connect to the SSL farm from the client's source address
8074 source 192.168.1.200 usesrc clientip
8075
8076 backend transparent_ssl2
8077 # Connect to the SSL farm from the client's source address and port
8078 # not recommended if IP conntrack is present on the local machine.
8079 source 192.168.1.200 usesrc client
8080
8081 backend transparent_ssl3
8082 # Connect to the SSL farm from the client's source address. It
8083 # is more conntrack-friendly.
8084 source 192.168.1.200 usesrc clientip
8085
8086 backend transparent_smtp
8087 # Connect to the SMTP farm from the client's source address/port
8088 # with Tproxy version 4.
8089 source 0.0.0.0 usesrc clientip
8090
Willy Tarreaubce70882009-09-07 11:51:47 +02008091 backend transparent_http
8092 # Connect to the servers using the client's IP as seen by previous
8093 # proxy.
8094 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8095
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008096 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008097 the Linux kernel on www.balabit.com, the "bind" keyword.
8098
Willy Tarreau844e3c52008-01-11 16:28:18 +01008099
Cyril Bonté66c327d2010-10-12 00:14:37 +02008100stats admin { if | unless } <cond>
8101 Enable statistics admin level if/unless a condition is matched
8102 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008103 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008104
8105 This statement enables the statistics admin level if/unless a condition is
8106 matched.
8107
8108 The admin level allows to enable/disable servers from the web interface. By
8109 default, statistics page is read-only for security reasons.
8110
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008111 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8112 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008113 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008114
Cyril Bonté23b39d92011-02-10 22:54:44 +01008115 Currently, the POST request is limited to the buffer size minus the reserved
8116 buffer space, which means that if the list of servers is too long, the
8117 request won't be processed. It is recommended to alter few servers at a
8118 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008119
8120 Example :
8121 # statistics admin level only for localhost
8122 backend stats_localhost
8123 stats enable
8124 stats admin if LOCALHOST
8125
8126 Example :
8127 # statistics admin level always enabled because of the authentication
8128 backend stats_auth
8129 stats enable
8130 stats auth admin:AdMiN123
8131 stats admin if TRUE
8132
8133 Example :
8134 # statistics admin level depends on the authenticated user
8135 userlist stats-auth
8136 group admin users admin
8137 user admin insecure-password AdMiN123
8138 group readonly users haproxy
8139 user haproxy insecure-password haproxy
8140
8141 backend stats_auth
8142 stats enable
8143 acl AUTH http_auth(stats-auth)
8144 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8145 stats http-request auth unless AUTH
8146 stats admin if AUTH_ADMIN
8147
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008148 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8149 "bind-process", section 3.4 about userlists and section 7 about
8150 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008151
8152
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008153stats auth <user>:<passwd>
8154 Enable statistics with authentication and grant access to an account
8155 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008156 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008157 Arguments :
8158 <user> is a user name to grant access to
8159
8160 <passwd> is the cleartext password associated to this user
8161
8162 This statement enables statistics with default settings, and restricts access
8163 to declared users only. It may be repeated as many times as necessary to
8164 allow as many users as desired. When a user tries to access the statistics
8165 without a valid account, a "401 Forbidden" response will be returned so that
8166 the browser asks the user to provide a valid user and password. The real
8167 which will be returned to the browser is configurable using "stats realm".
8168
8169 Since the authentication method is HTTP Basic Authentication, the passwords
8170 circulate in cleartext on the network. Thus, it was decided that the
8171 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008172 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008173
8174 It is also possible to reduce the scope of the proxies which appear in the
8175 report using "stats scope".
8176
8177 Though this statement alone is enough to enable statistics reporting, it is
8178 recommended to set all other settings in order to avoid relying on default
8179 unobvious parameters.
8180
8181 Example :
8182 # public access (limited to this backend only)
8183 backend public_www
8184 server srv1 192.168.0.1:80
8185 stats enable
8186 stats hide-version
8187 stats scope .
8188 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008189 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008190 stats auth admin1:AdMiN123
8191 stats auth admin2:AdMiN321
8192
8193 # internal monitoring access (unlimited)
8194 backend private_monitoring
8195 stats enable
8196 stats uri /admin?stats
8197 stats refresh 5s
8198
8199 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8200
8201
8202stats enable
8203 Enable statistics reporting with default settings
8204 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008205 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008206 Arguments : none
8207
8208 This statement enables statistics reporting with default settings defined
8209 at build time. Unless stated otherwise, these settings are used :
8210 - stats uri : /haproxy?stats
8211 - stats realm : "HAProxy Statistics"
8212 - stats auth : no authentication
8213 - stats scope : no restriction
8214
8215 Though this statement alone is enough to enable statistics reporting, it is
8216 recommended to set all other settings in order to avoid relying on default
8217 unobvious parameters.
8218
8219 Example :
8220 # public access (limited to this backend only)
8221 backend public_www
8222 server srv1 192.168.0.1:80
8223 stats enable
8224 stats hide-version
8225 stats scope .
8226 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008227 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008228 stats auth admin1:AdMiN123
8229 stats auth admin2:AdMiN321
8230
8231 # internal monitoring access (unlimited)
8232 backend private_monitoring
8233 stats enable
8234 stats uri /admin?stats
8235 stats refresh 5s
8236
8237 See also : "stats auth", "stats realm", "stats uri"
8238
8239
Willy Tarreaud63335a2010-02-26 12:56:52 +01008240stats hide-version
8241 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008242 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008243 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008244 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008245
Willy Tarreaud63335a2010-02-26 12:56:52 +01008246 By default, the stats page reports some useful status information along with
8247 the statistics. Among them is HAProxy's version. However, it is generally
8248 considered dangerous to report precise version to anyone, as it can help them
8249 target known weaknesses with specific attacks. The "stats hide-version"
8250 statement removes the version from the statistics report. This is recommended
8251 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008252
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008253 Though this statement alone is enough to enable statistics reporting, it is
8254 recommended to set all other settings in order to avoid relying on default
8255 unobvious parameters.
8256
Willy Tarreaud63335a2010-02-26 12:56:52 +01008257 Example :
8258 # public access (limited to this backend only)
8259 backend public_www
8260 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008261 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008262 stats hide-version
8263 stats scope .
8264 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008265 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008266 stats auth admin1:AdMiN123
8267 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008268
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008269 # internal monitoring access (unlimited)
8270 backend private_monitoring
8271 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008272 stats uri /admin?stats
8273 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008274
Willy Tarreaud63335a2010-02-26 12:56:52 +01008275 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008276
Willy Tarreau983e01e2010-01-11 18:42:06 +01008277
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008278stats http-request { allow | deny | auth [realm <realm>] }
8279 [ { if | unless } <condition> ]
8280 Access control for statistics
8281
8282 May be used in sections: defaults | frontend | listen | backend
8283 no | no | yes | yes
8284
8285 As "http-request", these set of options allow to fine control access to
8286 statistics. Each option may be followed by if/unless and acl.
8287 First option with matched condition (or option without condition) is final.
8288 For "deny" a 403 error will be returned, for "allow" normal processing is
8289 performed, for "auth" a 401/407 error code is returned so the client
8290 should be asked to enter a username and password.
8291
8292 There is no fixed limit to the number of http-request statements per
8293 instance.
8294
8295 See also : "http-request", section 3.4 about userlists and section 7
8296 about ACL usage.
8297
8298
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008299stats realm <realm>
8300 Enable statistics and set authentication realm
8301 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008302 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008303 Arguments :
8304 <realm> is the name of the HTTP Basic Authentication realm reported to
8305 the browser. The browser uses it to display it in the pop-up
8306 inviting the user to enter a valid username and password.
8307
8308 The realm is read as a single word, so any spaces in it should be escaped
8309 using a backslash ('\').
8310
8311 This statement is useful only in conjunction with "stats auth" since it is
8312 only related to authentication.
8313
8314 Though this statement alone is enough to enable statistics reporting, it is
8315 recommended to set all other settings in order to avoid relying on default
8316 unobvious parameters.
8317
8318 Example :
8319 # public access (limited to this backend only)
8320 backend public_www
8321 server srv1 192.168.0.1:80
8322 stats enable
8323 stats hide-version
8324 stats scope .
8325 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008326 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008327 stats auth admin1:AdMiN123
8328 stats auth admin2:AdMiN321
8329
8330 # internal monitoring access (unlimited)
8331 backend private_monitoring
8332 stats enable
8333 stats uri /admin?stats
8334 stats refresh 5s
8335
8336 See also : "stats auth", "stats enable", "stats uri"
8337
8338
8339stats refresh <delay>
8340 Enable statistics with automatic refresh
8341 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008342 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008343 Arguments :
8344 <delay> is the suggested refresh delay, specified in seconds, which will
8345 be returned to the browser consulting the report page. While the
8346 browser is free to apply any delay, it will generally respect it
8347 and refresh the page this every seconds. The refresh interval may
8348 be specified in any other non-default time unit, by suffixing the
8349 unit after the value, as explained at the top of this document.
8350
8351 This statement is useful on monitoring displays with a permanent page
8352 reporting the load balancer's activity. When set, the HTML report page will
8353 include a link "refresh"/"stop refresh" so that the user can select whether
8354 he wants automatic refresh of the page or not.
8355
8356 Though this statement alone is enough to enable statistics reporting, it is
8357 recommended to set all other settings in order to avoid relying on default
8358 unobvious parameters.
8359
8360 Example :
8361 # public access (limited to this backend only)
8362 backend public_www
8363 server srv1 192.168.0.1:80
8364 stats enable
8365 stats hide-version
8366 stats scope .
8367 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008368 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008369 stats auth admin1:AdMiN123
8370 stats auth admin2:AdMiN321
8371
8372 # internal monitoring access (unlimited)
8373 backend private_monitoring
8374 stats enable
8375 stats uri /admin?stats
8376 stats refresh 5s
8377
8378 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8379
8380
8381stats scope { <name> | "." }
8382 Enable statistics and limit access scope
8383 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008384 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008385 Arguments :
8386 <name> is the name of a listen, frontend or backend section to be
8387 reported. The special name "." (a single dot) designates the
8388 section in which the statement appears.
8389
8390 When this statement is specified, only the sections enumerated with this
8391 statement will appear in the report. All other ones will be hidden. This
8392 statement may appear as many times as needed if multiple sections need to be
8393 reported. Please note that the name checking is performed as simple string
8394 comparisons, and that it is never checked that a give section name really
8395 exists.
8396
8397 Though this statement alone is enough to enable statistics reporting, it is
8398 recommended to set all other settings in order to avoid relying on default
8399 unobvious parameters.
8400
8401 Example :
8402 # public access (limited to this backend only)
8403 backend public_www
8404 server srv1 192.168.0.1:80
8405 stats enable
8406 stats hide-version
8407 stats scope .
8408 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008409 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008410 stats auth admin1:AdMiN123
8411 stats auth admin2:AdMiN321
8412
8413 # internal monitoring access (unlimited)
8414 backend private_monitoring
8415 stats enable
8416 stats uri /admin?stats
8417 stats refresh 5s
8418
8419 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8420
Willy Tarreaud63335a2010-02-26 12:56:52 +01008421
Willy Tarreauc9705a12010-07-27 20:05:50 +02008422stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008423 Enable reporting of a description on the statistics page.
8424 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008425 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008426
Willy Tarreauc9705a12010-07-27 20:05:50 +02008427 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008428 description from global section is automatically used instead.
8429
8430 This statement is useful for users that offer shared services to their
8431 customers, where node or description should be different for each customer.
8432
8433 Though this statement alone is enough to enable statistics reporting, it is
8434 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008435 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008436
8437 Example :
8438 # internal monitoring access (unlimited)
8439 backend private_monitoring
8440 stats enable
8441 stats show-desc Master node for Europe, Asia, Africa
8442 stats uri /admin?stats
8443 stats refresh 5s
8444
8445 See also: "show-node", "stats enable", "stats uri" and "description" in
8446 global section.
8447
8448
8449stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008450 Enable reporting additional information on the statistics page
8451 May be used in sections : defaults | frontend | listen | backend
8452 yes | yes | yes | yes
8453 Arguments : none
8454
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008455 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008456 - cap: capabilities (proxy)
8457 - mode: one of tcp, http or health (proxy)
8458 - id: SNMP ID (proxy, socket, server)
8459 - IP (socket, server)
8460 - cookie (backend, server)
8461
8462 Though this statement alone is enough to enable statistics reporting, it is
8463 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008464 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008465
8466 See also: "stats enable", "stats uri".
8467
8468
8469stats show-node [ <name> ]
8470 Enable reporting of a host name on the statistics page.
8471 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008472 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008473 Arguments:
8474 <name> is an optional name to be reported. If unspecified, the
8475 node name from global section is automatically used instead.
8476
8477 This statement is useful for users that offer shared services to their
8478 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008479 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008480
8481 Though this statement alone is enough to enable statistics reporting, it is
8482 recommended to set all other settings in order to avoid relying on default
8483 unobvious parameters.
8484
8485 Example:
8486 # internal monitoring access (unlimited)
8487 backend private_monitoring
8488 stats enable
8489 stats show-node Europe-1
8490 stats uri /admin?stats
8491 stats refresh 5s
8492
8493 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8494 section.
8495
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008496
8497stats uri <prefix>
8498 Enable statistics and define the URI prefix to access them
8499 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008500 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008501 Arguments :
8502 <prefix> is the prefix of any URI which will be redirected to stats. This
8503 prefix may contain a question mark ('?') to indicate part of a
8504 query string.
8505
8506 The statistics URI is intercepted on the relayed traffic, so it appears as a
8507 page within the normal application. It is strongly advised to ensure that the
8508 selected URI will never appear in the application, otherwise it will never be
8509 possible to reach it in the application.
8510
8511 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008512 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008513 It is generally a good idea to include a question mark in the URI so that
8514 intermediate proxies refrain from caching the results. Also, since any string
8515 beginning with the prefix will be accepted as a stats request, the question
8516 mark helps ensuring that no valid URI will begin with the same words.
8517
8518 It is sometimes very convenient to use "/" as the URI prefix, and put that
8519 statement in a "listen" instance of its own. That makes it easy to dedicate
8520 an address or a port to statistics only.
8521
8522 Though this statement alone is enough to enable statistics reporting, it is
8523 recommended to set all other settings in order to avoid relying on default
8524 unobvious parameters.
8525
8526 Example :
8527 # public access (limited to this backend only)
8528 backend public_www
8529 server srv1 192.168.0.1:80
8530 stats enable
8531 stats hide-version
8532 stats scope .
8533 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008534 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008535 stats auth admin1:AdMiN123
8536 stats auth admin2:AdMiN321
8537
8538 # internal monitoring access (unlimited)
8539 backend private_monitoring
8540 stats enable
8541 stats uri /admin?stats
8542 stats refresh 5s
8543
8544 See also : "stats auth", "stats enable", "stats realm"
8545
8546
Willy Tarreaud63335a2010-02-26 12:56:52 +01008547stick match <pattern> [table <table>] [{if | unless} <cond>]
8548 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008549 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008550 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008551
8552 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008553 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008554 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008555 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008556 stickiness table. This rule is mandatory.
8557
8558 <table> is an optional stickiness table name. If unspecified, the same
8559 backend's table is used. A stickiness table is declared using
8560 the "stick-table" statement.
8561
8562 <cond> is an optional matching condition. It makes it possible to match
8563 on a certain criterion only when other conditions are met (or
8564 not met). For instance, it could be used to match on a source IP
8565 address except when a request passes through a known proxy, in
8566 which case we'd match on a header containing that IP address.
8567
8568 Some protocols or applications require complex stickiness rules and cannot
8569 always simply rely on cookies nor hashing. The "stick match" statement
8570 describes a rule to extract the stickiness criterion from an incoming request
8571 or connection. See section 7 for a complete list of possible patterns and
8572 transformation rules.
8573
8574 The table has to be declared using the "stick-table" statement. It must be of
8575 a type compatible with the pattern. By default it is the one which is present
8576 in the same backend. It is possible to share a table with other backends by
8577 referencing it using the "table" keyword. If another table is referenced,
8578 the server's ID inside the backends are used. By default, all server IDs
8579 start at 1 in each backend, so the server ordering is enough. But in case of
8580 doubt, it is highly recommended to force server IDs using their "id" setting.
8581
8582 It is possible to restrict the conditions where a "stick match" statement
8583 will apply, using "if" or "unless" followed by a condition. See section 7 for
8584 ACL based conditions.
8585
8586 There is no limit on the number of "stick match" statements. The first that
8587 applies and matches will cause the request to be directed to the same server
8588 as was used for the request which created the entry. That way, multiple
8589 matches can be used as fallbacks.
8590
8591 The stick rules are checked after the persistence cookies, so they will not
8592 affect stickiness if a cookie has already been used to select a server. That
8593 way, it becomes very easy to insert cookies and match on IP addresses in
8594 order to maintain stickiness between HTTP and HTTPS.
8595
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008596 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8597 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008598 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008599
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008600 Example :
8601 # forward SMTP users to the same server they just used for POP in the
8602 # last 30 minutes
8603 backend pop
8604 mode tcp
8605 balance roundrobin
8606 stick store-request src
8607 stick-table type ip size 200k expire 30m
8608 server s1 192.168.1.1:110
8609 server s2 192.168.1.1:110
8610
8611 backend smtp
8612 mode tcp
8613 balance roundrobin
8614 stick match src table pop
8615 server s1 192.168.1.1:25
8616 server s2 192.168.1.1:25
8617
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008618 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008619 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008620
8621
8622stick on <pattern> [table <table>] [{if | unless} <condition>]
8623 Define a request pattern to associate a user to a server
8624 May be used in sections : defaults | frontend | listen | backend
8625 no | no | yes | yes
8626
8627 Note : This form is exactly equivalent to "stick match" followed by
8628 "stick store-request", all with the same arguments. Please refer
8629 to both keywords for details. It is only provided as a convenience
8630 for writing more maintainable configurations.
8631
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008632 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8633 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008634 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008635
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008636 Examples :
8637 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008638 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008639
8640 # ...is strictly equivalent to this one :
8641 stick match src table pop if !localhost
8642 stick store-request src table pop if !localhost
8643
8644
8645 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8646 # well as HTTP without cookie. Share the same table between both accesses.
8647 backend http
8648 mode http
8649 balance roundrobin
8650 stick on src table https
8651 cookie SRV insert indirect nocache
8652 server s1 192.168.1.1:80 cookie s1
8653 server s2 192.168.1.1:80 cookie s2
8654
8655 backend https
8656 mode tcp
8657 balance roundrobin
8658 stick-table type ip size 200k expire 30m
8659 stick on src
8660 server s1 192.168.1.1:443
8661 server s2 192.168.1.1:443
8662
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008663 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008664
8665
8666stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8667 Define a request pattern used to create an entry in a stickiness table
8668 May be used in sections : defaults | frontend | listen | backend
8669 no | no | yes | yes
8670
8671 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008672 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008673 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008674 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008675 server is selected.
8676
8677 <table> is an optional stickiness table name. If unspecified, the same
8678 backend's table is used. A stickiness table is declared using
8679 the "stick-table" statement.
8680
8681 <cond> is an optional storage condition. It makes it possible to store
8682 certain criteria only when some conditions are met (or not met).
8683 For instance, it could be used to store the source IP address
8684 except when the request passes through a known proxy, in which
8685 case we'd store a converted form of a header containing that IP
8686 address.
8687
8688 Some protocols or applications require complex stickiness rules and cannot
8689 always simply rely on cookies nor hashing. The "stick store-request" statement
8690 describes a rule to decide what to extract from the request and when to do
8691 it, in order to store it into a stickiness table for further requests to
8692 match it using the "stick match" statement. Obviously the extracted part must
8693 make sense and have a chance to be matched in a further request. Storing a
8694 client's IP address for instance often makes sense. Storing an ID found in a
8695 URL parameter also makes sense. Storing a source port will almost never make
8696 any sense because it will be randomly matched. See section 7 for a complete
8697 list of possible patterns and transformation rules.
8698
8699 The table has to be declared using the "stick-table" statement. It must be of
8700 a type compatible with the pattern. By default it is the one which is present
8701 in the same backend. It is possible to share a table with other backends by
8702 referencing it using the "table" keyword. If another table is referenced,
8703 the server's ID inside the backends are used. By default, all server IDs
8704 start at 1 in each backend, so the server ordering is enough. But in case of
8705 doubt, it is highly recommended to force server IDs using their "id" setting.
8706
8707 It is possible to restrict the conditions where a "stick store-request"
8708 statement will apply, using "if" or "unless" followed by a condition. This
8709 condition will be evaluated while parsing the request, so any criteria can be
8710 used. See section 7 for ACL based conditions.
8711
8712 There is no limit on the number of "stick store-request" statements, but
8713 there is a limit of 8 simultaneous stores per request or response. This
8714 makes it possible to store up to 8 criteria, all extracted from either the
8715 request or the response, regardless of the number of rules. Only the 8 first
8716 ones which match will be kept. Using this, it is possible to feed multiple
8717 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008718 another protocol or access method. Using multiple store-request rules with
8719 the same table is possible and may be used to find the best criterion to rely
8720 on, by arranging the rules by decreasing preference order. Only the first
8721 extracted criterion for a given table will be stored. All subsequent store-
8722 request rules referencing the same table will be skipped and their ACLs will
8723 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008724
8725 The "store-request" rules are evaluated once the server connection has been
8726 established, so that the table will contain the real server that processed
8727 the request.
8728
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008729 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8730 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008731 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008732
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008733 Example :
8734 # forward SMTP users to the same server they just used for POP in the
8735 # last 30 minutes
8736 backend pop
8737 mode tcp
8738 balance roundrobin
8739 stick store-request src
8740 stick-table type ip size 200k expire 30m
8741 server s1 192.168.1.1:110
8742 server s2 192.168.1.1:110
8743
8744 backend smtp
8745 mode tcp
8746 balance roundrobin
8747 stick match src table pop
8748 server s1 192.168.1.1:25
8749 server s2 192.168.1.1:25
8750
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008751 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008752 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008753
8754
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008755stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008756 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8757 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008758 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008759 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008760 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008761
8762 Arguments :
8763 ip a table declared with "type ip" will only store IPv4 addresses.
8764 This form is very compact (about 50 bytes per entry) and allows
8765 very fast entry lookup and stores with almost no overhead. This
8766 is mainly used to store client source IP addresses.
8767
David du Colombier9a6d3c92011-03-17 10:40:24 +01008768 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8769 This form is very compact (about 60 bytes per entry) and allows
8770 very fast entry lookup and stores with almost no overhead. This
8771 is mainly used to store client source IP addresses.
8772
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008773 integer a table declared with "type integer" will store 32bit integers
8774 which can represent a client identifier found in a request for
8775 instance.
8776
8777 string a table declared with "type string" will store substrings of up
8778 to <len> characters. If the string provided by the pattern
8779 extractor is larger than <len>, it will be truncated before
8780 being stored. During matching, at most <len> characters will be
8781 compared between the string in the table and the extracted
8782 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008783 to 32 characters.
8784
8785 binary a table declared with "type binary" will store binary blocks
8786 of <len> bytes. If the block provided by the pattern
8787 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008788 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008789 is shorter than <len>, it will be padded by 0. When not
8790 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008791
8792 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008793 "string" type table (See type "string" above). Or the number
8794 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008795 changing this parameter as memory usage will proportionally
8796 increase.
8797
8798 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008799 value directly impacts memory usage. Count approximately
8800 50 bytes per entry, plus the size of a string if any. The size
8801 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008802
8803 [nopurge] indicates that we refuse to purge older entries when the table
8804 is full. When not specified and the table is full when haproxy
8805 wants to store an entry in it, it will flush a few of the oldest
8806 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008807 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008808 be desirable to refuse new entries instead of purging the older
8809 ones. That may be the case when the amount of data to store is
8810 far above the hardware limits and we prefer not to offer access
8811 to new clients than to reject the ones already connected. When
8812 using this parameter, be sure to properly set the "expire"
8813 parameter (see below).
8814
Emeric Brunf099e792010-09-27 12:05:28 +02008815 <peersect> is the name of the peers section to use for replication. Entries
8816 which associate keys to server IDs are kept synchronized with
8817 the remote peers declared in this section. All entries are also
8818 automatically learned from the local peer (old process) during a
8819 soft restart.
8820
Willy Tarreau1abc6732015-05-01 19:21:02 +02008821 NOTE : each peers section may be referenced only by tables
8822 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008823
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008824 <expire> defines the maximum duration of an entry in the table since it
8825 was last created, refreshed or matched. The expiration delay is
8826 defined using the standard time format, similarly as the various
8827 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008828 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008829 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008830 be removed once full. Be sure not to use the "nopurge" parameter
8831 if not expiration delay is specified.
8832
Willy Tarreau08d5f982010-06-06 13:34:54 +02008833 <data_type> is used to store additional information in the stick-table. This
8834 may be used by ACLs in order to control various criteria related
8835 to the activity of the client matching the stick-table. For each
8836 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008837 that the additional data can fit. Several data types may be
8838 stored with an entry. Multiple data types may be specified after
8839 the "store" keyword, as a comma-separated list. Alternatively,
8840 it is possible to repeat the "store" keyword followed by one or
8841 several data types. Except for the "server_id" type which is
8842 automatically detected and enabled, all data types must be
8843 explicitly declared to be stored. If an ACL references a data
8844 type which is not stored, the ACL will simply not match. Some
8845 data types require an argument which must be passed just after
8846 the type between parenthesis. See below for the supported data
8847 types and their arguments.
8848
8849 The data types that can be stored with an entry are the following :
8850 - server_id : this is an integer which holds the numeric ID of the server a
8851 request was assigned to. It is used by the "stick match", "stick store",
8852 and "stick on" rules. It is automatically enabled when referenced.
8853
8854 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
8855 integer which may be used for anything. Most of the time it will be used
8856 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01008857 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02008858
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008859 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
8860 over a period. It is a positive 32-bit integer integer which may be used
8861 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01008862 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008863 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01008864 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008865
Frédéric Lécaille6778b272018-01-29 15:22:53 +01008866 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
8867 integer which may be used for anything. Most of the time it will be used
8868 to put a special tag on some entries, for instance to note that a
8869 specific behavior was detected and must be known for future matches.
8870
8871 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
8872 over a period. It is a positive 32-bit integer integer which may be used
8873 for anything. Just like <gpc1>, it counts events, but instead of keeping
8874 a cumulative number, it maintains the rate at which the counter is
8875 incremented. Most of the time it will be used to measure the frequency of
8876 occurrence of certain events (e.g. requests to a specific URL).
8877
Willy Tarreauc9705a12010-07-27 20:05:50 +02008878 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
8879 the absolute number of connections received from clients which matched
8880 this entry. It does not mean the connections were accepted, just that
8881 they were received.
8882
8883 - conn_cur : Current Connections. It is a positive 32-bit integer which
8884 stores the concurrent connection counts for the entry. It is incremented
8885 once an incoming connection matches the entry, and decremented once the
8886 connection leaves. That way it is possible to know at any time the exact
8887 number of concurrent connections for an entry.
8888
8889 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8890 integer parameter <period> which indicates in milliseconds the length
8891 of the period over which the average is measured. It reports the average
8892 incoming connection rate over that period, in connections per period. The
8893 result is an integer which can be matched using ACLs.
8894
8895 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
8896 the absolute number of sessions received from clients which matched this
8897 entry. A session is a connection that was accepted by the layer 4 rules.
8898
8899 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8900 integer parameter <period> which indicates in milliseconds the length
8901 of the period over which the average is measured. It reports the average
8902 incoming session rate over that period, in sessions per period. The
8903 result is an integer which can be matched using ACLs.
8904
8905 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
8906 counts the absolute number of HTTP requests received from clients which
8907 matched this entry. It does not matter whether they are valid requests or
8908 not. Note that this is different from sessions when keep-alive is used on
8909 the client side.
8910
8911 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8912 integer parameter <period> which indicates in milliseconds the length
8913 of the period over which the average is measured. It reports the average
8914 HTTP request rate over that period, in requests per period. The result is
8915 an integer which can be matched using ACLs. It does not matter whether
8916 they are valid requests or not. Note that this is different from sessions
8917 when keep-alive is used on the client side.
8918
8919 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
8920 counts the absolute number of HTTP requests errors induced by clients
8921 which matched this entry. Errors are counted on invalid and truncated
8922 requests, as well as on denied or tarpitted requests, and on failed
8923 authentications. If the server responds with 4xx, then the request is
8924 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01008925 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02008926
8927 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8928 integer parameter <period> which indicates in milliseconds the length
8929 of the period over which the average is measured. It reports the average
8930 HTTP request error rate over that period, in requests per period (see
8931 http_err_cnt above for what is accounted as an error). The result is an
8932 integer which can be matched using ACLs.
8933
8934 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008935 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02008936 which matched this entry. Headers are included in the count. This may be
8937 used to limit abuse of upload features on photo or video servers.
8938
8939 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8940 integer parameter <period> which indicates in milliseconds the length
8941 of the period over which the average is measured. It reports the average
8942 incoming bytes rate over that period, in bytes per period. It may be used
8943 to detect users which upload too much and too fast. Warning: with large
8944 uploads, it is possible that the amount of uploaded data will be counted
8945 once upon termination, thus causing spikes in the average transfer speed
8946 instead of having a smooth one. This may partially be smoothed with
8947 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
8948 recommended for better fairness.
8949
8950 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008951 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02008952 matched this entry. Headers are included in the count. This may be used
8953 to limit abuse of bots sucking the whole site.
8954
8955 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
8956 an integer parameter <period> which indicates in milliseconds the length
8957 of the period over which the average is measured. It reports the average
8958 outgoing bytes rate over that period, in bytes per period. It may be used
8959 to detect users which download too much and too fast. Warning: with large
8960 transfers, it is possible that the amount of transferred data will be
8961 counted once upon termination, thus causing spikes in the average
8962 transfer speed instead of having a smooth one. This may partially be
8963 smoothed with "option contstats" though this is not perfect yet. Use of
8964 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02008965
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008966 There is only one stick-table per proxy. At the moment of writing this doc,
8967 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008968 to be required, simply create a dummy backend with a stick-table in it and
8969 reference it.
8970
8971 It is important to understand that stickiness based on learning information
8972 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01008973 lost upon restart unless peers are properly configured to transfer such
8974 information upon restart (recommended). In general it can be good as a
8975 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008976
Willy Tarreauc9705a12010-07-27 20:05:50 +02008977 Last, memory requirements may be important when storing many data types.
8978 Indeed, storing all indicators above at once in each entry requires 116 bytes
8979 per entry, or 116 MB for a 1-million entries table. This is definitely not
8980 something that can be ignored.
8981
8982 Example:
8983 # Keep track of counters of up to 1 million IP addresses over 5 minutes
8984 # and store a general purpose counter and the average connection rate
8985 # computed over a sliding window of 30 seconds.
8986 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
8987
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008988 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01008989 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008990
8991
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008992stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01008993 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008994 May be used in sections : defaults | frontend | listen | backend
8995 no | no | yes | yes
8996
8997 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008998 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008999 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009000 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009001 server is selected.
9002
9003 <table> is an optional stickiness table name. If unspecified, the same
9004 backend's table is used. A stickiness table is declared using
9005 the "stick-table" statement.
9006
9007 <cond> is an optional storage condition. It makes it possible to store
9008 certain criteria only when some conditions are met (or not met).
9009 For instance, it could be used to store the SSL session ID only
9010 when the response is a SSL server hello.
9011
9012 Some protocols or applications require complex stickiness rules and cannot
9013 always simply rely on cookies nor hashing. The "stick store-response"
9014 statement describes a rule to decide what to extract from the response and
9015 when to do it, in order to store it into a stickiness table for further
9016 requests to match it using the "stick match" statement. Obviously the
9017 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009018 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009019 See section 7 for a complete list of possible patterns and transformation
9020 rules.
9021
9022 The table has to be declared using the "stick-table" statement. It must be of
9023 a type compatible with the pattern. By default it is the one which is present
9024 in the same backend. It is possible to share a table with other backends by
9025 referencing it using the "table" keyword. If another table is referenced,
9026 the server's ID inside the backends are used. By default, all server IDs
9027 start at 1 in each backend, so the server ordering is enough. But in case of
9028 doubt, it is highly recommended to force server IDs using their "id" setting.
9029
9030 It is possible to restrict the conditions where a "stick store-response"
9031 statement will apply, using "if" or "unless" followed by a condition. This
9032 condition will be evaluated while parsing the response, so any criteria can
9033 be used. See section 7 for ACL based conditions.
9034
9035 There is no limit on the number of "stick store-response" statements, but
9036 there is a limit of 8 simultaneous stores per request or response. This
9037 makes it possible to store up to 8 criteria, all extracted from either the
9038 request or the response, regardless of the number of rules. Only the 8 first
9039 ones which match will be kept. Using this, it is possible to feed multiple
9040 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009041 another protocol or access method. Using multiple store-response rules with
9042 the same table is possible and may be used to find the best criterion to rely
9043 on, by arranging the rules by decreasing preference order. Only the first
9044 extracted criterion for a given table will be stored. All subsequent store-
9045 response rules referencing the same table will be skipped and their ACLs will
9046 not be evaluated. However, even if a store-request rule references a table, a
9047 store-response rule may also use the same table. This means that each table
9048 may learn exactly one element from the request and one element from the
9049 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009050
9051 The table will contain the real server that processed the request.
9052
9053 Example :
9054 # Learn SSL session ID from both request and response and create affinity.
9055 backend https
9056 mode tcp
9057 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009058 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009059 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009060
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009061 acl clienthello req_ssl_hello_type 1
9062 acl serverhello rep_ssl_hello_type 2
9063
9064 # use tcp content accepts to detects ssl client and server hello.
9065 tcp-request inspect-delay 5s
9066 tcp-request content accept if clienthello
9067
9068 # no timeout on response inspect delay by default.
9069 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009070
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009071 # SSL session ID (SSLID) may be present on a client or server hello.
9072 # Its length is coded on 1 byte at offset 43 and its value starts
9073 # at offset 44.
9074
9075 # Match and learn on request if client hello.
9076 stick on payload_lv(43,1) if clienthello
9077
9078 # Learn on response if server hello.
9079 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009080
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009081 server s1 192.168.1.1:443
9082 server s2 192.168.1.1:443
9083
9084 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9085 extraction.
9086
9087
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009088tcp-check connect [params*]
9089 Opens a new connection
9090 May be used in sections: defaults | frontend | listen | backend
9091 no | no | yes | yes
9092
9093 When an application lies on more than a single TCP port or when HAProxy
9094 load-balance many services in a single backend, it makes sense to probe all
9095 the services individually before considering a server as operational.
9096
9097 When there are no TCP port configured on the server line neither server port
9098 directive, then the 'tcp-check connect port <port>' must be the first step
9099 of the sequence.
9100
9101 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9102 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9103 do.
9104
9105 Parameters :
9106 They are optional and can be used to describe how HAProxy should open and
9107 use the TCP connection.
9108
9109 port if not set, check port or server port is used.
9110 It tells HAProxy where to open the connection to.
9111 <port> must be a valid TCP port source integer, from 1 to 65535.
9112
9113 send-proxy send a PROXY protocol string
9114
9115 ssl opens a ciphered connection
9116
9117 Examples:
9118 # check HTTP and HTTPs services on a server.
9119 # first open port 80 thanks to server line port directive, then
9120 # tcp-check opens port 443, ciphered and run a request on it:
9121 option tcp-check
9122 tcp-check connect
9123 tcp-check send GET\ /\ HTTP/1.0\r\n
9124 tcp-check send Host:\ haproxy.1wt.eu\r\n
9125 tcp-check send \r\n
9126 tcp-check expect rstring (2..|3..)
9127 tcp-check connect port 443 ssl
9128 tcp-check send GET\ /\ HTTP/1.0\r\n
9129 tcp-check send Host:\ haproxy.1wt.eu\r\n
9130 tcp-check send \r\n
9131 tcp-check expect rstring (2..|3..)
9132 server www 10.0.0.1 check port 80
9133
9134 # check both POP and IMAP from a single server:
9135 option tcp-check
9136 tcp-check connect port 110
9137 tcp-check expect string +OK\ POP3\ ready
9138 tcp-check connect port 143
9139 tcp-check expect string *\ OK\ IMAP4\ ready
9140 server mail 10.0.0.1 check
9141
9142 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9143
9144
9145tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009146 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009147 May be used in sections: defaults | frontend | listen | backend
9148 no | no | yes | yes
9149
9150 Arguments :
9151 <match> is a keyword indicating how to look for a specific pattern in the
9152 response. The keyword may be one of "string", "rstring" or
9153 binary.
9154 The keyword may be preceded by an exclamation mark ("!") to negate
9155 the match. Spaces are allowed between the exclamation mark and the
9156 keyword. See below for more details on the supported keywords.
9157
9158 <pattern> is the pattern to look for. It may be a string or a regular
9159 expression. If the pattern contains spaces, they must be escaped
9160 with the usual backslash ('\').
9161 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009162 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009163 two digits will represent a byte. The hexadecimal digits may be
9164 used upper or lower case.
9165
9166
9167 The available matches are intentionally similar to their http-check cousins :
9168
9169 string <string> : test the exact string matches in the response buffer.
9170 A health check response will be considered valid if the
9171 response's buffer contains this exact string. If the
9172 "string" keyword is prefixed with "!", then the response
9173 will be considered invalid if the body contains this
9174 string. This can be used to look for a mandatory pattern
9175 in a protocol response, or to detect a failure when a
9176 specific error appears in a protocol banner.
9177
9178 rstring <regex> : test a regular expression on the response buffer.
9179 A health check response will be considered valid if the
9180 response's buffer matches this expression. If the
9181 "rstring" keyword is prefixed with "!", then the response
9182 will be considered invalid if the body matches the
9183 expression.
9184
9185 binary <hexstring> : test the exact string in its hexadecimal form matches
9186 in the response buffer. A health check response will
9187 be considered valid if the response's buffer contains
9188 this exact hexadecimal string.
9189 Purpose is to match data on binary protocols.
9190
9191 It is important to note that the responses will be limited to a certain size
9192 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9193 Thus, too large responses may not contain the mandatory pattern when using
9194 "string", "rstring" or binary. If a large response is absolutely required, it
9195 is possible to change the default max size by setting the global variable.
9196 However, it is worth keeping in mind that parsing very large responses can
9197 waste some CPU cycles, especially when regular expressions are used, and that
9198 it is always better to focus the checks on smaller resources. Also, in its
9199 current state, the check will not find any string nor regex past a null
9200 character in the response. Similarly it is not possible to request matching
9201 the null character.
9202
9203 Examples :
9204 # perform a POP check
9205 option tcp-check
9206 tcp-check expect string +OK\ POP3\ ready
9207
9208 # perform an IMAP check
9209 option tcp-check
9210 tcp-check expect string *\ OK\ IMAP4\ ready
9211
9212 # look for the redis master server
9213 option tcp-check
9214 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009215 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009216 tcp-check send info\ replication\r\n
9217 tcp-check expect string role:master
9218 tcp-check send QUIT\r\n
9219 tcp-check expect string +OK
9220
9221
9222 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9223 "tcp-check send-binary", "http-check expect", tune.chksize
9224
9225
9226tcp-check send <data>
9227 Specify a string to be sent as a question during a generic health check
9228 May be used in sections: defaults | frontend | listen | backend
9229 no | no | yes | yes
9230
9231 <data> : the data to be sent as a question during a generic health check
9232 session. For now, <data> must be a string.
9233
9234 Examples :
9235 # look for the redis master server
9236 option tcp-check
9237 tcp-check send info\ replication\r\n
9238 tcp-check expect string role:master
9239
9240 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9241 "tcp-check send-binary", tune.chksize
9242
9243
Davor Ocelice9ed2812017-12-25 17:49:28 +01009244tcp-check send-binary <hexstring>
9245 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009246 tcp health check
9247 May be used in sections: defaults | frontend | listen | backend
9248 no | no | yes | yes
9249
9250 <data> : the data to be sent as a question during a generic health check
9251 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009252 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009253 response buffer. A health check response will be considered
9254 valid if the response's buffer contains this exact
9255 hexadecimal string.
9256 Purpose is to send binary data to ask on binary protocols.
9257
9258 Examples :
9259 # redis check in binary
9260 option tcp-check
9261 tcp-check send-binary 50494e470d0a # PING\r\n
9262 tcp-check expect binary 2b504F4e47 # +PONG
9263
9264
9265 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9266 "tcp-check send", tune.chksize
9267
9268
Willy Tarreaue9656522010-08-17 15:40:09 +02009269tcp-request connection <action> [{if | unless} <condition>]
9270 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009271 May be used in sections : defaults | frontend | listen | backend
9272 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009273 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009274 <action> defines the action to perform if the condition applies. See
9275 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009276
Willy Tarreaue9656522010-08-17 15:40:09 +02009277 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009278
9279 Immediately after acceptance of a new incoming connection, it is possible to
9280 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009281 or dropped or have its counters tracked. Those conditions cannot make use of
9282 any data contents because the connection has not been read from yet, and the
9283 buffers are not yet allocated. This is used to selectively and very quickly
9284 accept or drop connections from various sources with a very low overhead. If
9285 some contents need to be inspected in order to take the decision, the
9286 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009287
Willy Tarreaue9656522010-08-17 15:40:09 +02009288 The "tcp-request connection" rules are evaluated in their exact declaration
9289 order. If no rule matches or if there is no rule, the default action is to
9290 accept the incoming connection. There is no specific limit to the number of
9291 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009292
Willy Tarreaua9083d02015-05-08 15:27:59 +02009293 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009294 - accept :
9295 accepts the connection if the condition is true (when used with "if")
9296 or false (when used with "unless"). The first such rule executed ends
9297 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009298
Willy Tarreaue9656522010-08-17 15:40:09 +02009299 - reject :
9300 rejects the connection if the condition is true (when used with "if")
9301 or false (when used with "unless"). The first such rule executed ends
9302 the rules evaluation. Rejected connections do not even become a
9303 session, which is why they are accounted separately for in the stats,
9304 as "denied connections". They are not considered for the session
9305 rate-limit and are not logged either. The reason is that these rules
9306 should only be used to filter extremely high connection rates such as
9307 the ones encountered during a massive DDoS attack. Under these extreme
9308 conditions, the simple action of logging each event would make the
9309 system collapse and would considerably lower the filtering capacity. If
9310 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009311 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009312
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009313 - expect-proxy layer4 :
9314 configures the client-facing connection to receive a PROXY protocol
9315 header before any byte is read from the socket. This is equivalent to
9316 having the "accept-proxy" keyword on the "bind" line, except that using
9317 the TCP rule allows the PROXY protocol to be accepted only for certain
9318 IP address ranges using an ACL. This is convenient when multiple layers
9319 of load balancers are passed through by traffic coming from public
9320 hosts.
9321
Bertrand Jacquin90759682016-06-06 15:35:39 +01009322 - expect-netscaler-cip layer4 :
9323 configures the client-facing connection to receive a NetScaler Client
9324 IP insertion protocol header before any byte is read from the socket.
9325 This is equivalent to having the "accept-netscaler-cip" keyword on the
9326 "bind" line, except that using the TCP rule allows the PROXY protocol
9327 to be accepted only for certain IP address ranges using an ACL. This
9328 is convenient when multiple layers of load balancers are passed
9329 through by traffic coming from public hosts.
9330
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009331 - capture <sample> len <length> :
9332 This only applies to "tcp-request content" rules. It captures sample
9333 expression <sample> from the request buffer, and converts it to a
9334 string of at most <len> characters. The resulting string is stored into
9335 the next request "capture" slot, so it will possibly appear next to
9336 some captured HTTP headers. It will then automatically appear in the
9337 logs, and it will be possible to extract it using sample fetch rules to
9338 feed it into headers or anything. The length should be limited given
9339 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009340 session life. Please check section 7.3 (Fetching samples) and "capture
9341 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009342
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009343 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009344 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009345 rules do not stop evaluation and do not change default action. The
9346 number of counters that may be simultaneously tracked by the same
9347 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -05009348 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009349 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9350 tracking of the counters of the specified table as the first set. The
9351 first "track-sc1" rule executed enables tracking of the counters of the
9352 specified table as the second set. The first "track-sc2" rule executed
9353 enables tracking of the counters of the specified table as the third
9354 set. It is a recommended practice to use the first set of counters for
9355 the per-frontend counters and the second set for the per-backend ones.
9356 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009357
Willy Tarreaue9656522010-08-17 15:40:09 +02009358 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009359 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009360 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009361 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009362 and used to select which table entry to update the counters.
9363 Note that "tcp-request connection" cannot use content-based
9364 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009365
Willy Tarreaue9656522010-08-17 15:40:09 +02009366 <table> is an optional table to be used instead of the default one,
9367 which is the stick-table declared in the current proxy. All
9368 the counters for the matches and updates for the key will
9369 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009370
Willy Tarreaue9656522010-08-17 15:40:09 +02009371 Once a "track-sc*" rule is executed, the key is looked up in the table
9372 and if it is not found, an entry is allocated for it. Then a pointer to
9373 that entry is kept during all the session's life, and this entry's
9374 counters are updated as often as possible, every time the session's
9375 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009376 Counters are only updated for events that happen after the tracking has
9377 been started. For example, connection counters will not be updated when
9378 tracking layer 7 information, since the connection event happens before
9379 layer7 information is extracted.
9380
Willy Tarreaue9656522010-08-17 15:40:09 +02009381 If the entry tracks concurrent connection counters, one connection is
9382 counted for as long as the entry is tracked, and the entry will not
9383 expire during that time. Tracking counters also provides a performance
9384 advantage over just checking the keys, because only one table lookup is
9385 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009386
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009387 - sc-inc-gpc0(<sc-id>):
9388 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9389 counter designated by <sc-id>. If an error occurs, this action silently
9390 fails and the actions evaluation continues.
9391
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009392 - sc-inc-gpc1(<sc-id>):
9393 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9394 counter designated by <sc-id>. If an error occurs, this action silently
9395 fails and the actions evaluation continues.
9396
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009397 - sc-set-gpt0(<sc-id>) <int>:
9398 This action sets the GPT0 tag according to the sticky counter designated
9399 by <sc-id> and the value of <int>. The expected result is a boolean. If
9400 an error occurs, this action silently fails and the actions evaluation
9401 continues.
9402
William Lallemand2e785f22016-05-25 01:48:42 +02009403 - set-src <expr> :
9404 Is used to set the source IP address to the value of specified
9405 expression. Useful if you want to mask source IP for privacy.
9406 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009407 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +02009408
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009409 Arguments:
9410 <expr> Is a standard HAProxy expression formed by a sample-fetch
9411 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +02009412
9413 Example:
William Lallemand2e785f22016-05-25 01:48:42 +02009414 tcp-request connection set-src src,ipmask(24)
9415
Willy Tarreau0c630532016-10-21 17:52:58 +02009416 When possible, set-src preserves the original source port as long as the
9417 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009418
William Lallemand44be6402016-05-25 01:51:35 +02009419 - set-src-port <expr> :
9420 Is used to set the source port address to the value of specified
9421 expression.
9422
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009423 Arguments:
9424 <expr> Is a standard HAProxy expression formed by a sample-fetch
9425 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +02009426
9427 Example:
William Lallemand44be6402016-05-25 01:51:35 +02009428 tcp-request connection set-src-port int(4000)
9429
Willy Tarreau0c630532016-10-21 17:52:58 +02009430 When possible, set-src-port preserves the original source address as long
9431 as the address family supports a port, otherwise it forces the source
9432 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009433
William Lallemand13e9b0c2016-05-25 02:34:07 +02009434 - set-dst <expr> :
9435 Is used to set the destination IP address to the value of specified
9436 expression. Useful if you want to mask IP for privacy in log.
9437 If you want to provide an IP from a HTTP header use "http-request
9438 set-dst". If you want to connect to the new address/port, use
9439 '0.0.0.0:0' as a server address in the backend.
9440
9441 <expr> Is a standard HAProxy expression formed by a sample-fetch
9442 followed by some converters.
9443
9444 Example:
9445
9446 tcp-request connection set-dst dst,ipmask(24)
9447 tcp-request connection set-dst ipv4(10.0.0.1)
9448
Willy Tarreau0c630532016-10-21 17:52:58 +02009449 When possible, set-dst preserves the original destination port as long as
9450 the address family allows it, otherwise the destination port is set to 0.
9451
William Lallemand13e9b0c2016-05-25 02:34:07 +02009452 - set-dst-port <expr> :
9453 Is used to set the destination port address to the value of specified
9454 expression. If you want to connect to the new address/port, use
9455 '0.0.0.0:0' as a server address in the backend.
9456
9457
9458 <expr> Is a standard HAProxy expression formed by a sample-fetch
9459 followed by some converters.
9460
9461 Example:
9462
9463 tcp-request connection set-dst-port int(4000)
9464
Willy Tarreau0c630532016-10-21 17:52:58 +02009465 When possible, set-dst-port preserves the original destination address as
9466 long as the address family supports a port, otherwise it forces the
9467 destination address to IPv4 "0.0.0.0" before rewriting the port.
9468
Willy Tarreau2d392c22015-08-24 01:43:45 +02009469 - "silent-drop" :
9470 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009471 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009472 to prevent the client from being notified. The effect it then that the
9473 client still sees an established connection while there's none on
9474 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9475 except that it doesn't use any local resource at all on the machine
9476 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009477 slow down stronger attackers. It is important to understand the impact
9478 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009479 client and HAProxy (firewalls, proxies, load balancers) will also keep
9480 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009481 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009482 TCP_REPAIR socket option is used to block the emission of a TCP
9483 reset. On other systems, the socket's TTL is reduced to 1 so that the
9484 TCP reset doesn't pass the first router, though it's still delivered to
9485 local networks. Do not use it unless you fully understand how it works.
9486
Willy Tarreaue9656522010-08-17 15:40:09 +02009487 Note that the "if/unless" condition is optional. If no condition is set on
9488 the action, it is simply performed unconditionally. That can be useful for
9489 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009490
Willy Tarreaue9656522010-08-17 15:40:09 +02009491 Example: accept all connections from white-listed hosts, reject too fast
9492 connection without counting them, and track accepted connections.
9493 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009494
Willy Tarreaue9656522010-08-17 15:40:09 +02009495 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009496 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009497 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009498
Willy Tarreaue9656522010-08-17 15:40:09 +02009499 Example: accept all connections from white-listed hosts, count all other
9500 connections and reject too fast ones. This results in abusive ones
9501 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009502
Willy Tarreaue9656522010-08-17 15:40:09 +02009503 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009504 tcp-request connection track-sc0 src
9505 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009506
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009507 Example: enable the PROXY protocol for traffic coming from all known proxies.
9508
9509 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9510
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009511 See section 7 about ACL usage.
9512
Willy Tarreau4f614292016-10-21 17:49:36 +02009513 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009514
9515
Willy Tarreaue9656522010-08-17 15:40:09 +02009516tcp-request content <action> [{if | unless} <condition>]
9517 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009518 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009519 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009520 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009521 <action> defines the action to perform if the condition applies. See
9522 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009523
Willy Tarreaue9656522010-08-17 15:40:09 +02009524 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009525
Davor Ocelice9ed2812017-12-25 17:49:28 +01009526 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009527 called "TCP content inspection". During this stage, ACL-based rules are
9528 evaluated every time the request contents are updated, until either an
9529 "accept" or a "reject" rule matches, or the TCP request inspection delay
9530 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009531
Willy Tarreaue9656522010-08-17 15:40:09 +02009532 The first difference between these rules and "tcp-request connection" rules
9533 is that "tcp-request content" rules can make use of contents to take a
9534 decision. Most often, these decisions will consider a protocol recognition or
9535 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009536 both frontends and backends. In case of HTTP keep-alive with the client, all
9537 tcp-request content rules are evaluated again, so haproxy keeps a record of
9538 what sticky counters were assigned by a "tcp-request connection" versus a
9539 "tcp-request content" rule, and flushes all the content-related ones after
9540 processing an HTTP request, so that they may be evaluated again by the rules
9541 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009542 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009543 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009544
Willy Tarreaue9656522010-08-17 15:40:09 +02009545 Content-based rules are evaluated in their exact declaration order. If no
9546 rule matches or if there is no rule, the default action is to accept the
9547 contents. There is no specific limit to the number of rules which may be
9548 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009549
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009550 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009551 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +01009552 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009553 - reject : the request is rejected and the connection is closed
9554 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009555 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009556 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009557 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009558 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009559 - sc-set-gpt0(<sc-id>) <int>
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +02009560 - set-dst <expr>
9561 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009562 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009563 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009564 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009565 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009566
Willy Tarreaue9656522010-08-17 15:40:09 +02009567 They have the same meaning as their counter-parts in "tcp-request connection"
9568 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +01009569 For "do-resolve" action, please check the "http-request do-resolve"
9570 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +02009571
Willy Tarreauf3338342014-01-28 21:40:28 +01009572 While there is nothing mandatory about it, it is recommended to use the
9573 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9574 content" rules in the frontend, and track-sc2 for "tcp-request content"
9575 rules in the backend, because that makes the configuration more readable
9576 and easier to troubleshoot, but this is just a guideline and all counters
9577 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009578
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009579 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009580 the action, it is simply performed unconditionally. That can be useful for
9581 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009582
Willy Tarreaue9656522010-08-17 15:40:09 +02009583 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009584 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9585 contents of a buffer before extracting the required data. If the buffered
9586 contents do not parse as a valid HTTP message, then the ACL does not match.
9587 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009588 processing, so there is no risk of parsing something differently. In an HTTP
9589 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9590 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009591
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009592 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009593 are present when the rule is processed. The rule processing engine is able to
9594 wait until the inspect delay expires when the data to be tracked is not yet
9595 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009596
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +02009597 The "set-dst" and "set-dst-port" are used to set respectively the destination
9598 IP and port. More information on how to use it at "http-request set-dst".
9599
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009600 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009601 declared inline. For "tcp-request session" rules, only session-level
9602 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009603
Daniel Schneller0b547052016-03-21 20:46:57 +01009604 <var-name> The name of the variable starts with an indication about
9605 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009606 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009607 "sess" : the variable is shared with the whole session
9608 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009609 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009610 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009611 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009612 "res" : the variable is shared only during response
9613 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009614 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009615 The name may only contain characters 'a-z', 'A-Z', '0-9',
9616 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009617
9618 <expr> Is a standard HAProxy expression formed by a sample-fetch
9619 followed by some converters.
9620
Christopher Faulet85d79c92016-11-09 16:54:56 +01009621 The "unset-var" is used to unset a variable. See above for details about
9622 <var-name>.
9623
Patrick Hemmer268a7072018-05-11 12:52:31 -04009624 The "set-priority-class" is used to set the queue priority class of the
9625 current request. The value must be a sample expression which converts to an
9626 integer in the range -2047..2047. Results outside this range will be
9627 truncated. The priority class determines the order in which queued requests
9628 are processed. Lower values have higher priority.
9629
9630 The "set-priority-offset" is used to set the queue priority timestamp offset
9631 of the current request. The value must be a sample expression which converts
9632 to an integer in the range -524287..524287. Results outside this range will be
9633 truncated. When a request is queued, it is ordered first by the priority
9634 class, then by the current timestamp adjusted by the given offset in
9635 milliseconds. Lower values have higher priority.
9636 Note that the resulting timestamp is is only tracked with enough precision for
9637 524,287ms (8m44s287ms). If the request is queued long enough to where the
9638 adjusted timestamp exceeds this value, it will be misidentified as highest
9639 priority. Thus it is important to set "timeout queue" to a value, where when
9640 combined with the offset, does not exceed this limit.
9641
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009642 The "send-spoe-group" is used to trigger sending of a group of SPOE
9643 messages. To do so, the SPOE engine used to send messages must be defined, as
9644 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9645 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9646 the SPOE agent name must be used.
9647
9648 <engine-name> The SPOE engine name.
9649
9650 <group-name> The SPOE group name as specified in the engine configuration.
9651
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009652 Example:
9653
9654 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009655 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009656
Willy Tarreau62644772008-07-16 18:36:06 +02009657 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009658 # Accept HTTP requests containing a Host header saying "example.com"
9659 # and reject everything else.
9660 acl is_host_com hdr(Host) -i example.com
9661 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009662 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009663 tcp-request content reject
9664
9665 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009666 # reject SMTP connection if client speaks first
9667 tcp-request inspect-delay 30s
9668 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009669 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009670
9671 # Forward HTTPS connection only if client speaks
9672 tcp-request inspect-delay 30s
9673 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009674 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009675 tcp-request content reject
9676
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009677 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009678 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009679 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009680 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009681 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9682 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009683
9684 Example:
9685 # track request counts per "base" (concatenation of Host+URL)
9686 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009687 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009688
Willy Tarreaue9656522010-08-17 15:40:09 +02009689 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009690 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009691
9692 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009693 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009694 # protecting all our sites
9695 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009696 tcp-request connection track-sc0 src
9697 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009698 ...
9699 use_backend http_dynamic if { path_end .php }
9700
9701 backend http_dynamic
9702 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009703 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009704 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009705 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009706 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009707 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009708 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009709
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009710 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009711
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009712 See also : "tcp-request connection", "tcp-request session",
9713 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009714
9715
9716tcp-request inspect-delay <timeout>
9717 Set the maximum allowed time to wait for data during content inspection
9718 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009719 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009720 Arguments :
9721 <timeout> is the timeout value specified in milliseconds by default, but
9722 can be in any other unit if the number is suffixed by the unit,
9723 as explained at the top of this document.
9724
9725 People using haproxy primarily as a TCP relay are often worried about the
9726 risk of passing any type of protocol to a server without any analysis. In
9727 order to be able to analyze the request contents, we must first withhold
9728 the data then analyze them. This statement simply enables withholding of
9729 data for at most the specified amount of time.
9730
Willy Tarreaufb356202010-08-03 14:02:05 +02009731 TCP content inspection applies very early when a connection reaches a
9732 frontend, then very early when the connection is forwarded to a backend. This
9733 means that a connection may experience a first delay in the frontend and a
9734 second delay in the backend if both have tcp-request rules.
9735
Willy Tarreau62644772008-07-16 18:36:06 +02009736 Note that when performing content inspection, haproxy will evaluate the whole
9737 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009738 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009739 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009740 contents are definitive. If no delay is set, haproxy will not wait at all
9741 and will immediately apply a verdict based on the available information.
9742 Obviously this is unlikely to be very useful and might even be racy, so such
9743 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009744
9745 As soon as a rule matches, the request is released and continues as usual. If
9746 the timeout is reached and no rule matches, the default policy will be to let
9747 it pass through unaffected.
9748
9749 For most protocols, it is enough to set it to a few seconds, as most clients
9750 send the full request immediately upon connection. Add 3 or more seconds to
9751 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009752 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009753 before the server (e.g. SMTP), or to wait for a client to talk before passing
9754 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009755 least the inspection delay, otherwise it will expire first. If the client
9756 closes the connection or if the buffer is full, the delay immediately expires
9757 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009758
Willy Tarreau55165fe2009-05-10 12:02:55 +02009759 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009760 "timeout client".
9761
9762
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009763tcp-response content <action> [{if | unless} <condition>]
9764 Perform an action on a session response depending on a layer 4-7 condition
9765 May be used in sections : defaults | frontend | listen | backend
9766 no | no | yes | yes
9767 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009768 <action> defines the action to perform if the condition applies. See
9769 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009770
9771 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9772
Davor Ocelice9ed2812017-12-25 17:49:28 +01009773 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009774 called "TCP content inspection". During this stage, ACL-based rules are
9775 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009776 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9777 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009778
9779 Most often, these decisions will consider a protocol recognition or validity.
9780
9781 Content-based rules are evaluated in their exact declaration order. If no
9782 rule matches or if there is no rule, the default action is to accept the
9783 contents. There is no specific limit to the number of rules which may be
9784 inserted.
9785
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009786 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009787 - accept :
9788 accepts the response if the condition is true (when used with "if")
9789 or false (when used with "unless"). The first such rule executed ends
9790 the rules evaluation.
9791
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009792 - close :
9793 immediately closes the connection with the server if the condition is
9794 true (when used with "if"), or false (when used with "unless"). The
9795 first such rule executed ends the rules evaluation. The main purpose of
9796 this action is to force a connection to be finished between a client
9797 and a server after an exchange when the application protocol expects
9798 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009799 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009800 protocols.
9801
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009802 - reject :
9803 rejects the response if the condition is true (when used with "if")
9804 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009805 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009806
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009807 - set-var(<var-name>) <expr>
9808 Sets a variable.
9809
Christopher Faulet85d79c92016-11-09 16:54:56 +01009810 - unset-var(<var-name>)
9811 Unsets a variable.
9812
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009813 - sc-inc-gpc0(<sc-id>):
9814 This action increments the GPC0 counter according to the sticky
9815 counter designated by <sc-id>. If an error occurs, this action fails
9816 silently and the actions evaluation continues.
9817
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009818 - sc-inc-gpc1(<sc-id>):
9819 This action increments the GPC1 counter according to the sticky
9820 counter designated by <sc-id>. If an error occurs, this action fails
9821 silently and the actions evaluation continues.
9822
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009823 - sc-set-gpt0(<sc-id>) <int> :
9824 This action sets the GPT0 tag according to the sticky counter designated
9825 by <sc-id> and the value of <int>. The expected result is a boolean. If
9826 an error occurs, this action silently fails and the actions evaluation
9827 continues.
9828
Willy Tarreau2d392c22015-08-24 01:43:45 +02009829 - "silent-drop" :
9830 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009831 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009832 to prevent the client from being notified. The effect it then that the
9833 client still sees an established connection while there's none on
9834 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9835 except that it doesn't use any local resource at all on the machine
9836 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009837 slow down stronger attackers. It is important to understand the impact
9838 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009839 client and HAProxy (firewalls, proxies, load balancers) will also keep
9840 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009841 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009842 TCP_REPAIR socket option is used to block the emission of a TCP
9843 reset. On other systems, the socket's TTL is reduced to 1 so that the
9844 TCP reset doesn't pass the first router, though it's still delivered to
9845 local networks. Do not use it unless you fully understand how it works.
9846
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009847 - send-spoe-group <engine-name> <group-name>
9848 Send a group of SPOE messages.
9849
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009850 Note that the "if/unless" condition is optional. If no condition is set on
9851 the action, it is simply performed unconditionally. That can be useful for
9852 for changing the default action to a reject.
9853
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009854 It is perfectly possible to match layer 7 contents with "tcp-response
9855 content" rules, but then it is important to ensure that a full response has
9856 been buffered, otherwise no contents will match. In order to achieve this,
9857 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009858 period.
9859
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009860 The "set-var" is used to set the content of a variable. The variable is
9861 declared inline.
9862
Daniel Schneller0b547052016-03-21 20:46:57 +01009863 <var-name> The name of the variable starts with an indication about
9864 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009865 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009866 "sess" : the variable is shared with the whole session
9867 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009868 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009869 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009870 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009871 "res" : the variable is shared only during response
9872 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009873 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009874 The name may only contain characters 'a-z', 'A-Z', '0-9',
9875 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009876
9877 <expr> Is a standard HAProxy expression formed by a sample-fetch
9878 followed by some converters.
9879
9880 Example:
9881
9882 tcp-request content set-var(sess.my_var) src
9883
Christopher Faulet85d79c92016-11-09 16:54:56 +01009884 The "unset-var" is used to unset a variable. See above for details about
9885 <var-name>.
9886
9887 Example:
9888
9889 tcp-request content unset-var(sess.my_var)
9890
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009891 The "send-spoe-group" is used to trigger sending of a group of SPOE
9892 messages. To do so, the SPOE engine used to send messages must be defined, as
9893 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9894 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9895 the SPOE agent name must be used.
9896
9897 <engine-name> The SPOE engine name.
9898
9899 <group-name> The SPOE group name as specified in the engine configuration.
9900
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009901 See section 7 about ACL usage.
9902
9903 See also : "tcp-request content", "tcp-response inspect-delay"
9904
9905
Willy Tarreau4f614292016-10-21 17:49:36 +02009906tcp-request session <action> [{if | unless} <condition>]
9907 Perform an action on a validated session depending on a layer 5 condition
9908 May be used in sections : defaults | frontend | listen | backend
9909 no | yes | yes | no
9910 Arguments :
9911 <action> defines the action to perform if the condition applies. See
9912 below.
9913
9914 <condition> is a standard layer5-only ACL-based condition (see section 7).
9915
Davor Ocelice9ed2812017-12-25 17:49:28 +01009916 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +02009917 it is possible to evaluate some conditions to decide whether this session
9918 must be accepted or dropped or have its counters tracked. Those conditions
9919 cannot make use of any data contents because no buffers are allocated yet and
9920 the processing cannot wait at this stage. The main use case it to copy some
9921 early information into variables (since variables are accessible in the
9922 session), or to keep track of some information collected after the handshake,
9923 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +01009924 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +02009925 extracted information can thus be copied to a variable or tracked using
9926 "track-sc" rules. Of course it is also possible to decide to accept/reject as
9927 with other rulesets. Most operations performed here could also be performed
9928 in "tcp-request content" rules, except that in HTTP these rules are evaluated
9929 for each new request, and that might not always be acceptable. For example a
9930 rule might increment a counter on each evaluation. It would also be possible
9931 that a country is resolved by geolocation from the source IP address,
9932 assigned to a session-wide variable, then the source address rewritten from
9933 an HTTP header for all requests. If some contents need to be inspected in
9934 order to take the decision, the "tcp-request content" statements must be used
9935 instead.
9936
9937 The "tcp-request session" rules are evaluated in their exact declaration
9938 order. If no rule matches or if there is no rule, the default action is to
9939 accept the incoming session. There is no specific limit to the number of
9940 rules which may be inserted.
9941
9942 Several types of actions are supported :
9943 - accept : the request is accepted
9944 - reject : the request is rejected and the connection is closed
9945 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
9946 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009947 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009948 - sc-set-gpt0(<sc-id>) <int>
9949 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009950 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009951 - silent-drop
9952
9953 These actions have the same meaning as their respective counter-parts in
9954 "tcp-request connection" and "tcp-request content", so please refer to these
9955 sections for a complete description.
9956
9957 Note that the "if/unless" condition is optional. If no condition is set on
9958 the action, it is simply performed unconditionally. That can be useful for
9959 "track-sc*" actions as well as for changing the default action to a reject.
9960
9961 Example: track the original source address by default, or the one advertised
9962 in the PROXY protocol header for connection coming from the local
9963 proxies. The first connection-level rule enables receipt of the
9964 PROXY protocol for these ones, the second rule tracks whatever
9965 address we decide to keep after optional decoding.
9966
9967 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9968 tcp-request session track-sc0 src
9969
9970 Example: accept all sessions from white-listed hosts, reject too fast
9971 sessions without counting them, and track accepted sessions.
9972 This results in session rate being capped from abusive sources.
9973
9974 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9975 tcp-request session reject if { src_sess_rate gt 10 }
9976 tcp-request session track-sc0 src
9977
9978 Example: accept all sessions from white-listed hosts, count all other
9979 sessions and reject too fast ones. This results in abusive ones
9980 being blocked as long as they don't slow down.
9981
9982 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9983 tcp-request session track-sc0 src
9984 tcp-request session reject if { sc0_sess_rate gt 10 }
9985
9986 See section 7 about ACL usage.
9987
9988 See also : "tcp-request connection", "tcp-request content", "stick-table"
9989
9990
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009991tcp-response inspect-delay <timeout>
9992 Set the maximum allowed time to wait for a response during content inspection
9993 May be used in sections : defaults | frontend | listen | backend
9994 no | no | yes | yes
9995 Arguments :
9996 <timeout> is the timeout value specified in milliseconds by default, but
9997 can be in any other unit if the number is suffixed by the unit,
9998 as explained at the top of this document.
9999
10000 See also : "tcp-response content", "tcp-request inspect-delay".
10001
10002
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010003timeout check <timeout>
10004 Set additional check timeout, but only after a connection has been already
10005 established.
10006
10007 May be used in sections: defaults | frontend | listen | backend
10008 yes | no | yes | yes
10009 Arguments:
10010 <timeout> is the timeout value specified in milliseconds by default, but
10011 can be in any other unit if the number is suffixed by the unit,
10012 as explained at the top of this document.
10013
10014 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10015 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010016 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010017 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010018 (Please also note that there is no valid reason to have such long connect
10019 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10020 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010021
10022 If "timeout check" is not set haproxy uses "inter" for complete check
10023 timeout (connect + read) exactly like all <1.3.15 version.
10024
10025 In most cases check request is much simpler and faster to handle than normal
10026 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010027 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010028
10029 This parameter is specific to backends, but can be specified once for all in
10030 "defaults" sections. This is in fact one of the easiest solutions not to
10031 forget about it.
10032
Willy Tarreau41a340d2008-01-22 12:25:31 +010010033 See also: "timeout connect", "timeout queue", "timeout server",
10034 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010035
10036
Willy Tarreau0ba27502007-12-24 16:55:16 +010010037timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010038 Set the maximum inactivity time on the client side.
10039 May be used in sections : defaults | frontend | listen | backend
10040 yes | yes | yes | no
10041 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010042 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010043 can be in any other unit if the number is suffixed by the unit,
10044 as explained at the top of this document.
10045
10046 The inactivity timeout applies when the client is expected to acknowledge or
10047 send data. In HTTP mode, this timeout is particularly important to consider
10048 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010049 response while it is reading data sent by the server. That said, for the
10050 first phase, it is preferable to set the "timeout http-request" to better
10051 protect HAProxy from Slowloris like attacks. The value is specified in
10052 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010053 suffixed by the unit, as specified at the top of this document. In TCP mode
10054 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10055 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010056 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010057 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010058 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10059 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010060 which overrides "timeout client" and "timeout server" for tunnels, as well as
10061 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010062
10063 This parameter is specific to frontends, but can be specified once for all in
10064 "defaults" sections. This is in fact one of the easiest solutions not to
10065 forget about it. An unspecified timeout results in an infinite timeout, which
10066 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010067 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010068 the system if the system's timeouts are not configured either.
10069
Willy Tarreau95c4e142017-11-26 12:18:55 +010010070 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010071
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010072 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010073
Willy Tarreau0ba27502007-12-24 16:55:16 +010010074
Willy Tarreau05cdd962014-05-10 14:30:07 +020010075timeout client-fin <timeout>
10076 Set the inactivity timeout on the client side for half-closed connections.
10077 May be used in sections : defaults | frontend | listen | backend
10078 yes | yes | yes | no
10079 Arguments :
10080 <timeout> is the timeout value specified in milliseconds by default, but
10081 can be in any other unit if the number is suffixed by the unit,
10082 as explained at the top of this document.
10083
10084 The inactivity timeout applies when the client is expected to acknowledge or
10085 send data while one direction is already shut down. This timeout is different
10086 from "timeout client" in that it only applies to connections which are closed
10087 in one direction. This is particularly useful to avoid keeping connections in
10088 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10089 problem is particularly common long connections such as RDP or WebSocket.
10090 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010091 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10092 frame was sent, often indicating an expectation that the connection quickly
10093 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010094
10095 This parameter is specific to frontends, but can be specified once for all in
10096 "defaults" sections. By default it is not set, so half-closed connections
10097 will use the other timeouts (timeout.client or timeout.tunnel).
10098
10099 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10100
10101
Willy Tarreau0ba27502007-12-24 16:55:16 +010010102timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010103 Set the maximum time to wait for a connection attempt to a server to succeed.
10104 May be used in sections : defaults | frontend | listen | backend
10105 yes | no | yes | yes
10106 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010107 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010108 can be in any other unit if the number is suffixed by the unit,
10109 as explained at the top of this document.
10110
10111 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010112 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010113 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010114 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010115 connect timeout also presets both queue and tarpit timeouts to the same value
10116 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010117
10118 This parameter is specific to backends, but can be specified once for all in
10119 "defaults" sections. This is in fact one of the easiest solutions not to
10120 forget about it. An unspecified timeout results in an infinite timeout, which
10121 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010122 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010123 the system if the system's timeouts are not configured either.
10124
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010125 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010126
Willy Tarreau0ba27502007-12-24 16:55:16 +010010127
Willy Tarreaub16a5742010-01-10 14:46:16 +010010128timeout http-keep-alive <timeout>
10129 Set the maximum allowed time to wait for a new HTTP request to appear
10130 May be used in sections : defaults | frontend | listen | backend
10131 yes | yes | yes | yes
10132 Arguments :
10133 <timeout> is the timeout value specified in milliseconds by default, but
10134 can be in any other unit if the number is suffixed by the unit,
10135 as explained at the top of this document.
10136
10137 By default, the time to wait for a new request in case of keep-alive is set
10138 by "timeout http-request". However this is not always convenient because some
10139 people want very short keep-alive timeouts in order to release connections
10140 faster, and others prefer to have larger ones but still have short timeouts
10141 once the request has started to present itself.
10142
10143 The "http-keep-alive" timeout covers these needs. It will define how long to
10144 wait for a new HTTP request to start coming after a response was sent. Once
10145 the first byte of request has been seen, the "http-request" timeout is used
10146 to wait for the complete request to come. Note that empty lines prior to a
10147 new request do not refresh the timeout and are not counted as a new request.
10148
10149 There is also another difference between the two timeouts : when a connection
10150 expires during timeout http-keep-alive, no error is returned, the connection
10151 just closes. If the connection expires in "http-request" while waiting for a
10152 connection to complete, a HTTP 408 error is returned.
10153
10154 In general it is optimal to set this value to a few tens to hundreds of
10155 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010156 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010157 1 millisecond) it will probably only accept pipelined requests but not the
10158 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010159 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010160
10161 If this parameter is not set, the "http-request" timeout applies, and if both
10162 are not set, "timeout client" still applies at the lower level. It should be
10163 set in the frontend to take effect, unless the frontend is in TCP mode, in
10164 which case the HTTP backend's timeout will be used.
10165
Willy Tarreau95c4e142017-11-26 12:18:55 +010010166 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10167 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010168 (where we only have one connection per client and a connection setup).
10169
Willy Tarreaub16a5742010-01-10 14:46:16 +010010170 See also : "timeout http-request", "timeout client".
10171
10172
Willy Tarreau036fae02008-01-06 13:24:40 +010010173timeout http-request <timeout>
10174 Set the maximum allowed time to wait for a complete HTTP request
10175 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010176 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010177 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010178 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010179 can be in any other unit if the number is suffixed by the unit,
10180 as explained at the top of this document.
10181
10182 In order to offer DoS protection, it may be required to lower the maximum
10183 accepted time to receive a complete HTTP request without affecting the client
10184 timeout. This helps protecting against established connections on which
10185 nothing is sent. The client timeout cannot offer a good protection against
10186 this abuse because it is an inactivity timeout, which means that if the
10187 attacker sends one character every now and then, the timeout will not
10188 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010189 types, the request will be aborted if it does not complete in time. When the
10190 timeout expires, an HTTP 408 response is sent to the client to inform it
10191 about the problem, and the connection is closed. The logs will report
10192 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010193 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010194 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10195 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010196
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010197 By default, this timeout only applies to the header part of the request,
10198 and not to any data. As soon as the empty line is received, this timeout is
10199 not used anymore. When combined with "option http-buffer-request", this
10200 timeout also applies to the body of the request..
10201 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010202 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010203
10204 Generally it is enough to set it to a few seconds, as most clients send the
10205 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010206 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010207 generally work on local networks as long as there are no packet losses. This
10208 will prevent people from sending bare HTTP requests using telnet.
10209
10210 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010211 chunk of the incoming request. It should be set in the frontend to take
10212 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10213 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010214
Willy Tarreau0f228a02015-05-01 15:37:53 +020010215 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010216 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010217
Willy Tarreau844e3c52008-01-11 16:28:18 +010010218
10219timeout queue <timeout>
10220 Set the maximum time to wait in the queue for a connection slot to be free
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 When a server's maxconn is reached, connections are left pending in a queue
10229 which may be server-specific or global to the backend. In order not to wait
10230 indefinitely, a timeout is applied to requests pending in the queue. If the
10231 timeout is reached, it is considered that the request will almost never be
10232 served, so it is dropped and a 503 error is returned to the client.
10233
10234 The "timeout queue" statement allows to fix the maximum time for a request to
10235 be left pending in a queue. If unspecified, the same value as the backend's
10236 connection timeout ("timeout connect") is used, for backwards compatibility
10237 with older versions with no "timeout queue" parameter.
10238
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010239 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010240
10241
10242timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010010243 Set the maximum inactivity time on the server side.
10244 May be used in sections : defaults | frontend | listen | backend
10245 yes | no | yes | yes
10246 Arguments :
10247 <timeout> is the timeout value specified in milliseconds by default, but
10248 can be in any other unit if the number is suffixed by the unit,
10249 as explained at the top of this document.
10250
10251 The inactivity timeout applies when the server is expected to acknowledge or
10252 send data. In HTTP mode, this timeout is particularly important to consider
10253 during the first phase of the server's response, when it has to send the
10254 headers, as it directly represents the server's processing time for the
10255 request. To find out what value to put there, it's often good to start with
10256 what would be considered as unacceptable response times, then check the logs
10257 to observe the response time distribution, and adjust the value accordingly.
10258
10259 The value is specified in milliseconds by default, but can be in any other
10260 unit if the number is suffixed by the unit, as specified at the top of this
10261 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10262 recommended that the client timeout remains equal to the server timeout in
10263 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010264 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010265 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010266 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10267 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010268 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10269 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010270
10271 This parameter is specific to backends, but can be specified once for all in
10272 "defaults" sections. This is in fact one of the easiest solutions not to
10273 forget about it. An unspecified timeout results in an infinite timeout, which
10274 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010275 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010010276 the system if the system's timeouts are not configured either.
10277
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010278 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010279
Willy Tarreau05cdd962014-05-10 14:30:07 +020010280
10281timeout server-fin <timeout>
10282 Set the inactivity timeout on the server side for half-closed connections.
10283 May be used in sections : defaults | frontend | listen | backend
10284 yes | no | yes | yes
10285 Arguments :
10286 <timeout> is the timeout value specified in milliseconds by default, but
10287 can be in any other unit if the number is suffixed by the unit,
10288 as explained at the top of this document.
10289
10290 The inactivity timeout applies when the server is expected to acknowledge or
10291 send data while one direction is already shut down. This timeout is different
10292 from "timeout server" in that it only applies to connections which are closed
10293 in one direction. This is particularly useful to avoid keeping connections in
10294 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10295 This problem is particularly common long connections such as RDP or WebSocket.
10296 Note that this timeout can override "timeout tunnel" when a connection shuts
10297 down in one direction. This setting was provided for completeness, but in most
10298 situations, it should not be needed.
10299
10300 This parameter is specific to backends, but can be specified once for all in
10301 "defaults" sections. By default it is not set, so half-closed connections
10302 will use the other timeouts (timeout.server or timeout.tunnel).
10303
10304 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10305
Willy Tarreau844e3c52008-01-11 16:28:18 +010010306
10307timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010308 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010309 May be used in sections : defaults | frontend | listen | backend
10310 yes | yes | yes | yes
10311 Arguments :
10312 <timeout> is the tarpit duration specified in milliseconds by default, but
10313 can be in any other unit if the number is suffixed by the unit,
10314 as explained at the top of this document.
10315
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020010316 When a connection is tarpitted using "http-request tarpit", it is maintained
10317 open with no activity for a certain amount of time, then closed. "timeout
10318 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010319
10320 The value is specified in milliseconds by default, but can be in any other
10321 unit if the number is suffixed by the unit, as specified at the top of this
10322 document. If unspecified, the same value as the backend's connection timeout
10323 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010324 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010325
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010326 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010327
10328
Willy Tarreauce887fd2012-05-12 12:50:00 +020010329timeout tunnel <timeout>
10330 Set the maximum inactivity time on the client and server side for tunnels.
10331 May be used in sections : defaults | frontend | listen | backend
10332 yes | no | yes | yes
10333 Arguments :
10334 <timeout> is the timeout value specified in milliseconds by default, but
10335 can be in any other unit if the number is suffixed by the unit,
10336 as explained at the top of this document.
10337
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010338 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010339 between a client and a server, and the connection remains inactive in both
10340 directions. This timeout supersedes both the client and server timeouts once
10341 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010342 analyzer remains attached to either connection (e.g. tcp content rules are
10343 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010344 when switching to the WebSocket protocol, or forwarding a CONNECT request
10345 to a proxy), or after the first response when no keepalive/close option is
10346 specified.
10347
Willy Tarreau05cdd962014-05-10 14:30:07 +020010348 Since this timeout is usually used in conjunction with long-lived connections,
10349 it usually is a good idea to also set "timeout client-fin" to handle the
10350 situation where a client suddenly disappears from the net and does not
10351 acknowledge a close, or sends a shutdown and does not acknowledge pending
10352 data anymore. This can happen in lossy networks where firewalls are present,
10353 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10354 state.
10355
Willy Tarreauce887fd2012-05-12 12:50:00 +020010356 The value is specified in milliseconds by default, but can be in any other
10357 unit if the number is suffixed by the unit, as specified at the top of this
10358 document. Whatever the expected normal idle time, it is a good practice to
10359 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010360 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010361
10362 This parameter is specific to backends, but can be specified once for all in
10363 "defaults" sections. This is in fact one of the easiest solutions not to
10364 forget about it.
10365
10366 Example :
10367 defaults http
10368 option http-server-close
10369 timeout connect 5s
10370 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010371 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010372 timeout server 30s
10373 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10374
Willy Tarreau05cdd962014-05-10 14:30:07 +020010375 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010376
10377
Willy Tarreau844e3c52008-01-11 16:28:18 +010010378transparent (deprecated)
10379 Enable client-side transparent proxying
10380 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010381 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010382 Arguments : none
10383
10384 This keyword was introduced in order to provide layer 7 persistence to layer
10385 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10386 connection for a remote address to a local process (here HAProxy), and let
10387 this process know what address was initially requested. When this option is
10388 used, sessions without cookies will be forwarded to the original destination
10389 IP address of the incoming request (which should match that of another
10390 equipment), while requests with cookies will still be forwarded to the
10391 appropriate server.
10392
10393 The "transparent" keyword is deprecated, use "option transparent" instead.
10394
10395 Note that contrary to a common belief, this option does NOT make HAProxy
10396 present the client's IP to the server when establishing the connection.
10397
Willy Tarreau844e3c52008-01-11 16:28:18 +010010398 See also: "option transparent"
10399
William Lallemanda73203e2012-03-12 12:48:57 +010010400unique-id-format <string>
10401 Generate a unique ID for each request.
10402 May be used in sections : defaults | frontend | listen | backend
10403 yes | yes | yes | no
10404 Arguments :
10405 <string> is a log-format string.
10406
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010407 This keyword creates a ID for each request using the custom log format. A
10408 unique ID is useful to trace a request passing through many components of
10409 a complex infrastructure. The newly created ID may also be logged using the
10410 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010411
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010412 The format should be composed from elements that are guaranteed to be
10413 unique when combined together. For instance, if multiple haproxy instances
10414 are involved, it might be important to include the node name. It is often
10415 needed to log the incoming connection's source and destination addresses
10416 and ports. Note that since multiple requests may be performed over the same
10417 connection, including a request counter may help differentiate them.
10418 Similarly, a timestamp may protect against a rollover of the counter.
10419 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010420
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010421 It is recommended to use hexadecimal notation for many fields since it
10422 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010423
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010424 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010425
Julien Vehentf21be322014-03-07 08:27:34 -050010426 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010427
10428 will generate:
10429
10430 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10431
10432 See also: "unique-id-header"
10433
10434unique-id-header <name>
10435 Add a unique ID header in the HTTP request.
10436 May be used in sections : defaults | frontend | listen | backend
10437 yes | yes | yes | no
10438 Arguments :
10439 <name> is the name of the header.
10440
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010441 Add a unique-id header in the HTTP request sent to the server, using the
10442 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010443
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010444 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010445
Julien Vehentf21be322014-03-07 08:27:34 -050010446 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010447 unique-id-header X-Unique-ID
10448
10449 will generate:
10450
10451 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10452
10453 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010454
Willy Tarreauf51658d2014-04-23 01:21:56 +020010455use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010456 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010457 May be used in sections : defaults | frontend | listen | backend
10458 no | yes | yes | no
10459 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010460 <backend> is the name of a valid backend or "listen" section, or a
10461 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010462
Willy Tarreauf51658d2014-04-23 01:21:56 +020010463 <condition> is a condition composed of ACLs, as described in section 7. If
10464 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010465
10466 When doing content-switching, connections arrive on a frontend and are then
10467 dispatched to various backends depending on a number of conditions. The
10468 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010469 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010470 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010471 source address validation) or combined with a "tcp-request" rule to wait for
10472 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010473
10474 There may be as many "use_backend" rules as desired. All of these rules are
10475 evaluated in their declaration order, and the first one which matches will
10476 assign the backend.
10477
10478 In the first form, the backend will be used if the condition is met. In the
10479 second form, the backend will be used if the condition is not met. If no
10480 condition is valid, the backend defined with "default_backend" will be used.
10481 If no default backend is defined, either the servers in the same section are
10482 used (in case of a "listen" section) or, in case of a frontend, no server is
10483 used and a 503 service unavailable response is returned.
10484
Willy Tarreau51aecc72009-07-12 09:47:04 +020010485 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010486 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010487 and backend processing will immediately follow, or the backend will wait for
10488 a complete HTTP request to get in. This feature is useful when a frontend
10489 must decode several protocols on a unique port, one of them being HTTP.
10490
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010491 When <backend> is a simple name, it is resolved at configuration time, and an
10492 error is reported if the specified backend does not exist. If <backend> is
10493 a log-format string instead, no check may be done at configuration time, so
10494 the backend name is resolved dynamically at run time. If the resulting
10495 backend name does not correspond to any valid backend, no other rule is
10496 evaluated, and the default_backend directive is applied instead. Note that
10497 when using dynamic backend names, it is highly recommended to use a prefix
10498 that no other backend uses in order to ensure that an unauthorized backend
10499 cannot be forced from the request.
10500
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010501 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010502 used to detect the association between frontends and backends to compute the
10503 backend's "fullconn" setting. This cannot be done for dynamic names.
10504
10505 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10506 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010507
Christopher Fauletb30b3102019-09-12 23:03:09 +020010508use-fcgi-app <name>
10509 Defines the FastCGI application to use for the backend.
10510 May be used in sections : defaults | frontend | listen | backend
10511 no | no | yes | yes
10512 Arguments :
10513 <name> is the name of the FastCGI application to use.
10514
10515 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010010516
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010517use-server <server> if <condition>
10518use-server <server> unless <condition>
10519 Only use a specific server if/unless an ACL-based condition is matched.
10520 May be used in sections : defaults | frontend | listen | backend
10521 no | no | yes | yes
10522 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010523 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010524
10525 <condition> is a condition composed of ACLs, as described in section 7.
10526
10527 By default, connections which arrive to a backend are load-balanced across
10528 the available servers according to the configured algorithm, unless a
10529 persistence mechanism such as a cookie is used and found in the request.
10530
10531 Sometimes it is desirable to forward a particular request to a specific
10532 server without having to declare a dedicated backend for this server. This
10533 can be achieved using the "use-server" rules. These rules are evaluated after
10534 the "redirect" rules and before evaluating cookies, and they have precedence
10535 on them. There may be as many "use-server" rules as desired. All of these
10536 rules are evaluated in their declaration order, and the first one which
10537 matches will assign the server.
10538
10539 If a rule designates a server which is down, and "option persist" is not used
10540 and no force-persist rule was validated, it is ignored and evaluation goes on
10541 with the next rules until one matches.
10542
10543 In the first form, the server will be used if the condition is met. In the
10544 second form, the server will be used if the condition is not met. If no
10545 condition is valid, the processing continues and the server will be assigned
10546 according to other persistence mechanisms.
10547
10548 Note that even if a rule is matched, cookie processing is still performed but
10549 does not assign the server. This allows prefixed cookies to have their prefix
10550 stripped.
10551
10552 The "use-server" statement works both in HTTP and TCP mode. This makes it
10553 suitable for use with content-based inspection. For instance, a server could
10554 be selected in a farm according to the TLS SNI field. And if these servers
10555 have their weight set to zero, they will not be used for other traffic.
10556
10557 Example :
10558 # intercept incoming TLS requests based on the SNI field
10559 use-server www if { req_ssl_sni -i www.example.com }
10560 server www 192.168.0.1:443 weight 0
10561 use-server mail if { req_ssl_sni -i mail.example.com }
10562 server mail 192.168.0.1:587 weight 0
10563 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010564 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010565 # all the rest is forwarded to this server
10566 server default 192.168.0.2:443 check
10567
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010568 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010569
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010570
Davor Ocelice9ed2812017-12-25 17:49:28 +0100105715. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010572--------------------------
10573
10574The "bind", "server" and "default-server" keywords support a number of settings
10575depending on some build options and on the system HAProxy was built on. These
10576settings generally each consist in one word sometimes followed by a value,
10577written on the same line as the "bind" or "server" line. All these options are
10578described in this section.
10579
10580
105815.1. Bind options
10582-----------------
10583
10584The "bind" keyword supports a certain number of settings which are all passed
10585as arguments on the same line. The order in which those arguments appear makes
10586no importance, provided that they appear after the bind address. All of these
10587parameters are optional. Some of them consist in a single words (booleans),
10588while other ones expect a value after them. In this case, the value must be
10589provided immediately after the setting name.
10590
10591The currently supported settings are the following ones.
10592
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010593accept-netscaler-cip <magic number>
10594 Enforces the use of the NetScaler Client IP insertion protocol over any
10595 connection accepted by any of the TCP sockets declared on the same line. The
10596 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10597 the incoming connection to be used everywhere an address is used, with the
10598 only exception of "tcp-request connection" rules which will only see the
10599 real connection address. Logs will reflect the addresses indicated in the
10600 protocol, unless it is violated, in which case the real address will still
10601 be used. This keyword combined with support from external components can be
10602 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010603 mechanism which is not always reliable and not even always usable. See also
10604 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10605 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010606
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010607accept-proxy
10608 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010609 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10610 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010611 3/4 addresses of the incoming connection to be used everywhere an address is
10612 used, with the only exception of "tcp-request connection" rules which will
10613 only see the real connection address. Logs will reflect the addresses
10614 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010615 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010616 components can be used as an efficient and reliable alternative to the
10617 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010618 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10619 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010620
Olivier Houchardc2aae742017-09-22 18:26:28 +020010621allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010622 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010010623 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050010624 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010010625 that are idempotent. You can use the "wait-for-handshake" action for any
10626 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020010627
Willy Tarreauab861d32013-04-02 02:30:41 +020010628alpn <protocols>
10629 This enables the TLS ALPN extension and advertises the specified protocol
10630 list as supported on top of ALPN. The protocol list consists in a comma-
10631 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050010632 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020010633 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010634 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10635 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10636 now obsolete NPN extension. At the time of writing this, most browsers still
10637 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10638 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10639 are expected to be supported, both versions can be advertised, in order of
10640 preference, like below :
10641
10642 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010643
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010644backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010010645 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010646 backlog is used instead, which generally defaults to the maxconn value.
10647
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010648curves <curves>
10649 This setting is only available when support for OpenSSL was built in. It sets
10650 the string describing the list of elliptic curves algorithms ("curve suite")
10651 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10652 string is a colon-delimited list of curve name.
10653 Example: "X25519:P-256" (without quote)
10654 When "curves" is set, "ecdhe" parameter is ignored.
10655
Emeric Brun7fb34422012-09-28 15:26:15 +020010656ecdhe <named curve>
10657 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010658 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10659 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010660
Emeric Brunfd33a262012-10-11 16:28:27 +020010661ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010662 This setting is only available when support for OpenSSL was built in. It
10663 designates a PEM file from which to load CA certificates used to verify
10664 client's certificate.
10665
Emeric Brunb6dc9342012-09-28 17:55:37 +020010666ca-ignore-err [all|<errorID>,...]
10667 This setting is only available when support for OpenSSL was built in.
10668 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10669 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10670 error is ignored.
10671
Christopher Faulet31af49d2015-06-09 17:29:50 +020010672ca-sign-file <cafile>
10673 This setting is only available when support for OpenSSL was built in. It
10674 designates a PEM file containing both the CA certificate and the CA private
10675 key used to create and sign server's certificates. This is a mandatory
10676 setting when the dynamic generation of certificates is enabled. See
10677 'generate-certificates' for details.
10678
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010679ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010680 This setting is only available when support for OpenSSL was built in. It is
10681 the CA private key passphrase. This setting is optional and used only when
10682 the dynamic generation of certificates is enabled. See
10683 'generate-certificates' for details.
10684
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010685ciphers <ciphers>
10686 This setting is only available when support for OpenSSL was built in. It sets
10687 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000010688 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010689 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010690 information and recommendations see e.g.
10691 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10692 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10693 cipher configuration, please check the "ciphersuites" keyword.
10694
10695ciphersuites <ciphersuites>
10696 This setting is only available when support for OpenSSL was built in and
10697 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10698 the list of cipher algorithms ("cipher suite") that are negotiated during the
10699 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010700 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
10701 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010702
Emeric Brunfd33a262012-10-11 16:28:27 +020010703crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010704 This setting is only available when support for OpenSSL was built in. It
10705 designates a PEM file from which to load certificate revocation list used
10706 to verify client's certificate.
10707
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010708crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010709 This setting is only available when support for OpenSSL was built in. It
10710 designates a PEM file containing both the required certificates and any
10711 associated private keys. This file can be built by concatenating multiple
10712 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10713 requires an intermediate certificate, this can also be concatenated into this
10714 file.
10715
10716 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10717 are loaded.
10718
10719 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010720 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010721 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10722 specified multiple times in order to load certificates from multiple files or
10723 directories. The certificates will be presented to clients who provide a
10724 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010725 subjects. Wildcards are supported, where a wildcard character '*' is used
10726 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010727 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010728
10729 If no SNI is provided by the client or if the SSL library does not support
10730 TLS extensions, or if the client provides an SNI hostname which does not
10731 match any certificate, then the first loaded certificate will be presented.
10732 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010733 recommended to load the default one first as a file or to ensure that it will
10734 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010735
Emeric Brune032bfa2012-09-28 13:01:45 +020010736 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010737
Davor Ocelice9ed2812017-12-25 17:49:28 +010010738 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010739 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010740 choose a web server that the CA believes requires an intermediate CA (for
10741 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010742 others, e.g. nginx, result in a wrong bundle that will not work for some
10743 clients).
10744
Emeric Brun4147b2e2014-06-16 18:36:30 +020010745 For each PEM file, haproxy checks for the presence of file at the same path
10746 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10747 Status Request extension (also known as "OCSP stapling") is automatically
10748 enabled. The content of this file is optional. If not empty, it must contain
10749 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10750 must comply with the following rules: it has to indicate a good status,
10751 it has to be a single response for the certificate of the PEM file, and it
10752 has to be valid at the moment of addition. If these rules are not respected
10753 the OCSP Response is ignored and a warning is emitted. In order to identify
10754 which certificate an OCSP Response applies to, the issuer's certificate is
10755 necessary. If the issuer's certificate is not found in the PEM file, it will
10756 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10757 if it exists otherwise it will fail with an error.
10758
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010759 For each PEM file, haproxy also checks for the presence of file at the same
10760 path suffixed by ".sctl". If such file is found, support for Certificate
10761 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10762 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10763 to check basic syntax, but no signatures are verified.
10764
yanbzhu6c25e9e2016-01-05 12:52:02 -050010765 There are cases where it is desirable to support multiple key types, e.g. RSA
10766 and ECDSA in the cipher suites offered to the clients. This allows clients
10767 that support EC certificates to be able to use EC ciphers, while
10768 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010769
10770 In order to provide this functionality, multiple PEM files, each with a
10771 different key type, are required. To associate these PEM files into a
10772 "cert bundle" that is recognized by haproxy, they must be named in the
10773 following way: All PEM files that are to be bundled must have the same base
10774 name, with a suffix indicating the key type. Currently, three suffixes are
10775 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10776 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10777 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10778 suffix matters. To load this bundle into haproxy, specify the base name only:
10779
10780 Example : bind :8443 ssl crt example.pem
10781
yanbzhu6c25e9e2016-01-05 12:52:02 -050010782 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010783 a cert bundle.
10784
Davor Ocelice9ed2812017-12-25 17:49:28 +010010785 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010786 support multiple key types. PEM files are combined based on Common Name
10787 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10788 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10789 entries in the certificates in that bundle, haproxy will not be able to
10790 provide multi-cert support.
10791
10792 Assuming bundle in the example above contained the following:
10793
10794 Filename | CN | SAN
10795 -------------------+-----------------+-------------------
10796 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010797 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010798 example.pem.ecdsa | www.example.com | ecdsa.example.com
10799 -------------------+-----------------+-------------------
10800
10801 Users connecting with an SNI of "www.example.com" will be able
10802 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10803 "rsa.example.com" will only be able to use RSA cipher suites, and users
10804 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010805 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10806 no need to bundle certificates. ECDSA certificate will be preferred if client
10807 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010808
10809 If a directory name is given as the <cert> argument, haproxy will
10810 automatically search and load bundled files in that directory.
10811
10812 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10813 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10814 time, sctl is not supported in multi-certificate bundling.
10815
Emeric Brunb6dc9342012-09-28 17:55:37 +020010816crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010817 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010818 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010819 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010820 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010821
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010822crt-list <file>
10823 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010824 designates a list of PEM file with an optional ssl configuration and a SNI
10825 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010826
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010827 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10828
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010829 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10830 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010831 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010832 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010833
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010834 Wildcards are supported in the SNI filter. Negative filter are also supported,
10835 only useful in combination with a wildcard filter to exclude a particular SNI.
10836 The certificates will be presented to clients who provide a valid TLS Server
10837 Name Indication field matching one of the SNI filters. If no SNI filter is
10838 specified, the CN and alt subjects are used. This directive may be specified
10839 multiple times. See the "crt" option for more information. The default
10840 certificate is still needed to meet OpenSSL expectations. If it is not used,
10841 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010842
yanbzhu6c25e9e2016-01-05 12:52:02 -050010843 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010844 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010845 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10846 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10847 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010848
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010849 crt-list file example:
10850 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010851 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010852 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010853 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010854
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010855defer-accept
10856 Is an optional keyword which is supported only on certain Linux kernels. It
10857 states that a connection will only be accepted once some data arrive on it,
10858 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010010859 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010860 performance by ensuring that most of the request is already available when
10861 the connection is accepted. On the other hand, it will not be able to detect
10862 connections which don't talk. It is important to note that this option is
10863 broken in all kernels up to 2.6.31, as the connection is never accepted until
10864 the client talks. This can cause issues with front firewalls which would see
10865 an established connection while the proxy will only see it in SYN_RECV. This
10866 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
10867
William Lallemandf6975e92017-05-26 17:42:10 +020010868expose-fd listeners
10869 This option is only usable with the stats socket. It gives your stats socket
10870 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020010871 During a reload with the master-worker mode, the process is automatically
10872 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010010873 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020010874
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010875force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010876 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010877 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010878 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010879 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010880
10881force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010882 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010883 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010884 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010885
10886force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010887 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010888 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010889 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010890
10891force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010892 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010893 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010894 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010895
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010896force-tlsv13
10897 This option enforces use of TLSv1.3 only on SSL connections instantiated from
10898 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010899 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010900
Christopher Faulet31af49d2015-06-09 17:29:50 +020010901generate-certificates
10902 This setting is only available when support for OpenSSL was built in. It
10903 enables the dynamic SSL certificates generation. A CA certificate and its
10904 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
10905 a transparent forward proxy, SSL requests generate errors because of a common
10906 name mismatch on the certificate presented to the client. With this option
10907 enabled, HAProxy will try to forge a certificate using the SNI hostname
10908 indicated by the client. This is done only if no certificate matches the SNI
10909 hostname (see 'crt-list'). If an error occurs, the default certificate is
10910 used, else the 'strict-sni' option is set.
10911 It can also be used when HAProxy is configured as a reverse proxy to ease the
10912 deployment of an architecture with many backends.
10913
10914 Creating a SSL certificate is an expensive operation, so a LRU cache is used
10915 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010010916 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020010917 certificate is used many times.
10918
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010919gid <gid>
10920 Sets the group of the UNIX sockets to the designated system gid. It can also
10921 be set by default in the global section's "unix-bind" statement. Note that
10922 some platforms simply ignore this. This setting is equivalent to the "group"
10923 setting except that the group ID is used instead of its name. This setting is
10924 ignored by non UNIX sockets.
10925
10926group <group>
10927 Sets the group of the UNIX sockets to the designated system group. It can
10928 also be set by default in the global section's "unix-bind" statement. Note
10929 that some platforms simply ignore this. This setting is equivalent to the
10930 "gid" setting except that the group name is used instead of its gid. This
10931 setting is ignored by non UNIX sockets.
10932
10933id <id>
10934 Fixes the socket ID. By default, socket IDs are automatically assigned, but
10935 sometimes it is more convenient to fix them to ease monitoring. This value
10936 must be strictly positive and unique within the listener/frontend. This
10937 option can only be used when defining only a single socket.
10938
10939interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010010940 Restricts the socket to a specific interface. When specified, only packets
10941 received from that particular interface are processed by the socket. This is
10942 currently only supported on Linux. The interface must be a primary system
10943 interface, not an aliased interface. It is also possible to bind multiple
10944 frontends to the same address if they are bound to different interfaces. Note
10945 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010010946 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
10947 uses the same interface as inbound traffic, and its associated routing table,
10948 even if there are explicit routes through different interfaces configured.
10949 This can prove useful to address asymmetric routing issues when the same
10950 client IP addresses need to be able to reach frontends hosted on different
10951 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010952
Willy Tarreauabb175f2012-09-24 12:43:26 +020010953level <level>
10954 This setting is used with the stats sockets only to restrict the nature of
10955 the commands that can be issued on the socket. It is ignored by other
10956 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010010957 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020010958 read, and no change is allowed. It would make sense on systems where it
10959 is not easy to restrict access to the socket.
10960 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010961 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020010962 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010010963 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020010964 all counters).
10965
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020010966severity-output <format>
10967 This setting is used with the stats sockets only to configure severity
10968 level output prepended to informational feedback messages. Severity
10969 level of messages can range between 0 and 7, conforming to syslog
10970 rfc5424. Valid and successful socket commands requesting data
10971 (i.e. "show map", "get acl foo" etc.) will never have a severity level
10972 prepended. It is ignored by other sockets. <format> can be one of :
10973 - "none" (default) no severity level is prepended to feedback messages.
10974 - "number" severity level is prepended as a number.
10975 - "string" severity level is prepended as a string following the
10976 rfc5424 convention.
10977
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010978maxconn <maxconn>
10979 Limits the sockets to this number of concurrent connections. Extraneous
10980 connections will remain in the system's backlog until a connection is
10981 released. If unspecified, the limit will be the same as the frontend's
10982 maxconn. Note that in case of port ranges or multiple addresses, the same
10983 value will be applied to each socket. This setting enables different
10984 limitations on expensive sockets, for instance SSL entries which may easily
10985 eat all memory.
10986
10987mode <mode>
10988 Sets the octal mode used to define access permissions on the UNIX socket. It
10989 can also be set by default in the global section's "unix-bind" statement.
10990 Note that some platforms simply ignore this. This setting is ignored by non
10991 UNIX sockets.
10992
10993mss <maxseg>
10994 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
10995 connections. This can be used to force a lower MSS for certain specific
10996 ports, for instance for connections passing through a VPN. Note that this
10997 relies on a kernel feature which is theoretically supported under Linux but
10998 was buggy in all versions prior to 2.6.28. It may or may not work on other
10999 operating systems. It may also not change the advertised value but change the
11000 effective size of outgoing segments. The commonly advertised value for TCPv4
11001 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11002 positive, it will be used as the advertised MSS. If it is negative, it will
11003 indicate by how much to reduce the incoming connection's advertised MSS for
11004 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11005
11006name <name>
11007 Sets an optional name for these sockets, which will be reported on the stats
11008 page.
11009
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011010namespace <name>
11011 On Linux, it is possible to specify which network namespace a socket will
11012 belong to. This directive makes it possible to explicitly bind a listener to
11013 a namespace different from the default one. Please refer to your operating
11014 system's documentation to find more details about network namespaces.
11015
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011016nice <nice>
11017 Sets the 'niceness' of connections initiated from the socket. Value must be
11018 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11019 means that such connections are more friendly to others and easily offer
11020 their place in the scheduler. On the opposite, negative values mean that
11021 connections want to run with a higher priority than others. The difference
11022 only happens under high loads when the system is close to saturation.
11023 Negative values are appropriate for low-latency or administration services,
11024 and high values are generally recommended for CPU intensive tasks such as SSL
11025 processing or bulk transfers which are less sensible to latency. For example,
11026 it may make sense to use a positive value for an SMTP socket and a negative
11027 one for an RDP socket.
11028
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011029no-ca-names
11030 This setting is only available when support for OpenSSL was built in. It
11031 prevents from send CA names in server hello message when ca-file is used.
11032
Emeric Brun9b3009b2012-10-05 11:55:06 +020011033no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +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 SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011036 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011037 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011038 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11039 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011040
Emeric Brun90ad8722012-10-02 14:00:59 +020011041no-tls-tickets
11042 This setting is only available when support for OpenSSL was built in. It
11043 disables the stateless session resumption (RFC 5077 TLS Ticket
11044 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011045 session resumption is more expensive in CPU usage. This option is also
11046 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011047
Emeric Brun9b3009b2012-10-05 11:55:06 +020011048no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011049 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011050 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011051 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011052 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011053 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11054 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011055
Emeric Brun9b3009b2012-10-05 11:55:06 +020011056no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011057 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011058 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011059 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011060 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011061 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11062 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011063
Emeric Brun9b3009b2012-10-05 11:55:06 +020011064no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011065 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011066 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011067 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011068 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011069 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11070 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011071
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011072no-tlsv13
11073 This setting is only available when support for OpenSSL was built in. It
11074 disables support for TLSv1.3 on any sockets instantiated from the listener
11075 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11076 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011077 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11078 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011079
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011080npn <protocols>
11081 This enables the NPN TLS extension and advertises the specified protocol list
11082 as supported on top of NPN. The protocol list consists in a comma-delimited
11083 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050011084 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011085 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011086 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11087 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11088 version of OpenSSL, NPN might still be used as most clients still support it
11089 at the time of writing this. It is possible to enable both NPN and ALPN
11090 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011091
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011092prefer-client-ciphers
11093 Use the client's preference when selecting the cipher suite, by default
11094 the server's preference is enforced. This option is also available on
11095 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011096 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11097 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11098 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011099
Christopher Fauletc644fa92017-11-23 22:44:11 +010011100process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010011101 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010011102 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011103 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011104 between the two is applied. If in the end the listener is not allowed to run
11105 on any remaining process, a warning is emitted, and the listener will either
11106 run on the first process of the listener if a single process was specified,
11107 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011108 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010011109 connections for this listener, for the the process set. If multiple processes
11110 and threads are configured, a warning is emitted, as it either results from a
11111 configuration error or a misunderstanding of these models. For the unlikely
11112 case where several ranges are needed, this directive may be repeated.
11113 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010011114
11115 all | odd | even | number[-[number]]
11116
11117 Ranges can be partially defined. The higher bound can be omitted. In such
11118 case, it is replaced by the corresponding maximum value. The main purpose of
11119 this directive is to be used with the stats sockets and have one different
11120 socket per process. The second purpose is to have multiple bind lines sharing
11121 the same IP:port but not the same process in a listener, so that the system
11122 can distribute the incoming connections into multiple queues and allow a
11123 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11124 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011125
Christopher Fauleta717b992018-04-10 14:43:00 +020011126proto <name>
11127 Forces the multiplexer's protocol to use for the incoming connections. It
11128 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11129 be usable on the frontend side. The list of available protocols is reported
11130 in haproxy -vv.
11131 Idea behind this optipon is to bypass the selection of the best multiplexer's
11132 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011133 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011134 h2" on the bind line.
11135
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011136ssl
11137 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011138 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011139 certificate is necessary (see "crt" above). All contents in the buffers will
11140 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011141 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11142 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011143
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011144ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11145 This option enforces use of <version> or lower on SSL connections instantiated
11146 from this listener. This option is also available on global statement
11147 "ssl-default-bind-options". See also "ssl-min-ver".
11148
11149ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11150 This option enforces use of <version> or upper on SSL connections instantiated
11151 from this listener. This option is also available on global statement
11152 "ssl-default-bind-options". See also "ssl-max-ver".
11153
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011154strict-sni
11155 This setting is only available when support for OpenSSL was built in. The
11156 SSL/TLS negotiation is allow only if the client provided an SNI which match
11157 a certificate. The default certificate is not used.
11158 See the "crt" option for more information.
11159
Willy Tarreau2af207a2015-02-04 00:45:58 +010011160tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011161 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011162 listening socket. This option is available on Linux since version 2.6.37. It
11163 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011164 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011165 useful on long-lived connections experiencing long idle periods such as
11166 remote terminals or database connection pools, where the client and server
11167 timeouts must remain high to allow a long period of idle, but where it is
11168 important to detect that the client has disappeared in order to release all
11169 resources associated with its connection (and the server's session). The
11170 argument is a delay expressed in milliseconds by default. This only works
11171 for regular TCP connections, and is ignored for other protocols.
11172
Willy Tarreau1c862c52012-10-05 16:21:00 +020011173tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011174 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011175 enables TCP Fast Open on the listening socket, which means that clients which
11176 support this feature will be able to send a request and receive a response
11177 during the 3-way handshake starting from second connection, thus saving one
11178 round-trip after the first connection. This only makes sense with protocols
11179 that use high connection rates and where each round trip matters. This can
11180 possibly cause issues with many firewalls which do not accept data on SYN
11181 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011182 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11183 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11184 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011185
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011186tls-ticket-keys <keyfile>
11187 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010011188 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
11189 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
11190 The first key determines the key length used for next keys: you can't mix
11191 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
11192 build option (default 3) and at least as many keys need to be present in
11193 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
11194 penultimate one for encryption. This enables easy key rotation by just
11195 appending new key to the file and reloading the process. Keys must be
11196 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
11197 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011198 storage such as hard drives (hint: use tmpfs and don't swap those files).
11199 Lifetime hint can be changed using tune.ssl.timeout.
11200
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011201transparent
11202 Is an optional keyword which is supported only on certain Linux kernels. It
11203 indicates that the addresses will be bound even if they do not belong to the
11204 local machine, and that packets targeting any of these addresses will be
11205 intercepted just as if the addresses were locally configured. This normally
11206 requires that IP forwarding is enabled. Caution! do not use this with the
11207 default address '*', as it would redirect any traffic for the specified port.
11208 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11209 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11210 kernel version. Some distribution kernels include backports of the feature,
11211 so check for support with your vendor.
11212
Willy Tarreau77e3af92012-11-24 15:07:23 +010011213v4v6
11214 Is an optional keyword which is supported only on most recent systems
11215 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11216 and IPv6 when it uses the default address. Doing so is sometimes necessary
11217 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011218 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011219
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011220v6only
11221 Is an optional keyword which is supported only on most recent systems
11222 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11223 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011224 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11225 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011226
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011227uid <uid>
11228 Sets the owner of the UNIX sockets to the designated system uid. It can also
11229 be set by default in the global section's "unix-bind" statement. Note that
11230 some platforms simply ignore this. This setting is equivalent to the "user"
11231 setting except that the user numeric ID is used instead of its name. This
11232 setting is ignored by non UNIX sockets.
11233
11234user <user>
11235 Sets the owner of the UNIX sockets to the designated system user. It can also
11236 be set by default in the global section's "unix-bind" statement. Note that
11237 some platforms simply ignore this. This setting is equivalent to the "uid"
11238 setting except that the user name is used instead of its uid. This setting is
11239 ignored by non UNIX sockets.
11240
Emeric Brun1a073b42012-09-28 17:07:34 +020011241verify [none|optional|required]
11242 This setting is only available when support for OpenSSL was built in. If set
11243 to 'none', client certificate is not requested. This is the default. In other
11244 cases, a client certificate is requested. If the client does not provide a
11245 certificate after the request and if 'verify' is set to 'required', then the
11246 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011247 certificate provided by the client is always verified using CAs from
11248 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11249 is aborted, regardless of the 'verify' option, unless the error code exactly
11250 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011251
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200112525.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011253------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011254
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011255The "server" and "default-server" keywords support a certain number of settings
11256which are all passed as arguments on the server line. The order in which those
11257arguments appear does not count, and they are all optional. Some of those
11258settings are single words (booleans) while others expect one or several values
11259after them. In this case, the values must immediately follow the setting name.
11260Except default-server, all those settings must be specified after the server's
11261address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011262
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011263 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011264 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011265
Frédéric Lécailled2376272017-03-21 18:52:12 +010011266Note that all these settings are supported both by "server" and "default-server"
11267keywords, except "id" which is only supported by "server".
11268
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011269The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011270
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011271addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011272 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011273 to send health-checks or to probe the agent-check. On some servers, it may be
11274 desirable to dedicate an IP address to specific component able to perform
11275 complex tests which are more suitable to health-checks than the application.
11276 This parameter is ignored if the "check" parameter is not set. See also the
11277 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011278
Simon Hormand60d6912013-11-25 10:46:36 +090011279agent-check
11280 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011281 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010011282 to the port set by the "agent-port" parameter and reading an ASCII string
11283 terminated by the first '\r' or '\n' met. The string is made of a series of
11284 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011285
Willy Tarreau81f5d942013-12-09 20:51:51 +010011286 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011287 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011288 weight of a server as configured when haproxy starts. Note that a zero
11289 weight is reported on the stats page as "DRAIN" since it has the same
11290 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011291
Davor Ocelice9ed2812017-12-25 17:49:28 +010011292 - The string "maxconn:" followed by an integer (no space between). Values
11293 in this format will set the maxconn of a server. The maximum number of
11294 connections advertised needs to be multiplied by the number of load
11295 balancers and different backends that use this health check to get the
11296 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011297
Willy Tarreau81f5d942013-12-09 20:51:51 +010011298 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011299 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011300
Willy Tarreau81f5d942013-12-09 20:51:51 +010011301 - The word "drain". This will turn the server's administrative state to the
11302 DRAIN mode, thus it will not accept any new connections other than those
11303 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011304
Willy Tarreau81f5d942013-12-09 20:51:51 +010011305 - The word "maint". This will turn the server's administrative state to the
11306 MAINT mode, thus it will not accept any new connections at all, and health
11307 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011308
Willy Tarreau81f5d942013-12-09 20:51:51 +010011309 - The words "down", "failed", or "stopped", optionally followed by a
11310 description string after a sharp ('#'). All of these mark the server's
11311 operating state as DOWN, but since the word itself is reported on the stats
11312 page, the difference allows an administrator to know if the situation was
11313 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011314 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011315 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011316
Willy Tarreau81f5d942013-12-09 20:51:51 +010011317 - The word "up" sets back the server's operating state as UP if health checks
11318 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011319
Willy Tarreau81f5d942013-12-09 20:51:51 +010011320 Parameters which are not advertised by the agent are not changed. For
11321 example, an agent might be designed to monitor CPU usage and only report a
11322 relative weight and never interact with the operating status. Similarly, an
11323 agent could be designed as an end-user interface with 3 radio buttons
11324 allowing an administrator to change only the administrative state. However,
11325 it is important to consider that only the agent may revert its own actions,
11326 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11327 agent must implement the other equivalent actions to bring the service into
11328 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011329
Simon Horman2f1f9552013-11-25 10:46:37 +090011330 Failure to connect to the agent is not considered an error as connectivity
11331 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011332 parameter. Warning though, it is not a good idea to stop an agent after it
11333 reports "down", since only an agent reporting "up" will be able to turn the
11334 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011335 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011336
Willy Tarreau81f5d942013-12-09 20:51:51 +010011337 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011338 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011339
James Brown55f9ff12015-10-21 18:19:05 -070011340agent-send <string>
11341 If this option is specified, haproxy will send the given string (verbatim)
11342 to the agent server upon connection. You could, for example, encode
11343 the backend name into this string, which would enable your agent to send
11344 different responses based on the backend. Make sure to include a '\n' if
11345 you want to terminate your request with a newline.
11346
Simon Hormand60d6912013-11-25 10:46:36 +090011347agent-inter <delay>
11348 The "agent-inter" parameter sets the interval between two agent checks
11349 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11350
11351 Just as with every other time-based parameter, it may be entered in any
11352 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11353 parameter also serves as a timeout for agent checks "timeout check" is
11354 not set. In order to reduce "resonance" effects when multiple servers are
11355 hosted on the same hardware, the agent and health checks of all servers
11356 are started with a small time offset between them. It is also possible to
11357 add some random noise in the agent and health checks interval using the
11358 global "spread-checks" keyword. This makes sense for instance when a lot
11359 of backends use the same servers.
11360
11361 See also the "agent-check" and "agent-port" parameters.
11362
Misiek768d8602017-01-09 09:52:43 +010011363agent-addr <addr>
11364 The "agent-addr" parameter sets address for agent check.
11365
11366 You can offload agent-check to another target, so you can make single place
11367 managing status and weights of servers defined in haproxy in case you can't
11368 make self-aware and self-managing services. You can specify both IP or
11369 hostname, it will be resolved.
11370
Simon Hormand60d6912013-11-25 10:46:36 +090011371agent-port <port>
11372 The "agent-port" parameter sets the TCP port used for agent checks.
11373
11374 See also the "agent-check" and "agent-inter" parameters.
11375
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020011376allow-0rtt
11377 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020011378 Note that early data will be sent only if the client used early data, or
11379 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020011380
Olivier Houchardc7566002018-11-20 23:33:50 +010011381alpn <protocols>
11382 This enables the TLS ALPN extension and advertises the specified protocol
11383 list as supported on top of ALPN. The protocol list consists in a comma-
11384 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050011385 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010011386 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
11387 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
11388 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11389 now obsolete NPN extension.
11390 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
11391 be advertised, in order of preference, like below :
11392
11393 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
11394
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011395backup
11396 When "backup" is present on a server line, the server is only used in load
11397 balancing when all other non-backup servers are unavailable. Requests coming
11398 with a persistence cookie referencing the server will always be served
11399 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011400 the "allbackups" option is set in the backend. See also the "no-backup" and
11401 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011402
Emeric Brunef42d922012-10-11 16:11:36 +020011403ca-file <cafile>
11404 This setting is only available when support for OpenSSL was built in. It
11405 designates a PEM file from which to load CA certificates used to verify
11406 server's certificate.
11407
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011408check
11409 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011410 always considered available. If "check" is set, the server is available when
11411 accepting periodic TCP connections, to ensure that it is really able to serve
11412 requests. The default address and port to send the tests to are those of the
11413 server, and the default source is the same as the one defined in the
11414 backend. It is possible to change the address using the "addr" parameter, the
11415 port using the "port" parameter, the source address using the "source"
11416 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011417 parameters. The request method is define in the backend using the "httpchk",
11418 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011419 refer to those options and parameters for more information. See also
11420 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011421
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011422check-send-proxy
11423 This option forces emission of a PROXY protocol line with outgoing health
11424 checks, regardless of whether the server uses send-proxy or not for the
11425 normal traffic. By default, the PROXY protocol is enabled for health checks
11426 if it is already enabled for normal traffic and if no "port" nor "addr"
11427 directive is present. However, if such a directive is present, the
11428 "check-send-proxy" option needs to be used to force the use of the
11429 protocol. See also the "send-proxy" option for more information.
11430
Olivier Houchard92150142018-12-21 19:47:01 +010011431check-alpn <protocols>
11432 Defines which protocols to advertise with ALPN. The protocol list consists in
11433 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
11434 (without quotes). If it is not set, the server ALPN is used.
11435
Jérôme Magninae9bb762018-12-09 16:08:26 +010011436check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020011437 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010011438 over SSL. It is only possible to use a string to set <sni>. If you want to
11439 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020011440
Willy Tarreau763a95b2012-10-04 23:15:39 +020011441check-ssl
11442 This option forces encryption of all health checks over SSL, regardless of
11443 whether the server uses SSL or not for the normal traffic. This is generally
11444 used when an explicit "port" or "addr" directive is specified and SSL health
11445 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011446 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011447 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11448 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011449 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011450 See the "ssl" option for more information and "no-check-ssl" to disable
11451 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011452
Alexander Liu2a54bb72019-05-22 19:44:48 +080011453check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050011454 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080011455 default, the health checks won't go through socks tunnel even it was enabled
11456 for normal traffic.
11457
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011458ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011459 This setting is only available when support for OpenSSL was built in. This
11460 option sets the string describing the list of cipher algorithms that is
11461 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011462 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
11463 information and recommendations see e.g.
11464 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
11465 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
11466 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011467
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011468ciphersuites <ciphersuites>
11469 This setting is only available when support for OpenSSL was built in and
11470 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11471 describing the list of cipher algorithms that is negotiated during the TLS
11472 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011473 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
11474 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
11475 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011476
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011477cookie <value>
11478 The "cookie" parameter sets the cookie value assigned to the server to
11479 <value>. This value will be checked in incoming requests, and the first
11480 operational server possessing the same value will be selected. In return, in
11481 cookie insertion or rewrite modes, this value will be assigned to the cookie
11482 sent to the client. There is nothing wrong in having several servers sharing
11483 the same cookie value, and it is in fact somewhat common between normal and
11484 backup servers. See also the "cookie" keyword in backend section.
11485
Emeric Brunef42d922012-10-11 16:11:36 +020011486crl-file <crlfile>
11487 This setting is only available when support for OpenSSL was built in. It
11488 designates a PEM file from which to load certificate revocation list used
11489 to verify server's certificate.
11490
Emeric Bruna7aa3092012-10-26 12:58:00 +020011491crt <cert>
11492 This setting is only available when support for OpenSSL was built in.
11493 It designates a PEM file from which to load both a certificate and the
11494 associated private key. This file can be built by concatenating both PEM
11495 files into one. This certificate will be sent if the server send a client
11496 certificate request.
11497
Willy Tarreau96839092010-03-29 10:02:24 +020011498disabled
11499 The "disabled" keyword starts the server in the "disabled" state. That means
11500 that it is marked down in maintenance mode, and no connection other than the
11501 ones allowed by persist mode will reach it. It is very well suited to setup
11502 new servers, because normal traffic will never reach them, while it is still
11503 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011504 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011505
Frédéric Lécailled2376272017-03-21 18:52:12 +010011506enabled
11507 This option may be used as 'server' setting to reset any 'disabled'
11508 setting which would have been inherited from 'default-server' directive as
11509 default value.
11510 It may also be used as 'default-server' setting to reset any previous
11511 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011512
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011513error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011514 If health observing is enabled, the "error-limit" parameter specifies the
11515 number of consecutive errors that triggers event selected by the "on-error"
11516 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011517
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011518 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011519
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011520fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011521 The "fall" parameter states that a server will be considered as dead after
11522 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11523 unspecified. See also the "check", "inter" and "rise" parameters.
11524
Emeric Brun8694b9a2012-10-05 14:39:07 +020011525force-sslv3
11526 This option enforces use of SSLv3 only when SSL is used to communicate with
11527 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011528 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011529 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011530
11531force-tlsv10
11532 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011533 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011534 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011535
11536force-tlsv11
11537 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011538 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011539 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011540
11541force-tlsv12
11542 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011543 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011544 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011545
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011546force-tlsv13
11547 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11548 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011549 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011550
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011551id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011552 Set a persistent ID for the server. This ID must be positive and unique for
11553 the proxy. An unused ID will automatically be assigned if unset. The first
11554 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011555
Willy Tarreau6a031d12016-11-07 19:42:35 +010011556init-addr {last | libc | none | <ip>},[...]*
11557 Indicate in what order the server's address should be resolved upon startup
11558 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011559 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011560 method which succeeds is used. If the end of the list is reached without
11561 finding a working method, an error is thrown. Method "last" suggests to pick
11562 the address which appears in the state file (see "server-state-file"). Method
11563 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11564 depending on the operating system and build options). Method "none"
11565 specifically indicates that the server should start without any valid IP
11566 address in a down state. It can be useful to ignore some DNS issues upon
11567 startup, waiting for the situation to get fixed later. Finally, an IP address
11568 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011569 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011570 server used to catch old sessions and present them with a decent error
11571 message for example. When the "first" load balancing algorithm is used, this
11572 IP address could point to a fake server used to trigger the creation of new
11573 instances on the fly. This option defaults to "last,libc" indicating that the
11574 previous address found in the state file (if any) is used first, otherwise
11575 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011576 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011577
11578 Example:
11579 defaults
11580 # never fail on address resolution
11581 default-server init-addr last,libc,none
11582
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011583inter <delay>
11584fastinter <delay>
11585downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011586 The "inter" parameter sets the interval between two consecutive health checks
11587 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11588 It is also possible to use "fastinter" and "downinter" to optimize delays
11589 between checks depending on the server state :
11590
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011591 Server state | Interval used
11592 ----------------------------------------+----------------------------------
11593 UP 100% (non-transitional) | "inter"
11594 ----------------------------------------+----------------------------------
11595 Transitionally UP (going down "fall"), | "fastinter" if set,
11596 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11597 or yet unchecked. |
11598 ----------------------------------------+----------------------------------
11599 DOWN 100% (non-transitional) | "downinter" if set,
11600 | "inter" otherwise.
11601 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011602
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011603 Just as with every other time-based parameter, they can be entered in any
11604 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11605 serves as a timeout for health checks sent to servers if "timeout check" is
11606 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011607 hosted on the same hardware, the agent and health checks of all servers
11608 are started with a small time offset between them. It is also possible to
11609 add some random noise in the agent and health checks interval using the
11610 global "spread-checks" keyword. This makes sense for instance when a lot
11611 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011612
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011613maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011614 The "maxconn" parameter specifies the maximal number of concurrent
11615 connections that will be sent to this server. If the number of incoming
11616 concurrent requests goes higher than this value, they will be queued, waiting
11617 for a connection to be released. This parameter is very important as it can
11618 save fragile servers from going down under extreme loads. If a "minconn"
11619 parameter is specified, the limit becomes dynamic. The default value is "0"
11620 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11621 the backend's "fullconn" keyword.
11622
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011623maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011624 The "maxqueue" parameter specifies the maximal number of connections which
11625 will wait in the queue for this server. If this limit is reached, next
11626 requests will be redispatched to other servers instead of indefinitely
11627 waiting to be served. This will break persistence but may allow people to
11628 quickly re-log in when the server they try to connect to is dying. The
11629 default value is "0" which means the queue is unlimited. See also the
11630 "maxconn" and "minconn" parameters.
11631
Willy Tarreau9c538e02019-01-23 10:21:49 +010011632max-reuse <count>
11633 The "max-reuse" argument indicates the HTTP connection processors that they
11634 should not reuse a server connection more than this number of times to send
11635 new requests. Permitted values are -1 (the default), which disables this
11636 limit, or any positive value. Value zero will effectively disable keep-alive.
11637 This is only used to work around certain server bugs which cause them to leak
11638 resources over time. The argument is not necessarily respected by the lower
11639 layers as there might be technical limitations making it impossible to
11640 enforce. At least HTTP/2 connections to servers will respect it.
11641
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011642minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011643 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11644 limit following the backend's load. The server will always accept at least
11645 <minconn> connections, never more than <maxconn>, and the limit will be on
11646 the ramp between both values when the backend has less than <fullconn>
11647 concurrent connections. This makes it possible to limit the load on the
11648 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011649 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011650 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011651
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011652namespace <name>
11653 On Linux, it is possible to specify which network namespace a socket will
11654 belong to. This directive makes it possible to explicitly bind a server to
11655 a namespace different from the default one. Please refer to your operating
11656 system's documentation to find more details about network namespaces.
11657
Frédéric Lécailled2376272017-03-21 18:52:12 +010011658no-agent-check
11659 This option may be used as "server" setting to reset any "agent-check"
11660 setting which would have been inherited from "default-server" directive as
11661 default value.
11662 It may also be used as "default-server" setting to reset any previous
11663 "default-server" "agent-check" setting.
11664
11665no-backup
11666 This option may be used as "server" setting to reset any "backup"
11667 setting which would have been inherited from "default-server" directive as
11668 default value.
11669 It may also be used as "default-server" setting to reset any previous
11670 "default-server" "backup" setting.
11671
11672no-check
11673 This option may be used as "server" setting to reset any "check"
11674 setting which would have been inherited from "default-server" directive as
11675 default value.
11676 It may also be used as "default-server" setting to reset any previous
11677 "default-server" "check" setting.
11678
11679no-check-ssl
11680 This option may be used as "server" setting to reset any "check-ssl"
11681 setting which would have been inherited from "default-server" directive as
11682 default value.
11683 It may also be used as "default-server" setting to reset any previous
11684 "default-server" "check-ssl" setting.
11685
Frédéric Lécailled2376272017-03-21 18:52:12 +010011686no-send-proxy
11687 This option may be used as "server" setting to reset any "send-proxy"
11688 setting which would have been inherited from "default-server" directive as
11689 default value.
11690 It may also be used as "default-server" setting to reset any previous
11691 "default-server" "send-proxy" setting.
11692
11693no-send-proxy-v2
11694 This option may be used as "server" setting to reset any "send-proxy-v2"
11695 setting which would have been inherited from "default-server" directive as
11696 default value.
11697 It may also be used as "default-server" setting to reset any previous
11698 "default-server" "send-proxy-v2" setting.
11699
11700no-send-proxy-v2-ssl
11701 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11702 setting which would have been inherited from "default-server" directive as
11703 default value.
11704 It may also be used as "default-server" setting to reset any previous
11705 "default-server" "send-proxy-v2-ssl" setting.
11706
11707no-send-proxy-v2-ssl-cn
11708 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11709 setting which would have been inherited from "default-server" directive as
11710 default value.
11711 It may also be used as "default-server" setting to reset any previous
11712 "default-server" "send-proxy-v2-ssl-cn" setting.
11713
11714no-ssl
11715 This option may be used as "server" setting to reset any "ssl"
11716 setting which would have been inherited from "default-server" directive as
11717 default value.
11718 It may also be used as "default-server" setting to reset any previous
11719 "default-server" "ssl" setting.
11720
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011721no-ssl-reuse
11722 This option disables SSL session reuse when SSL is used to communicate with
11723 the server. It will force the server to perform a full handshake for every
11724 new connection. It's probably only useful for benchmarking, troubleshooting,
11725 and for paranoid users.
11726
Emeric Brun9b3009b2012-10-05 11:55:06 +020011727no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011728 This option disables support for SSLv3 when SSL is used to communicate with
11729 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011730 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011731
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011732 Supported in default-server: No
11733
Emeric Brunf9c5c472012-10-11 15:28:34 +020011734no-tls-tickets
11735 This setting is only available when support for OpenSSL was built in. It
11736 disables the stateless session resumption (RFC 5077 TLS Ticket
11737 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011738 session resumption is more expensive in CPU usage for servers. This option
11739 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011740 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011741
Emeric Brun9b3009b2012-10-05 11:55:06 +020011742no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011743 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011744 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11745 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011746 often makes sense to disable it when communicating with local servers. This
11747 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011748 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011749
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011750 Supported in default-server: No
11751
Emeric Brun9b3009b2012-10-05 11:55:06 +020011752no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011753 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011754 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11755 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011756 often makes sense to disable it when communicating with local servers. This
11757 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011758 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011759
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011760 Supported in default-server: No
11761
Emeric Brun9b3009b2012-10-05 11:55:06 +020011762no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011763 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011764 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11765 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011766 often makes sense to disable it when communicating with local servers. This
11767 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011768 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011769
11770 Supported in default-server: No
11771
11772no-tlsv13
11773 This option disables support for TLSv1.3 when SSL is used to communicate with
11774 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11775 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11776 often makes sense to disable it when communicating with local servers. This
11777 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011778 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011779
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011780 Supported in default-server: No
11781
Frédéric Lécailled2376272017-03-21 18:52:12 +010011782no-verifyhost
11783 This option may be used as "server" setting to reset any "verifyhost"
11784 setting which would have been inherited from "default-server" directive as
11785 default value.
11786 It may also be used as "default-server" setting to reset any previous
11787 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011788
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020011789no-tfo
11790 This option may be used as "server" setting to reset any "tfo"
11791 setting which would have been inherited from "default-server" directive as
11792 default value.
11793 It may also be used as "default-server" setting to reset any previous
11794 "default-server" "tfo" setting.
11795
Simon Hormanfa461682011-06-25 09:39:49 +090011796non-stick
11797 Never add connections allocated to this sever to a stick-table.
11798 This may be used in conjunction with backup to ensure that
11799 stick-table persistence is disabled for backup servers.
11800
Olivier Houchardc7566002018-11-20 23:33:50 +010011801npn <protocols>
11802 This enables the NPN TLS extension and advertises the specified protocol list
11803 as supported on top of NPN. The protocol list consists in a comma-delimited
11804 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050011805 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010011806 enabled (check with haproxy -vv). Note that the NPN extension has been
11807 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11808 only available starting with OpenSSL 1.0.2.
11809
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011810observe <mode>
11811 This option enables health adjusting based on observing communication with
11812 the server. By default this functionality is disabled and enabling it also
11813 requires to enable health checks. There are two supported modes: "layer4" and
11814 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11815 significant. In layer7, which is only allowed for http proxies, responses
11816 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011817 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011818
11819 See also the "check", "on-error" and "error-limit".
11820
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011821on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011822 Select what should happen when enough consecutive errors are detected.
11823 Currently, four modes are available:
11824 - fastinter: force fastinter
11825 - fail-check: simulate a failed check, also forces fastinter (default)
11826 - sudden-death: simulate a pre-fatal failed health check, one more failed
11827 check will mark a server down, forces fastinter
11828 - mark-down: mark the server immediately down and force fastinter
11829
11830 See also the "check", "observe" and "error-limit".
11831
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011832on-marked-down <action>
11833 Modify what occurs when a server is marked down.
11834 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011835 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11836 all connections to the server are immediately terminated when the server
11837 goes down. It might be used if the health check detects more complex cases
11838 than a simple connection status, and long timeouts would cause the service
11839 to remain unresponsive for too long a time. For instance, a health check
11840 might detect that a database is stuck and that there's no chance to reuse
11841 existing connections anymore. Connections killed this way are logged with
11842 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011843
11844 Actions are disabled by default
11845
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011846on-marked-up <action>
11847 Modify what occurs when a server is marked up.
11848 Currently one action is available:
11849 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11850 done only if the server is not in backup state and if it is not disabled
11851 (it must have an effective weight > 0). This can be used sometimes to force
11852 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011853 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11854 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011855 with extreme care. Sessions killed because a server comes up are logged
11856 with an 'U' termination code (for "Up").
11857
11858 Actions are disabled by default
11859
Olivier Houchard006e3102018-12-10 18:30:32 +010011860pool-max-conn <max>
11861 Set the maximum number of idling connections for a server. -1 means unlimited
11862 connections, 0 means no idle connections. The default is -1. When idle
11863 connections are enabled, orphaned idle connections which do not belong to any
11864 client session anymore are moved to a dedicated pool so that they remain
11865 usable by future clients. This only applies to connections that can be shared
11866 according to the same principles as those applying to "http-reuse".
11867
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010011868pool-purge-delay <delay>
11869 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010011870 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020011871 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010011872
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011873port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011874 Using the "port" parameter, it becomes possible to use a different port to
11875 send health-checks. On some servers, it may be desirable to dedicate a port
11876 to a specific component able to perform complex tests which are more suitable
11877 to health-checks than the application. It is common to run a simple script in
11878 inetd for instance. This parameter is ignored if the "check" parameter is not
11879 set. See also the "addr" parameter.
11880
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020011881proto <name>
11882
11883 Forces the multiplexer's protocol to use for the outgoing connections to this
11884 server. It must be compatible with the mode of the backend (TCP or HTTP). It
11885 must also be usable on the backend side. The list of available protocols is
11886 reported in haproxy -vv.
11887 Idea behind this optipon is to bypass the selection of the best multiplexer's
11888 protocol for all connections established to this server.
11889
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011890redir <prefix>
11891 The "redir" parameter enables the redirection mode for all GET and HEAD
11892 requests addressing this server. This means that instead of having HAProxy
11893 forward the request to the server, it will send an "HTTP 302" response with
11894 the "Location" header composed of this prefix immediately followed by the
11895 requested URI beginning at the leading '/' of the path component. That means
11896 that no trailing slash should be used after <prefix>. All invalid requests
11897 will be rejected, and all non-GET or HEAD requests will be normally served by
11898 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011899 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011900 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011901 users to a remote location in case of local disaster. Main use consists in
11902 increasing bandwidth for static servers by having the clients directly
11903 connect to them. Note: never use a relative location here, it would cause a
11904 loop between the client and HAProxy!
11905
11906 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
11907
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011908rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011909 The "rise" parameter states that a server will be considered as operational
11910 after <count> consecutive successful health checks. This value defaults to 2
11911 if unspecified. See also the "check", "inter" and "fall" parameters.
11912
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020011913resolve-opts <option>,<option>,...
11914 Comma separated list of options to apply to DNS resolution linked to this
11915 server.
11916
11917 Available options:
11918
11919 * allow-dup-ip
11920 By default, HAProxy prevents IP address duplication in a backend when DNS
11921 resolution at runtime is in operation.
11922 That said, for some cases, it makes sense that two servers (in the same
11923 backend, being resolved by the same FQDN) have the same IP address.
11924 For such case, simply enable this option.
11925 This is the opposite of prevent-dup-ip.
11926
11927 * prevent-dup-ip
11928 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
11929 an IP address already set to a server in the same backend and sharing the
11930 same fqdn.
11931 This is the opposite of allow-dup-ip.
11932
11933 Example:
11934 backend b_myapp
11935 default-server init-addr none resolvers dns
11936 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
11937 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
11938
11939 With the option allow-dup-ip set:
11940 * if the nameserver returns a single IP address, then both servers will use
11941 it
11942 * If the nameserver returns 2 IP addresses, then each server will pick up a
11943 different address
11944
11945 Default value: not set
11946
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011947resolve-prefer <family>
11948 When DNS resolution is enabled for a server and multiple IP addresses from
11949 different families are returned, HAProxy will prefer using an IP address
11950 from the family mentioned in the "resolve-prefer" parameter.
11951 Available families: "ipv4" and "ipv6"
11952
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020011953 Default value: ipv6
11954
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011955 Example:
11956
11957 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011958
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011959resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050011960 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011961 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010011962 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010011963 different datacenters. The latency between datacenter is not negligible, so
11964 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011965 configured network, another address is selected.
11966
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011967 Example:
11968
11969 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011970
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011971resolvers <id>
11972 Points to an existing "resolvers" section to resolve current server's
11973 hostname.
11974
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011975 Example:
11976
11977 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011978
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011979 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011980
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011981send-proxy
11982 The "send-proxy" parameter enforces use of the PROXY protocol over any
11983 connection established to this server. The PROXY protocol informs the other
11984 end about the layer 3/4 addresses of the incoming connection, so that it can
11985 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011986 upper layer protocol. For connections accepted by an "accept-proxy" or
11987 "accept-netscaler-cip" listener, the advertised address will be used. Only
11988 TCPv4 and TCPv6 address families are supported. Other families such as
11989 Unix sockets, will report an UNKNOWN family. Servers using this option can
11990 fully be chained to another instance of haproxy listening with an
11991 "accept-proxy" setting. This setting must not be used if the server isn't
11992 aware of the protocol. When health checks are sent to the server, the PROXY
11993 protocol is automatically used when this option is set, unless there is an
11994 explicit "port" or "addr" directive, in which case an explicit
11995 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011996 See also the "no-send-proxy" option of this section and "accept-proxy" and
11997 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011998
David Safb76832014-05-08 23:42:08 -040011999send-proxy-v2
12000 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12001 over any connection established to this server. The PROXY protocol informs
12002 the other end about the layer 3/4 addresses of the incoming connection, so
12003 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012004 whatever the upper layer protocol. It also send ALPN information if an alpn
12005 have been negotiated. This setting must not be used if the server isn't aware
12006 of this version of the protocol. See also the "no-send-proxy-v2" option of
12007 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012008
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012009proxy-v2-options <option>[,<option>]*
12010 The "proxy-v2-options" parameter add option to send in PROXY protocol version
12011 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010012012 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
12013 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010012014 certificate, "cert-key": key algorithm of the used certificate), "authority":
12015 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010012016 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012017
David Safb76832014-05-08 23:42:08 -040012018send-proxy-v2-ssl
12019 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12020 2 over any connection established to this server. The PROXY protocol informs
12021 the other end about the layer 3/4 addresses of the incoming connection, so
12022 that it can know the client's address or the public address it accessed to,
12023 whatever the upper layer protocol. In addition, the SSL information extension
12024 of the PROXY protocol is added to the PROXY protocol header. This setting
12025 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012026 See also the "no-send-proxy-v2-ssl" option of this section and the
12027 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012028
12029send-proxy-v2-ssl-cn
12030 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12031 2 over any connection established to this server. The PROXY protocol informs
12032 the other end about the layer 3/4 addresses of the incoming connection, so
12033 that it can know the client's address or the public address it accessed to,
12034 whatever the upper layer protocol. In addition, the SSL information extension
12035 of the PROXY protocol, along along with the Common Name from the subject of
12036 the client certificate (if any), is added to the PROXY protocol header. This
12037 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012038 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12039 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012040
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012041slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012042 The "slowstart" parameter for a server accepts a value in milliseconds which
12043 indicates after how long a server which has just come back up will run at
12044 full speed. Just as with every other time-based parameter, it can be entered
12045 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12046 linearly from 0 to 100% during this time. The limitation applies to two
12047 parameters :
12048
12049 - maxconn: the number of connections accepted by the server will grow from 1
12050 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12051
12052 - weight: when the backend uses a dynamic weighted algorithm, the weight
12053 grows linearly from 1 to 100%. In this case, the weight is updated at every
12054 health-check. For this reason, it is important that the "inter" parameter
12055 is smaller than the "slowstart", in order to maximize the number of steps.
12056
12057 The slowstart never applies when haproxy starts, otherwise it would cause
12058 trouble to running servers. It only applies when a server has been previously
12059 seen as failed.
12060
Willy Tarreau732eac42015-07-09 11:40:25 +020012061sni <expression>
12062 The "sni" parameter evaluates the sample fetch expression, converts it to a
12063 string and uses the result as the host name sent in the SNI TLS extension to
12064 the server. A typical use case is to send the SNI received from the client in
12065 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012066 expression, though alternatives such as req.hdr(host) can also make sense. If
12067 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012068 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012069 "verify" directive for more details. If you want to set a SNI for health
12070 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012071
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012072source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012073source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012074source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012075 The "source" parameter sets the source address which will be used when
12076 connecting to the server. It follows the exact same parameters and principle
12077 as the backend "source" keyword, except that it only applies to the server
12078 referencing it. Please consult the "source" keyword for details.
12079
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012080 Additionally, the "source" statement on a server line allows one to specify a
12081 source port range by indicating the lower and higher bounds delimited by a
12082 dash ('-'). Some operating systems might require a valid IP address when a
12083 source port range is specified. It is permitted to have the same IP/range for
12084 several servers. Doing so makes it possible to bypass the maximum of 64k
12085 total concurrent connections. The limit will then reach 64k connections per
12086 server.
12087
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012088 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12089 specifying the source address without port(s).
12090
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012091ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012092 This option enables SSL ciphering on outgoing connections to the server. It
12093 is critical to verify server certificates using "verify" when using SSL to
12094 connect to servers, otherwise the communication is prone to trivial man in
12095 the-middle attacks rendering SSL useless. When this option is used, health
12096 checks are automatically sent in SSL too unless there is a "port" or an
12097 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012098 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12099 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012100
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012101ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12102 This option enforces use of <version> or lower when SSL is used to communicate
12103 with the server. This option is also available on global statement
12104 "ssl-default-server-options". See also "ssl-min-ver".
12105
12106ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12107 This option enforces use of <version> or upper when SSL is used to communicate
12108 with the server. This option is also available on global statement
12109 "ssl-default-server-options". See also "ssl-max-ver".
12110
Frédéric Lécailled2376272017-03-21 18:52:12 +010012111ssl-reuse
12112 This option may be used as "server" setting to reset any "no-ssl-reuse"
12113 setting which would have been inherited from "default-server" directive as
12114 default value.
12115 It may also be used as "default-server" setting to reset any previous
12116 "default-server" "no-ssl-reuse" setting.
12117
12118stick
12119 This option may be used as "server" setting to reset any "non-stick"
12120 setting which would have been inherited from "default-server" directive as
12121 default value.
12122 It may also be used as "default-server" setting to reset any previous
12123 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012124
Alexander Liu2a54bb72019-05-22 19:44:48 +080012125socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050012126 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080012127 server. Using this option won't force the health check to go via socks4 by
12128 default. You will have to use the keyword "check-via-socks4" to enable it.
12129
Willy Tarreau163d4622015-10-13 16:16:41 +020012130tcp-ut <delay>
12131 Sets the TCP User Timeout for all outgoing connections to this server. This
12132 option is available on Linux since version 2.6.37. It allows haproxy to
12133 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012134 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012135 long-lived connections experiencing long idle periods such as remote
12136 terminals or database connection pools, where the client and server timeouts
12137 must remain high to allow a long period of idle, but where it is important to
12138 detect that the server has disappeared in order to release all resources
12139 associated with its connection (and the client's session). One typical use
12140 case is also to force dead server connections to die when health checks are
12141 too slow or during a soft reload since health checks are then disabled. The
12142 argument is a delay expressed in milliseconds by default. This only works for
12143 regular TCP connections, and is ignored for other protocols.
12144
Willy Tarreau034c88c2017-01-23 23:36:45 +010012145tfo
12146 This option enables using TCP fast open when connecting to servers, on
12147 systems that support it (currently only the Linux kernel >= 4.11).
12148 See the "tfo" bind option for more information about TCP fast open.
12149 Please note that when using tfo, you should also use the "conn-failure",
12150 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020012151 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010012152
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012153track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012154 This option enables ability to set the current state of the server by tracking
12155 another one. It is possible to track a server which itself tracks another
12156 server, provided that at the end of the chain, a server has health checks
12157 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012158 used, it has to be enabled on both proxies.
12159
Frédéric Lécailled2376272017-03-21 18:52:12 +010012160tls-tickets
12161 This option may be used as "server" setting to reset any "no-tls-tickets"
12162 setting which would have been inherited from "default-server" directive as
12163 default value.
12164 It may also be used as "default-server" setting to reset any previous
12165 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012166
Emeric Brunef42d922012-10-11 16:11:36 +020012167verify [none|required]
12168 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012169 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012170 certificate provided by the server is verified using CAs from 'ca-file' and
12171 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012172 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012173 the name passed using the "sni" directive, or if not provided, the static
12174 host name passed using the "verifyhost" directive. When no name is found, the
12175 certificate's names are ignored. For this reason, without SNI it's important
12176 to use "verifyhost". On verification failure the handshake is aborted. It is
12177 critically important to verify server certificates when using SSL to connect
12178 to servers, otherwise the communication is prone to trivial man-in-the-middle
12179 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12180 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012181
Evan Broderbe554312013-06-27 00:05:25 -070012182verifyhost <hostname>
12183 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012184 only takes effect if 'verify required' is also specified. This directive sets
12185 a default static hostname to check the server's certificate against when no
12186 SNI was used to connect to the server. If SNI is not used, this is the only
12187 way to enable hostname verification. This static hostname, when set, will
12188 also be used for health checks (which cannot provide an SNI value). If none
12189 of the hostnames in the certificate match the specified hostname, the
12190 handshake is aborted. The hostnames in the server-provided certificate may
12191 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012192
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012193weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012194 The "weight" parameter is used to adjust the server's weight relative to
12195 other servers. All servers will receive a load proportional to their weight
12196 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012197 load. The default weight is 1, and the maximal value is 256. A value of 0
12198 means the server will not participate in load-balancing but will still accept
12199 persistent connections. If this parameter is used to distribute the load
12200 according to server's capacity, it is recommended to start with values which
12201 can both grow and shrink, for instance between 10 and 100 to leave enough
12202 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012203
12204
Cyril Bonté46175dd2015-07-02 22:45:32 +0200122055.3. Server IP address resolution using DNS
12206-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012207
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012208HAProxy allows using a host name on the server line to retrieve its IP address
12209using name servers. By default, HAProxy resolves the name when parsing the
12210configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012211This is not sufficient in some cases, such as in Amazon where a server's IP
12212can change after a reboot or an ELB Virtual IP can change based on current
12213workload.
12214This chapter describes how HAProxy can be configured to process server's name
12215resolution at run time.
12216Whether run time server name resolution has been enable or not, HAProxy will
12217carry on doing the first resolution when parsing the configuration.
12218
12219
Cyril Bonté46175dd2015-07-02 22:45:32 +0200122205.3.1. Global overview
12221----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012222
12223As we've seen in introduction, name resolution in HAProxy occurs at two
12224different steps of the process life:
12225
12226 1. when starting up, HAProxy parses the server line definition and matches a
12227 host name. It uses libc functions to get the host name resolved. This
12228 resolution relies on /etc/resolv.conf file.
12229
Christopher Faulet67957bd2017-09-27 11:00:59 +020012230 2. at run time, HAProxy performs periodically name resolutions for servers
12231 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012232
12233A few other events can trigger a name resolution at run time:
12234 - when a server's health check ends up in a connection timeout: this may be
12235 because the server has a new IP address. So we need to trigger a name
12236 resolution to know this new IP.
12237
Christopher Faulet67957bd2017-09-27 11:00:59 +020012238When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012239HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012240SRV label is specified, then the corresponding SRV records will be retrieved
12241from the DNS server, and the provided hostnames will be used. The SRV label
12242will be checked periodically, and if any server are added or removed, haproxy
12243will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012244
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012245A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050012246 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012247 first valid response.
12248
12249 - a resolution is considered as invalid (NX, timeout, refused), when all the
12250 servers return an error.
12251
12252
Cyril Bonté46175dd2015-07-02 22:45:32 +0200122535.3.2. The resolvers section
12254----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012255
12256This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012257HAProxy. There can be as many as resolvers section as needed. Each section can
12258contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012259
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012260When multiple name servers are configured in a resolvers section, then HAProxy
12261uses the first valid response. In case of invalid responses, only the last one
12262is treated. Purpose is to give the chance to a slow server to deliver a valid
12263answer after a fast faulty or outdated server.
12264
12265When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012266used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012267
Christopher Faulet67957bd2017-09-27 11:00:59 +020012268 1. HAProxy retries the same DNS query with a new query type. The A queries are
12269 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12270 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012271
Christopher Faulet67957bd2017-09-27 11:00:59 +020012272 2. When the fallback on the query type was done (or not applicable), HAProxy
12273 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012274
Christopher Faulet67957bd2017-09-27 11:00:59 +020012275 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12276 response is received after that, it stops the DNS resolution and reports
12277 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012278
Christopher Faulet67957bd2017-09-27 11:00:59 +020012279For example, with 2 name servers configured in a resolvers section, the
12280following scenarios are possible:
12281
12282 - First response is valid and is applied directly, second response is
12283 ignored
12284
12285 - First response is invalid and second one is valid, then second response is
12286 applied
12287
12288 - First response is a NX domain and second one a truncated response, then
12289 HAProxy retries the query with a new type
12290
12291 - First response is a NX domain and second one is a timeout, then HAProxy
12292 retries the query with a new type
12293
12294 - Query timed out for both name servers, then HAProxy retries it with the
12295 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012296
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012297As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12298a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012299<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012300
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012301
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012302resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012303 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012304
12305A resolvers section accept the following parameters:
12306
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012307accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012308 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012309 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012310 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12311 by RFC 6891)
12312
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012313 Note: the maximum allowed value is 8192.
12314
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012315nameserver <id> <ip>:<port>
12316 DNS server description:
12317 <id> : label of the server, should be unique
12318 <ip> : IP address of the server
12319 <port> : port where the DNS service actually runs
12320
Ben Draut44e609b2018-05-29 15:40:08 -060012321parse-resolv-conf
12322 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12323 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12324 placed in the resolvers section in place of this directive.
12325
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012326hold <status> <period>
12327 Defines <period> during which the last name resolution should be kept based
12328 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012329 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012330 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012331 <period> : interval between two successive name resolution when the last
12332 answer was in <status>. It follows the HAProxy time format.
12333 <period> is in milliseconds by default.
12334
Baptiste Assmann686408b2017-08-18 10:15:42 +020012335 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012336
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012337resolve_retries <nb>
12338 Defines the number <nb> of queries to send to resolve a server name before
12339 giving up.
12340 Default value: 3
12341
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012342 A retry occurs on name server timeout or when the full sequence of DNS query
12343 type failover is over and we need to start up from the default ANY query
12344 type.
12345
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012346timeout <event> <time>
12347 Defines timeouts related to name resolution
12348 <event> : the event on which the <time> timeout period applies to.
12349 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012350 - resolve : default time to trigger name resolutions when no
12351 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012352 Default value: 1s
12353 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012354 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012355 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012356 <time> : time related to the event. It follows the HAProxy time format.
12357 <time> is expressed in milliseconds.
12358
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012359 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012360
12361 resolvers mydns
12362 nameserver dns1 10.0.0.1:53
12363 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012364 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012365 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012366 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012367 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012368 hold other 30s
12369 hold refused 30s
12370 hold nx 30s
12371 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012372 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012373 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012374
12375
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200123766. Cache
12377---------
12378
12379HAProxy provides a cache, which was designed to perform cache on small objects
12380(favicon, css...). This is a minimalist low-maintenance cache which runs in
12381RAM.
12382
12383The cache is based on a memory which is shared between processes and threads,
12384this memory is split in blocks of 1k.
12385
12386If an object is not used anymore, it can be deleted to store a new object
12387independently of its expiration date. The oldest objects are deleted first
12388when we try to allocate a new one.
12389
12390The cache uses a hash of the host header and the URI as the key.
12391
12392It's possible to view the status of a cache using the Unix socket command
12393"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
12394for more details.
12395
12396When an object is delivered from the cache, the server name in the log is
12397replaced by "<CACHE>".
12398
12399
124006.1. Limitation
12401----------------
12402
12403The cache won't store and won't deliver objects in these cases:
12404
12405- If the response is not a 200
12406- If the response contains a Vary header
12407- If the Content-Length + the headers size is greater than "max-object-size"
12408- If the response is not cacheable
12409
12410- If the request is not a GET
12411- If the HTTP version of the request is smaller than 1.1
12412- If the request contains an Authorization header
12413
12414
124156.2. Setup
12416-----------
12417
12418To setup a cache, you must define a cache section and use it in a proxy with
12419the corresponding http-request and response actions.
12420
12421
124226.2.1. Cache section
12423---------------------
12424
12425cache <name>
12426 Declare a cache section, allocate a shared cache memory named <name>, the
12427 size of cache is mandatory.
12428
12429total-max-size <megabytes>
12430 Define the size in RAM of the cache in megabytes. This size is split in
12431 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
12432
12433max-object-size <bytes>
12434 Define the maximum size of the objects to be cached. Must not be greater than
12435 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
12436 All objects with sizes larger than "max-object-size" will not be cached.
12437
12438max-age <seconds>
12439 Define the maximum expiration duration. The expiration is set has the lowest
12440 value between the s-maxage or max-age (in this order) directive in the
12441 Cache-Control response header and this value. The default value is 60
12442 seconds, which means that you can't cache an object more than 60 seconds by
12443 default.
12444
12445
124466.2.2. Proxy section
12447---------------------
12448
12449http-request cache-use <name> [ { if | unless } <condition> ]
12450 Try to deliver a cached object from the cache <name>. This directive is also
12451 mandatory to store the cache as it calculates the cache hash. If you want to
12452 use a condition for both storage and delivering that's a good idea to put it
12453 after this one.
12454
12455http-response cache-store <name> [ { if | unless } <condition> ]
12456 Store an http-response within the cache. The storage of the response headers
12457 is done at this step, which means you can use others http-response actions
12458 to modify headers before or after the storage of the response. This action
12459 is responsible for the setup of the cache storage filter.
12460
12461
12462Example:
12463
12464 backend bck1
12465 mode http
12466
12467 http-request cache-use foobar
12468 http-response cache-store foobar
12469 server srv1 127.0.0.1:80
12470
12471 cache foobar
12472 total-max-size 4
12473 max-age 240
12474
12475
Willy Tarreau74ca5042013-06-11 23:12:07 +0200124767. Using ACLs and fetching samples
12477----------------------------------
12478
Davor Ocelice9ed2812017-12-25 17:49:28 +010012479HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012480client or server information, from tables, environmental information etc...
12481The action of extracting such data is called fetching a sample. Once retrieved,
12482these samples may be used for various purposes such as a key to a stick-table,
12483but most common usages consist in matching them against predefined constant
12484data called patterns.
12485
12486
124877.1. ACL basics
12488---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012489
12490The use of Access Control Lists (ACL) provides a flexible solution to perform
12491content switching and generally to take decisions based on content extracted
12492from the request, the response or any environmental status. The principle is
12493simple :
12494
Willy Tarreau74ca5042013-06-11 23:12:07 +020012495 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012496 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012497 - apply one or multiple pattern matching methods on this sample
12498 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012499
Willy Tarreau74ca5042013-06-11 23:12:07 +020012500The actions generally consist in blocking a request, selecting a backend, or
12501adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012502
12503In order to define a test, the "acl" keyword is used. The syntax is :
12504
Willy Tarreau74ca5042013-06-11 23:12:07 +020012505 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012506
12507This creates a new ACL <aclname> or completes an existing one with new tests.
12508Those tests apply to the portion of request/response specified in <criterion>
12509and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012510an operator which may be specified before the set of values. Optionally some
12511conversion operators may be applied to the sample, and they will be specified
12512as a comma-delimited list of keywords just after the first keyword. The values
12513are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012514
12515ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12516'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12517which means that "my_acl" and "My_Acl" are two different ACLs.
12518
12519There is no enforced limit to the number of ACLs. The unused ones do not affect
12520performance, they just consume a small amount of memory.
12521
Willy Tarreau74ca5042013-06-11 23:12:07 +020012522The criterion generally is the name of a sample fetch method, or one of its ACL
12523specific declinations. The default test method is implied by the output type of
12524this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012525methods of a same sample fetch method. The sample fetch methods are the only
12526ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012527
12528Sample fetch methods return data which can be of the following types :
12529 - boolean
12530 - integer (signed or unsigned)
12531 - IPv4 or IPv6 address
12532 - string
12533 - data block
12534
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012535Converters transform any of these data into any of these. For example, some
12536converters might convert a string to a lower-case string while other ones
12537would turn a string to an IPv4 address, or apply a netmask to an IP address.
12538The resulting sample is of the type of the last converter applied to the list,
12539which defaults to the type of the sample fetch method.
12540
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012541Each sample or converter returns data of a specific type, specified with its
12542keyword in this documentation. When an ACL is declared using a standard sample
12543fetch method, certain types automatically involved a default matching method
12544which are summarized in the table below :
12545
12546 +---------------------+-----------------+
12547 | Sample or converter | Default |
12548 | output type | matching method |
12549 +---------------------+-----------------+
12550 | boolean | bool |
12551 +---------------------+-----------------+
12552 | integer | int |
12553 +---------------------+-----------------+
12554 | ip | ip |
12555 +---------------------+-----------------+
12556 | string | str |
12557 +---------------------+-----------------+
12558 | binary | none, use "-m" |
12559 +---------------------+-----------------+
12560
12561Note that in order to match a binary samples, it is mandatory to specify a
12562matching method, see below.
12563
Willy Tarreau74ca5042013-06-11 23:12:07 +020012564The ACL engine can match these types against patterns of the following types :
12565 - boolean
12566 - integer or integer range
12567 - IP address / network
12568 - string (exact, substring, suffix, prefix, subdir, domain)
12569 - regular expression
12570 - hex block
12571
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012572The following ACL flags are currently supported :
12573
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012574 -i : ignore case during matching of all subsequent patterns.
12575 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012576 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012577 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012578 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012579 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012580 -- : force end of flags. Useful when a string looks like one of the flags.
12581
Willy Tarreau74ca5042013-06-11 23:12:07 +020012582The "-f" flag is followed by the name of a file from which all lines will be
12583read as individual values. It is even possible to pass multiple "-f" arguments
12584if the patterns are to be loaded from multiple files. Empty lines as well as
12585lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12586will be stripped. If it is absolutely necessary to insert a valid pattern
12587beginning with a sharp, just prefix it with a space so that it is not taken for
12588a comment. Depending on the data type and match method, haproxy may load the
12589lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12590exact string matching. In this case, duplicates will automatically be removed.
12591
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012592The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12593parsed as two column file. The first column contains the patterns used by the
12594ACL, and the second column contain the samples. The sample can be used later by
12595a map. This can be useful in some rare cases where an ACL would just be used to
12596check for the existence of a pattern in a map before a mapping is applied.
12597
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012598The "-u" flag forces the unique id of the ACL. This unique id is used with the
12599socket interface to identify ACL and dynamically change its values. Note that a
12600file is always identified by its name even if an id is set.
12601
Willy Tarreau74ca5042013-06-11 23:12:07 +020012602Also, note that the "-i" flag applies to subsequent entries and not to entries
12603loaded from files preceding it. For instance :
12604
12605 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12606
12607In this example, each line of "exact-ua.lst" will be exactly matched against
12608the "user-agent" header of the request. Then each line of "generic-ua" will be
12609case-insensitively matched. Then the word "test" will be insensitively matched
12610as well.
12611
12612The "-m" flag is used to select a specific pattern matching method on the input
12613sample. All ACL-specific criteria imply a pattern matching method and generally
12614do not need this flag. However, this flag is useful with generic sample fetch
12615methods to describe how they're going to be matched against the patterns. This
12616is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012617obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012618followed by a pattern matching method name, this method is used instead of the
12619default one for the criterion. This makes it possible to match contents in ways
12620that were not initially planned, or with sample fetch methods which return a
12621string. The matching method also affects the way the patterns are parsed.
12622
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012623The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12624By default, if the parser cannot parse ip address it considers that the parsed
12625string is maybe a domain name and try dns resolution. The flag "-n" disable this
12626resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12627server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050012628waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012629flag "-n" disable this behavior. Note also that during the runtime, this
12630function is disabled for the dynamic acl modifications.
12631
Willy Tarreau74ca5042013-06-11 23:12:07 +020012632There are some restrictions however. Not all methods can be used with all
12633sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12634be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012635
12636 - "found" : only check if the requested sample could be found in the stream,
12637 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012638 to pass any pattern to avoid confusion. This matching method is
12639 particularly useful to detect presence of certain contents such
12640 as headers, cookies, etc... even if they are empty and without
12641 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012642
12643 - "bool" : check the value as a boolean. It can only be applied to fetches
12644 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012645 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012646
12647 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012648 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012649
12650 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012651 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012652
Davor Ocelice9ed2812017-12-25 17:49:28 +010012653 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012654 binary sequence. This may be used with binary or string samples.
12655
12656 - "len" : match the sample's length as an integer. This may be used with
12657 binary or string samples.
12658
Willy Tarreau74ca5042013-06-11 23:12:07 +020012659 - "str" : exact match : match the contents against a string. This may be
12660 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012661
Willy Tarreau74ca5042013-06-11 23:12:07 +020012662 - "sub" : substring match : check that the contents contain at least one of
12663 the provided string patterns. This may be used with binary or
12664 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012665
Willy Tarreau74ca5042013-06-11 23:12:07 +020012666 - "reg" : regex match : match the contents against a list of regular
12667 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012668
Willy Tarreau74ca5042013-06-11 23:12:07 +020012669 - "beg" : prefix match : check that the contents begin like the provided
12670 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012671
Willy Tarreau74ca5042013-06-11 23:12:07 +020012672 - "end" : suffix match : check that the contents end like the provided
12673 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012674
Willy Tarreau74ca5042013-06-11 23:12:07 +020012675 - "dir" : subdir match : check that a slash-delimited portion of the
12676 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012677 This may be used with binary or string samples.
12678
Willy Tarreau74ca5042013-06-11 23:12:07 +020012679 - "dom" : domain match : check that a dot-delimited portion of the contents
12680 exactly match one of the provided string patterns. This may be
12681 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012682
12683For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12684request, it is possible to do :
12685
12686 acl jsess_present cook(JSESSIONID) -m found
12687
12688In order to apply a regular expression on the 500 first bytes of data in the
12689buffer, one would use the following acl :
12690
12691 acl script_tag payload(0,500) -m reg -i <script>
12692
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012693On systems where the regex library is much slower when using "-i", it is
12694possible to convert the sample to lowercase before matching, like this :
12695
12696 acl script_tag payload(0,500),lower -m reg <script>
12697
Willy Tarreau74ca5042013-06-11 23:12:07 +020012698All ACL-specific criteria imply a default matching method. Most often, these
12699criteria are composed by concatenating the name of the original sample fetch
12700method and the matching method. For example, "hdr_beg" applies the "beg" match
12701to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12702criteria rely on a sample fetch method, it is always possible instead to use
12703the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012704
Willy Tarreau74ca5042013-06-11 23:12:07 +020012705If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012706the matching method is simply applied to the underlying sample fetch method.
12707For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012708
Willy Tarreau74ca5042013-06-11 23:12:07 +020012709 acl short_form hdr_beg(host) www.
12710 acl alternate1 hdr_beg(host) -m beg www.
12711 acl alternate2 hdr_dom(host) -m beg www.
12712 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012713
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012714
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012715The table below summarizes the compatibility matrix between sample or converter
12716types and the pattern types to fetch against. It indicates for each compatible
12717combination the name of the matching method to be used, surrounded with angle
12718brackets ">" and "<" when the method is the default one and will work by
12719default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012720
Willy Tarreau74ca5042013-06-11 23:12:07 +020012721 +-------------------------------------------------+
12722 | Input sample type |
12723 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012724 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012725 +----------------------+---------+---------+---------+---------+---------+
12726 | none (presence only) | found | found | found | found | found |
12727 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012728 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012729 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012730 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012731 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012732 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012733 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012734 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012735 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012736 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012737 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012738 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012739 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012740 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012741 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012742 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012743 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012744 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012745 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012746 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012747 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012748 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012749 +----------------------+---------+---------+---------+---------+---------+
12750 | hex block | | | | bin | bin |
12751 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012752
12753
Willy Tarreau74ca5042013-06-11 23:12:07 +0200127547.1.1. Matching booleans
12755------------------------
12756
12757In order to match a boolean, no value is needed and all values are ignored.
12758Boolean matching is used by default for all fetch methods of type "boolean".
12759When boolean matching is used, the fetched value is returned as-is, which means
12760that a boolean "true" will always match and a boolean "false" will never match.
12761
12762Boolean matching may also be enforced using "-m bool" on fetch methods which
12763return an integer value. Then, integer value 0 is converted to the boolean
12764"false" and all other values are converted to "true".
12765
Willy Tarreau6a06a402007-07-15 20:15:28 +020012766
Willy Tarreau74ca5042013-06-11 23:12:07 +0200127677.1.2. Matching integers
12768------------------------
12769
12770Integer matching applies by default to integer fetch methods. It can also be
12771enforced on boolean fetches using "-m int". In this case, "false" is converted
12772to the integer 0, and "true" is converted to the integer 1.
12773
12774Integer matching also supports integer ranges and operators. Note that integer
12775matching only applies to positive values. A range is a value expressed with a
12776lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012777
12778For instance, "1024:65535" is a valid range to represent a range of
12779unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12780representation of privileged ports, and ":1023" would also work.
12781
Willy Tarreau62644772008-07-16 18:36:06 +020012782As a special case, some ACL functions support decimal numbers which are in fact
12783two integers separated by a dot. This is used with some version checks for
12784instance. All integer properties apply to those decimal numbers, including
12785ranges and operators.
12786
Willy Tarreau6a06a402007-07-15 20:15:28 +020012787For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012788operators with ranges does not make much sense and is strongly discouraged.
12789Similarly, it does not make much sense to perform order comparisons with a set
12790of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012791
Willy Tarreau0ba27502007-12-24 16:55:16 +010012792Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012793
12794 eq : true if the tested value equals at least one value
12795 ge : true if the tested value is greater than or equal to at least one value
12796 gt : true if the tested value is greater than at least one value
12797 le : true if the tested value is less than or equal to at least one value
12798 lt : true if the tested value is less than at least one value
12799
Willy Tarreau0ba27502007-12-24 16:55:16 +010012800For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012801
12802 acl negative-length hdr_val(content-length) lt 0
12803
Willy Tarreau62644772008-07-16 18:36:06 +020012804This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12805
12806 acl sslv3 req_ssl_ver 3:3.1
12807
Willy Tarreau6a06a402007-07-15 20:15:28 +020012808
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128097.1.3. Matching strings
12810-----------------------
12811
12812String matching applies to string or binary fetch methods, and exists in 6
12813different forms :
12814
12815 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012816 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012817
12818 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012819 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012820
12821 - prefix match (-m beg) : the patterns are compared with the beginning of
12822 the extracted string, and the ACL matches if any of them matches.
12823
12824 - suffix match (-m end) : the patterns are compared with the end of the
12825 extracted string, and the ACL matches if any of them matches.
12826
Baptiste Assmann33db6002016-03-06 23:32:10 +010012827 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012828 string, delimited with slashes ("/"), and the ACL matches if any of them
12829 matches.
12830
12831 - domain match (-m dom) : the patterns are looked up inside the extracted
12832 string, delimited with dots ("."), and the ACL matches if any of them
12833 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012834
12835String matching applies to verbatim strings as they are passed, with the
12836exception of the backslash ("\") which makes it possible to escape some
12837characters such as the space. If the "-i" flag is passed before the first
12838string, then the matching will be performed ignoring the case. In order
12839to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012840before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012841
12842
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128437.1.4. Matching regular expressions (regexes)
12844---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012845
12846Just like with string matching, regex matching applies to verbatim strings as
12847they are passed, with the exception of the backslash ("\") which makes it
12848possible to escape some characters such as the space. If the "-i" flag is
12849passed before the first regex, then the matching will be performed ignoring
12850the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012851the "--" flag before the first string. Same principle applies of course to
12852match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012853
12854
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128557.1.5. Matching arbitrary data blocks
12856-------------------------------------
12857
12858It is possible to match some extracted samples against a binary block which may
12859not safely be represented as a string. For this, the patterns must be passed as
12860a series of hexadecimal digits in an even number, when the match method is set
12861to binary. Each sequence of two digits will represent a byte. The hexadecimal
12862digits may be used upper or lower case.
12863
12864Example :
12865 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12866 acl hello payload(0,6) -m bin 48656c6c6f0a
12867
12868
128697.1.6. Matching IPv4 and IPv6 addresses
12870---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012871
12872IPv4 addresses values can be specified either as plain addresses or with a
12873netmask appended, in which case the IPv4 address matches whenever it is
12874within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012875host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012876difficult to read and debug configurations. If hostnames are used, you should
12877at least ensure that they are present in /etc/hosts so that the configuration
12878does not depend on any random DNS match at the moment the configuration is
12879parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012880
Daniel Schnellereba56342016-04-13 00:26:52 +020012881The dotted IPv4 address notation is supported in both regular as well as the
12882abbreviated form with all-0-octets omitted:
12883
12884 +------------------+------------------+------------------+
12885 | Example 1 | Example 2 | Example 3 |
12886 +------------------+------------------+------------------+
12887 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12888 | 192.168.1 | 10.12 | 127.1 |
12889 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
12890 | 192.168.1/22 | 10.12/8 | 127.1/8 |
12891 +------------------+------------------+------------------+
12892
12893Notice that this is different from RFC 4632 CIDR address notation in which
12894192.168.42/24 would be equivalent to 192.168.42.0/24.
12895
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012896IPv6 may be entered in their usual form, with or without a netmask appended.
12897Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
12898trouble with randomly resolved IP addresses, host names are never allowed in
12899IPv6 patterns.
12900
12901HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
12902following situations :
12903 - tested address is IPv4, pattern address is IPv4, the match applies
12904 in IPv4 using the supplied mask if any.
12905 - tested address is IPv6, pattern address is IPv6, the match applies
12906 in IPv6 using the supplied mask if any.
12907 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
12908 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
12909 ::IPV4 or ::ffff:IPV4, otherwise it fails.
12910 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
12911 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
12912 applied in IPv6 using the supplied IPv6 mask.
12913
Willy Tarreau74ca5042013-06-11 23:12:07 +020012914
129157.2. Using ACLs to form conditions
12916----------------------------------
12917
12918Some actions are only performed upon a valid condition. A condition is a
12919combination of ACLs with operators. 3 operators are supported :
12920
12921 - AND (implicit)
12922 - OR (explicit with the "or" keyword or the "||" operator)
12923 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020012924
Willy Tarreau74ca5042013-06-11 23:12:07 +020012925A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020012926
Willy Tarreau74ca5042013-06-11 23:12:07 +020012927 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020012928
Willy Tarreau74ca5042013-06-11 23:12:07 +020012929Such conditions are generally used after an "if" or "unless" statement,
12930indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020012931
Willy Tarreau74ca5042013-06-11 23:12:07 +020012932For instance, to block HTTP requests to the "*" URL with methods other than
12933"OPTIONS", as well as POST requests without content-length, and GET or HEAD
12934requests with a content-length greater than 0, and finally every request which
12935is not either GET/HEAD/POST/OPTIONS !
12936
12937 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012938 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
12939 http-request deny if METH_GET HTTP_CONTENT
12940 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020012941
12942To select a different backend for requests to static contents on the "www" site
12943and to every request on the "img", "video", "download" and "ftp" hosts :
12944
12945 acl url_static path_beg /static /images /img /css
12946 acl url_static path_end .gif .png .jpg .css .js
12947 acl host_www hdr_beg(host) -i www
12948 acl host_static hdr_beg(host) -i img. video. download. ftp.
12949
Davor Ocelice9ed2812017-12-25 17:49:28 +010012950 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020012951 # of host "www". Use backend "www" for the rest.
12952 use_backend static if host_static or host_www url_static
12953 use_backend www if host_www
12954
12955It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
12956expressions that are built on the fly without needing to be declared. They must
12957be enclosed between braces, with a space before and after each brace (because
12958the braces must be seen as independent words). Example :
12959
12960 The following rule :
12961
12962 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012963 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020012964
12965 Can also be written that way :
12966
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012967 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020012968
12969It is generally not recommended to use this construct because it's a lot easier
12970to leave errors in the configuration when written that way. However, for very
12971simple rules matching only one source IP address for instance, it can make more
12972sense to use them than to declare ACLs with random names. Another example of
12973good use is the following :
12974
12975 With named ACLs :
12976
12977 acl site_dead nbsrv(dynamic) lt 2
12978 acl site_dead nbsrv(static) lt 2
12979 monitor fail if site_dead
12980
12981 With anonymous ACLs :
12982
12983 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
12984
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012985See section 4.2 for detailed help on the "http-request deny" and "use_backend"
12986keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012987
12988
129897.3. Fetching samples
12990---------------------
12991
12992Historically, sample fetch methods were only used to retrieve data to match
12993against patterns using ACLs. With the arrival of stick-tables, a new class of
12994sample fetch methods was created, most often sharing the same syntax as their
12995ACL counterpart. These sample fetch methods are also known as "fetches". As
12996of now, ACLs and fetches have converged. All ACL fetch methods have been made
12997available as fetch methods, and ACLs may use any sample fetch method as well.
12998
12999This section details all available sample fetch methods and their output type.
13000Some sample fetch methods have deprecated aliases that are used to maintain
13001compatibility with existing configurations. They are then explicitly marked as
13002deprecated and should not be used in new setups.
13003
13004The ACL derivatives are also indicated when available, with their respective
13005matching methods. These ones all have a well defined default pattern matching
13006method, so it is never necessary (though allowed) to pass the "-m" option to
13007indicate how the sample will be matched using ACLs.
13008
13009As indicated in the sample type versus matching compatibility matrix above,
13010when using a generic sample fetch method in an ACL, the "-m" option is
13011mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13012the same keyword exists as an ACL keyword and as a standard fetch method, the
13013ACL engine will automatically pick the ACL-only one by default.
13014
13015Some of these keywords support one or multiple mandatory arguments, and one or
13016multiple optional arguments. These arguments are strongly typed and are checked
13017when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013018incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13019are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013020is optional, it will be indicated below between square brackets ('[ ]'). When
13021all arguments are optional, the parenthesis may be omitted.
13022
13023Thus, the syntax of a standard sample fetch method is one of the following :
13024 - name
13025 - name(arg1)
13026 - name(arg1,arg2)
13027
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013028
130297.3.1. Converters
13030-----------------
13031
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013032Sample fetch methods may be combined with transformations to be applied on top
13033of the fetched sample (also called "converters"). These combinations form what
13034is called "sample expressions" and the result is a "sample". Initially this
13035was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013036has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013037unique-id-format, add-header, ...).
13038
13039These transformations are enumerated as a series of specific keywords after the
13040sample fetch method. These keywords may equally be appended immediately after
13041the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013042support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013043
Willy Tarreau97707872015-01-27 15:12:13 +010013044A certain category of converters are bitwise and arithmetic operators which
13045support performing basic operations on integers. Some bitwise operations are
13046supported (and, or, xor, cpl) and some arithmetic operations are supported
13047(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13048bool) which make it possible to report a match without having to write an ACL.
13049
Willy Tarreau74ca5042013-06-11 23:12:07 +020013050The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013051
Ben Shillitof25e8e52016-12-02 14:25:37 +00001305251d.single(<prop>[,<prop>*])
13053 Returns values for the properties requested as a string, where values are
13054 separated by the delimiter specified with "51degrees-property-separator".
13055 The device is identified using the User-Agent header passed to the
13056 converter. The function can be passed up to five property names, and if a
13057 property name can't be found, the value "NoData" is returned.
13058
13059 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013060 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13061 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013062 # User-Agent passed to the converter.
13063 frontend http-in
13064 bind *:8081
13065 default_backend servers
13066 http-request set-header X-51D-DeviceTypeMobileTablet \
13067 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13068
Willy Tarreau97707872015-01-27 15:12:13 +010013069add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013070 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013071 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013072 name. The name of the variable starts with an indication about its scope. The
13073 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013074 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013075 "sess" : the variable is shared with the whole session
13076 "txn" : the variable is shared with the transaction (request and response)
13077 "req" : the variable is shared only during request processing
13078 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013079 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013080 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013081
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010013082aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
13083 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
13084 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
13085 need to be base64 encoded and the returned result is in raw byte format.
13086 If the <aead_tag> validation fails, the converter doesn't return any data.
13087 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
13088 converter requires at least OpenSSL 1.0.1.
13089
13090 Example:
13091 http-response set-header X-Decrypted-Text %[var(txn.enc),\
13092 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
13093
Willy Tarreau97707872015-01-27 15:12:13 +010013094and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013095 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013096 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013097 numeric value or a variable name. The name of the variable starts with an
13098 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013099 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013100 "sess" : the variable is shared with the whole session
13101 "txn" : the variable is shared with the transaction (request and response)
13102 "req" : the variable is shared only during request processing
13103 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013104 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013105 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013106
Holger Just1bfc24b2017-05-06 00:56:53 +020013107b64dec
13108 Converts (decodes) a base64 encoded input string to its binary
13109 representation. It performs the inverse operation of base64().
13110
Emeric Brun53d1a982014-04-30 18:21:37 +020013111base64
13112 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013113 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013114 an SSL ID can be copied in a header).
13115
Willy Tarreau97707872015-01-27 15:12:13 +010013116bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013117 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013118 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013119 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013120 presence of a flag).
13121
Emeric Brun54c4ac82014-11-03 15:32:43 +010013122bytes(<offset>[,<length>])
13123 Extracts some bytes from an input binary sample. The result is a binary
13124 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013125 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013126
Willy Tarreau280f42b2018-02-19 15:34:12 +010013127concat([<start>],[<var>],[<end>])
13128 Concatenates up to 3 fields after the current sample which is then turned to
13129 a string. The first one, <start>, is a constant string, that will be appended
13130 immediately after the existing sample. It may be omitted if not used. The
13131 second one, <var>, is a variable name. The variable will be looked up, its
13132 contents converted to a string, and it will be appended immediately after the
13133 <first> part. If the variable is not found, nothing is appended. It may be
13134 omitted as well. The third field, <end> is a constant string that will be
13135 appended after the variable. It may also be omitted. Together, these elements
13136 allow to concatenate variables with delimiters to an existing set of
13137 variables. This can be used to build new variables made of a succession of
John Roeslerfb2fce12019-07-10 15:45:51 -050013138 other variables, such as colon-delimited values. Note that due to the config
Willy Tarreau280f42b2018-02-19 15:34:12 +010013139 parser, it is not possible to use a comma nor a closing parenthesis as
John Roeslerfb2fce12019-07-10 15:45:51 -050013140 delimiters.
Willy Tarreau280f42b2018-02-19 15:34:12 +010013141
13142 Example:
13143 tcp-request session set-var(sess.src) src
13144 tcp-request session set-var(sess.dn) ssl_c_s_dn
13145 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13146 http-request set-header x-hap-sig %[var(txn.sig)]
13147
Willy Tarreau97707872015-01-27 15:12:13 +010013148cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013149 Takes the input value of type signed integer, applies a ones-complement
13150 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013151
Willy Tarreau80599772015-01-20 19:35:24 +010013152crc32([<avalanche>])
13153 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13154 hash function. Optionally, it is possible to apply a full avalanche hash
13155 function to the output if the optional <avalanche> argument equals 1. This
13156 converter uses the same functions as used by the various hash-based load
13157 balancing algorithms, so it will provide exactly the same results. It is
13158 provided for compatibility with other software which want a CRC32 to be
13159 computed on some input keys, so it follows the most common implementation as
13160 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13161 but may provide a better or at least less predictable distribution. It must
13162 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013163 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13164
13165crc32c([<avalanche>])
13166 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13167 hash function. Optionally, it is possible to apply a full avalanche hash
13168 function to the output if the optional <avalanche> argument equals 1. This
13169 converter uses the same functions as described in RFC4960, Appendix B [8].
13170 It is provided for compatibility with other software which want a CRC32C to be
13171 computed on some input keys. It is slower than the other algorithms and it must
13172 not be used for security purposes as a 32-bit hash is trivial to break. See
13173 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013174
David Carlier29b3ca32015-09-25 14:09:21 +010013175da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013176 Asks the DeviceAtlas converter to identify the User Agent string passed on
13177 input, and to emit a string made of the concatenation of the properties
13178 enumerated in argument, delimited by the separator defined by the global
13179 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013180 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013181 configuration language.
13182
13183 Example:
13184 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013185 bind *:8881
13186 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013187 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 +020013188
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013189debug
13190 This converter is used as debug tool. It dumps on screen the content and the
13191 type of the input sample. The sample is returned as is on its output. This
13192 converter only exists when haproxy was built with debugging enabled.
13193
Willy Tarreau97707872015-01-27 15:12:13 +010013194div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013195 Divides the input value of type signed integer by <value>, and returns the
13196 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013197 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013198 variable name. The name of the variable starts with an indication about its
13199 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013200 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013201 "sess" : the variable is shared with the whole session
13202 "txn" : the variable is shared with the transaction (request and response)
13203 "req" : the variable is shared only during request processing
13204 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013205 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013206 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013207
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013208djb2([<avalanche>])
13209 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13210 hash function. Optionally, it is possible to apply a full avalanche hash
13211 function to the output if the optional <avalanche> argument equals 1. This
13212 converter uses the same functions as used by the various hash-based load
13213 balancing algorithms, so it will provide exactly the same results. It is
13214 mostly intended for debugging, but can be used as a stick-table entry to
13215 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013216 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13217 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013218
Willy Tarreau97707872015-01-27 15:12:13 +010013219even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013220 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013221 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13222
Marcin Deranek9631a282018-04-16 14:30:46 +020013223field(<index>,<delimiters>[,<count>])
13224 Extracts the substring at the given index counting from the beginning
13225 (positive index) or from the end (negative index) considering given delimiters
13226 from an input string. Indexes start at 1 or -1 and delimiters are a string
13227 formatted list of chars. Optionally you can specify <count> of fields to
13228 extract (default: 1). Value of 0 indicates extraction of all remaining
13229 fields.
13230
13231 Example :
13232 str(f1_f2_f3__f5),field(5,_) # f5
13233 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13234 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13235 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13236 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013237
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013238hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013239 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013240 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013241 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 +020013242 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013243
Dragan Dosen3f957b22017-10-24 09:27:34 +020013244hex2i
13245 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050013246 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020013247
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013248http_date([<offset>])
13249 Converts an integer supposed to contain a date since epoch to a string
13250 representing this date in a format suitable for use in HTTP header fields. If
13251 an offset value is specified, then it is a number of seconds that is added to
13252 the date before the conversion is operated. This is particularly useful to
13253 emit Date header fields, Expires values in responses when combined with a
13254 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013255
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013256in_table(<table>)
13257 Uses the string representation of the input sample to perform a look up in
13258 the specified table. If the key is not found in the table, a boolean false
13259 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013260 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013261 or not a source IP address or an Authorization header was already seen).
13262
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013263ipmask(<mask4>, [<mask6>])
13264 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013265 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013266 table entries and as such use the same server. The mask4 can be passed in
13267 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13268 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13269 If no mask6 is given IPv6 addresses will fail to convert for backwards
13270 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013271
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013272json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013273 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013274 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013275 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013276 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13277 of errors:
13278 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13279 bytes, ...)
13280 - invalid range (the decoded value is within a UTF-8 prohibited range),
13281 - code overlong (the value is encoded with more bytes than necessary).
13282
13283 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13284 character is greater than 0xffff because the JSON string escape specification
13285 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13286 in 4 variants designated by a combination of two suffix letters : "p" for
13287 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13288 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013289 - "ascii" : never fails;
13290 - "utf8" : fails on any detected errors;
13291 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013292 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013293 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013294 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13295 characters corresponding to the other errors.
13296
13297 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013298 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013299
13300 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013301 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013302 capture request header user-agent len 150
13303 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013304
13305 Input request from client 127.0.0.1:
13306 GET / HTTP/1.0
13307 User-Agent: Very "Ugly" UA 1/2
13308
13309 Output log:
13310 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13311
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013312language(<value>[,<default>])
13313 Returns the value with the highest q-factor from a list as extracted from the
13314 "accept-language" header using "req.fhdr". Values with no q-factor have a
13315 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13316 belong to the list of semi-colon delimited <values> will be considered. The
13317 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13318 given list and a default value is provided, it is returned. Note that language
13319 names may have a variant after a dash ('-'). If this variant is present in the
13320 list, it will be matched, but if it is not, only the base language is checked.
13321 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013322 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013323 ordering of the values in the request counts, as the first value among
13324 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013325
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013326 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013327
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013328 # this configuration switches to the backend matching a
13329 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013330
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013331 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13332 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13333 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13334 use_backend spanish if es
13335 use_backend french if fr
13336 use_backend english if en
13337 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013338
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013339length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013340 Get the length of the string. This can only be placed after a string
13341 sample fetch function or after a transformation keyword returning a string
13342 type. The result is of type integer.
13343
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013344lower
13345 Convert a string sample to lower case. This can only be placed after a string
13346 sample fetch function or after a transformation keyword returning a string
13347 type. The result is of type string.
13348
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013349ltime(<format>[,<offset>])
13350 Converts an integer supposed to contain a date since epoch to a string
13351 representing this date in local time using a format defined by the <format>
13352 string using strftime(3). The purpose is to allow any date format to be used
13353 in logs. An optional <offset> in seconds may be applied to the input date
13354 (positive or negative). See the strftime() man page for the format supported
13355 by your operating system. See also the utime converter.
13356
13357 Example :
13358
13359 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013360 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013361 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13362
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013363map(<map_file>[,<default_value>])
13364map_<match_type>(<map_file>[,<default_value>])
13365map_<match_type>_<output_type>(<map_file>[,<default_value>])
13366 Search the input value from <map_file> using the <match_type> matching method,
13367 and return the associated value converted to the type <output_type>. If the
13368 input value cannot be found in the <map_file>, the converter returns the
13369 <default_value>. If the <default_value> is not set, the converter fails and
13370 acts as if no input value could be fetched. If the <match_type> is not set, it
13371 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13372 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13373 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013374
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013375 It is important to avoid overlapping between the keys : IP addresses and
13376 strings are stored in trees, so the first of the finest match will be used.
13377 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013378
Tim Düsterhus4896c442016-11-29 02:15:19 +010013379 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013380 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013381
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013382 input type | match method | output type str | output type int | output type ip
13383 -----------+--------------+-----------------+-----------------+---------------
13384 str | str | map_str | map_str_int | map_str_ip
13385 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013386 str | beg | map_beg | map_beg_int | map_end_ip
13387 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013388 str | sub | map_sub | map_sub_int | map_sub_ip
13389 -----------+--------------+-----------------+-----------------+---------------
13390 str | dir | map_dir | map_dir_int | map_dir_ip
13391 -----------+--------------+-----------------+-----------------+---------------
13392 str | dom | map_dom | map_dom_int | map_dom_ip
13393 -----------+--------------+-----------------+-----------------+---------------
13394 str | end | map_end | map_end_int | map_end_ip
13395 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013396 str | reg | map_reg | map_reg_int | map_reg_ip
13397 -----------+--------------+-----------------+-----------------+---------------
13398 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013399 -----------+--------------+-----------------+-----------------+---------------
13400 int | int | map_int | map_int_int | map_int_ip
13401 -----------+--------------+-----------------+-----------------+---------------
13402 ip | ip | map_ip | map_ip_int | map_ip_ip
13403 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013404
Thierry Fournier8feaa662016-02-10 22:55:20 +010013405 The special map called "map_regm" expect matching zone in the regular
13406 expression and modify the output replacing back reference (like "\1") by
13407 the corresponding match text.
13408
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013409 The file contains one key + value per line. Lines which start with '#' are
13410 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13411 is then the first "word" (series of non-space/tabs characters), and the value
13412 is what follows this series of space/tab till the end of the line excluding
13413 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013414
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013415 Example :
13416
13417 # this is a comment and is ignored
13418 2.22.246.0/23 United Kingdom \n
13419 <-><-----------><--><------------><---->
13420 | | | | `- trailing spaces ignored
13421 | | | `---------- value
13422 | | `-------------------- middle spaces ignored
13423 | `---------------------------- key
13424 `------------------------------------ leading spaces ignored
13425
Willy Tarreau97707872015-01-27 15:12:13 +010013426mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013427 Divides the input value of type signed integer by <value>, and returns the
13428 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013429 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013430 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013431 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013432 "sess" : the variable is shared with the whole session
13433 "txn" : the variable is shared with the transaction (request and response)
13434 "req" : the variable is shared only during request processing
13435 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013436 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013437 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013438
13439mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013440 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013441 the product as an signed integer. In case of overflow, the largest possible
13442 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013443 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013444 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013445 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013446 "sess" : the variable is shared with the whole session
13447 "txn" : the variable is shared with the transaction (request and response)
13448 "req" : the variable is shared only during request processing
13449 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013450 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013451 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013452
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013453nbsrv
13454 Takes an input value of type string, interprets it as a backend name and
13455 returns the number of usable servers in that backend. Can be used in places
13456 where we want to look up a backend from a dynamic name, like a result of a
13457 map lookup.
13458
Willy Tarreau97707872015-01-27 15:12:13 +010013459neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013460 Takes the input value of type signed integer, computes the opposite value,
13461 and returns the remainder as an signed integer. 0 is identity. This operator
13462 is provided for reversed subtracts : in order to subtract the input from a
13463 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013464
13465not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013466 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013467 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013468 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013469 absence of a flag).
13470
13471odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013472 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013473 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13474
13475or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013476 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013477 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013478 numeric value or a variable name. The name of the variable starts with an
13479 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013480 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013481 "sess" : the variable is shared with the whole session
13482 "txn" : the variable is shared with the transaction (request and response)
13483 "req" : the variable is shared only during request processing
13484 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013485 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013486 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013487
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013488protobuf(<field_number>,[<field_type>])
13489 This extracts the protocol buffers message field in raw mode of an input binary
13490 sample representation of a protocol buffer message with <field_number> as field
13491 number (dotted notation) if <field_type> is not present, or as an integer sample
13492 if this field is present (see also "ungrpc" below).
13493 The list of the authorized types is the following one: "int32", "int64", "uint32",
13494 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13495 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13496 "float" for the wire type 5. Note that "string" is considered as a length-delimited
13497 type, so it does not require any <field_type> argument to be extracted.
13498 More information may be found here about the protocol buffers message field types:
13499 https://developers.google.com/protocol-buffers/docs/encoding
13500
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013501regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013502 Applies a regex-based substitution to the input string. It does the same
13503 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13504 default it will replace in the input string the first occurrence of the
13505 largest part matching the regular expression <regex> with the substitution
13506 string <subst>. It is possible to replace all occurrences instead by adding
13507 the flag "g" in the third argument <flags>. It is also possible to make the
13508 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13509 string, it is made up from the concatenation of all desired flags. Thus if
13510 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13511 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013512 configuration parser, some characters such as closing parenthesis, closing
13513 square brackets or comma are not possible to use in the arguments. The first
13514 use of this converter is to replace certain characters or sequence of
13515 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013516
13517 Example :
13518
13519 # de-duplicate "/" in header "x-path".
13520 # input: x-path: /////a///b/c/xzxyz/
13521 # output: x-path: /a/b/c/xzxyz/
13522 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13523
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013524capture-req(<id>)
13525 Capture the string entry in the request slot <id> and returns the entry as
13526 is. If the slot doesn't exist, the capture fails silently.
13527
13528 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013529 "http-response capture", "capture.req.hdr" and
13530 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013531
13532capture-res(<id>)
13533 Capture the string entry in the response slot <id> and returns the entry as
13534 is. If the slot doesn't exist, the capture fails silently.
13535
13536 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013537 "http-response capture", "capture.req.hdr" and
13538 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013539
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013540sdbm([<avalanche>])
13541 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13542 hash function. Optionally, it is possible to apply a full avalanche hash
13543 function to the output if the optional <avalanche> argument equals 1. This
13544 converter uses the same functions as used by the various hash-based load
13545 balancing algorithms, so it will provide exactly the same results. It is
13546 mostly intended for debugging, but can be used as a stick-table entry to
13547 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013548 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13549 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013550
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013551set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013552 Sets a variable with the input content and returns the content on the output
13553 as-is. The variable keeps the value and the associated input type. The name of
13554 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013555 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013556 "sess" : the variable is shared with the whole session
13557 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013558 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013559 "req" : the variable is shared only during request processing,
13560 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013561 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013562 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013563
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013564sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020013565 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013566 sample with length of 20 bytes.
13567
Tim Duesterhusd4376302019-06-17 12:41:44 +020013568sha2([<bits>])
13569 Converts a binary input sample to a digest in the SHA-2 family. The result
13570 is a binary sample with length of <bits>/8 bytes.
13571
13572 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
13573 SHA-<bits>. The default value is 256.
13574
13575 Please note that this converter is only available when haproxy has been
13576 compiled with USE_OPENSSL.
13577
Nenad Merdanovic177adc92019-08-27 01:58:13 +020013578srv_queue
13579 Takes an input value of type string, either a server name or <backend>/<server>
13580 format and returns the number of queued sessions on that server. Can be used
13581 in places where we want to look up queued sessions from a dynamic name, like a
13582 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
13583 persistence or direct a request elsewhere.
13584
Tim Duesterhusca097c12018-04-27 21:18:45 +020013585strcmp(<var>)
13586 Compares the contents of <var> with the input value of type string. Returns
13587 the result as a signed integer compatible with strcmp(3): 0 if both strings
13588 are identical. A value less than 0 if the left string is lexicographically
13589 smaller than the right string or if the left string is shorter. A value greater
13590 than 0 otherwise (right string greater than left string or the right string is
13591 shorter).
13592
13593 Example :
13594
13595 http-request set-var(txn.host) hdr(host)
13596 # Check whether the client is attempting domain fronting.
13597 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13598
13599
Willy Tarreau97707872015-01-27 15:12:13 +010013600sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013601 Subtracts <value> from the input value of type signed integer, and returns
13602 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013603 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013604 or a variable name. The name of the variable starts with an indication about
13605 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013606 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013607 "sess" : the variable is shared with the whole session
13608 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013609 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013610 "req" : the variable is shared only during request processing,
13611 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013612 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013613 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013614
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013615table_bytes_in_rate(<table>)
13616 Uses the string representation of the input sample to perform a look up in
13617 the specified table. If the key is not found in the table, integer value zero
13618 is returned. Otherwise the converter returns the average client-to-server
13619 bytes rate associated with the input sample in the designated table, measured
13620 in amount of bytes over the period configured in the table. See also the
13621 sc_bytes_in_rate sample fetch keyword.
13622
13623
13624table_bytes_out_rate(<table>)
13625 Uses the string representation of the input sample to perform a look up in
13626 the specified table. If the key is not found in the table, integer value zero
13627 is returned. Otherwise the converter returns the average server-to-client
13628 bytes rate associated with the input sample in the designated table, measured
13629 in amount of bytes over the period configured in the table. See also the
13630 sc_bytes_out_rate sample fetch keyword.
13631
13632table_conn_cnt(<table>)
13633 Uses the string representation of the input sample to perform a look up in
13634 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013635 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013636 connections associated with the input sample in the designated table. See
13637 also the sc_conn_cnt sample fetch keyword.
13638
13639table_conn_cur(<table>)
13640 Uses the string representation of the input sample to perform a look up in
13641 the specified table. If the key is not found in the table, integer value zero
13642 is returned. Otherwise the converter returns the current amount of concurrent
13643 tracked connections associated with the input sample in the designated table.
13644 See also the sc_conn_cur sample fetch keyword.
13645
13646table_conn_rate(<table>)
13647 Uses the string representation of the input sample to perform a look up in
13648 the specified table. If the key is not found in the table, integer value zero
13649 is returned. Otherwise the converter returns the average incoming connection
13650 rate associated with the input sample in the designated table. See also the
13651 sc_conn_rate sample fetch keyword.
13652
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013653table_gpt0(<table>)
13654 Uses the string representation of the input sample to perform a look up in
13655 the specified table. If the key is not found in the table, boolean value zero
13656 is returned. Otherwise the converter returns the current value of the first
13657 general purpose tag associated with the input sample in the designated table.
13658 See also the sc_get_gpt0 sample fetch keyword.
13659
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013660table_gpc0(<table>)
13661 Uses the string representation of the input sample to perform a look up in
13662 the specified table. If the key is not found in the table, integer value zero
13663 is returned. Otherwise the converter returns the current value of the first
13664 general purpose counter associated with the input sample in the designated
13665 table. See also the sc_get_gpc0 sample fetch keyword.
13666
13667table_gpc0_rate(<table>)
13668 Uses the string representation of the input sample to perform a look up in
13669 the specified table. If the key is not found in the table, integer value zero
13670 is returned. Otherwise the converter returns the frequency which the gpc0
13671 counter was incremented over the configured period in the table, associated
13672 with the input sample in the designated table. See also the sc_get_gpc0_rate
13673 sample fetch keyword.
13674
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013675table_gpc1(<table>)
13676 Uses the string representation of the input sample to perform a look up in
13677 the specified table. If the key is not found in the table, integer value zero
13678 is returned. Otherwise the converter returns the current value of the second
13679 general purpose counter associated with the input sample in the designated
13680 table. See also the sc_get_gpc1 sample fetch keyword.
13681
13682table_gpc1_rate(<table>)
13683 Uses the string representation of the input sample to perform a look up in
13684 the specified table. If the key is not found in the table, integer value zero
13685 is returned. Otherwise the converter returns the frequency which the gpc1
13686 counter was incremented over the configured period in the table, associated
13687 with the input sample in the designated table. See also the sc_get_gpc1_rate
13688 sample fetch keyword.
13689
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013690table_http_err_cnt(<table>)
13691 Uses the string representation of the input sample to perform a look up in
13692 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013693 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013694 errors associated with the input sample in the designated table. See also the
13695 sc_http_err_cnt sample fetch keyword.
13696
13697table_http_err_rate(<table>)
13698 Uses the string representation of the input sample to perform a look up in
13699 the specified table. If the key is not found in the table, integer value zero
13700 is returned. Otherwise the average rate of HTTP errors associated with the
13701 input sample in the designated table, measured in amount of errors over the
13702 period configured in the table. See also the sc_http_err_rate sample fetch
13703 keyword.
13704
13705table_http_req_cnt(<table>)
13706 Uses the string representation of the input sample to perform a look up in
13707 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013708 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013709 requests associated with the input sample in the designated table. See also
13710 the sc_http_req_cnt sample fetch keyword.
13711
13712table_http_req_rate(<table>)
13713 Uses the string representation of the input sample to perform a look up in
13714 the specified table. If the key is not found in the table, integer value zero
13715 is returned. Otherwise the average rate of HTTP requests associated with the
13716 input sample in the designated table, measured in amount of requests over the
13717 period configured in the table. See also the sc_http_req_rate sample fetch
13718 keyword.
13719
13720table_kbytes_in(<table>)
13721 Uses the string representation of the input sample to perform a look up in
13722 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013723 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013724 to-server data associated with the input sample in the designated table,
13725 measured in kilobytes. The test is currently performed on 32-bit integers,
13726 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13727 keyword.
13728
13729table_kbytes_out(<table>)
13730 Uses the string representation of the input sample to perform a look up in
13731 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013732 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013733 to-client data associated with the input sample in the designated table,
13734 measured in kilobytes. The test is currently performed on 32-bit integers,
13735 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13736 keyword.
13737
13738table_server_id(<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 server ID associated with
13742 the input sample in the designated table. A server ID is associated to a
13743 sample by a "stick" rule when a connection to a server succeeds. A server ID
13744 zero means that no server is associated with this key.
13745
13746table_sess_cnt(<table>)
13747 Uses the string representation of the input sample to perform a look up in
13748 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013749 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013750 sessions associated with the input sample in the designated table. Note that
13751 a session here refers to an incoming connection being accepted by the
13752 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13753 keyword.
13754
13755table_sess_rate(<table>)
13756 Uses the string representation of the input sample to perform a look up in
13757 the specified table. If the key is not found in the table, integer value zero
13758 is returned. Otherwise the converter returns the average incoming session
13759 rate associated with the input sample in the designated table. Note that a
13760 session here refers to an incoming connection being accepted by the
13761 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13762 keyword.
13763
13764table_trackers(<table>)
13765 Uses the string representation of the input sample to perform a look up in
13766 the specified table. If the key is not found in the table, integer value zero
13767 is returned. Otherwise the converter returns the current amount of concurrent
13768 connections tracking the same key as the input sample in the designated
13769 table. It differs from table_conn_cur in that it does not rely on any stored
13770 information but on the table's reference count (the "use" value which is
13771 returned by "show table" on the CLI). This may sometimes be more suited for
13772 layer7 tracking. It can be used to tell a server how many concurrent
13773 connections there are from a given address for example. See also the
13774 sc_trackers sample fetch keyword.
13775
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013776upper
13777 Convert a string sample to upper case. This can only be placed after a string
13778 sample fetch function or after a transformation keyword returning a string
13779 type. The result is of type string.
13780
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013781url_dec
13782 Takes an url-encoded string provided as input and returns the decoded
13783 version as output. The input and the output are of type string.
13784
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013785ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013786 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013787 sample representation of a gRPC message with <field_number> as field number
13788 (dotted notation) if <field_type> is not present, or as an integer sample if this
13789 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013790 The list of the authorized types is the following one: "int32", "int64", "uint32",
13791 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13792 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13793 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013794 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013795 More information may be found here about the protocol buffers message field types:
13796 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013797
13798 Example:
13799 // with such a protocol buffer .proto file content adapted from
13800 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
13801
13802 message Point {
13803 int32 latitude = 1;
13804 int32 longitude = 2;
13805 }
13806
13807 message PPoint {
13808 Point point = 59;
13809 }
13810
13811 message Rectangle {
13812 // One corner of the rectangle.
13813 PPoint lo = 48;
13814 // The other corner of the rectangle.
13815 PPoint hi = 49;
13816 }
13817
13818 let's say a body request is made of a "Rectangle" object value (two PPoint
13819 protocol buffers messages), the four protocol buffers fields could be
13820 extracted with these "ungrpc" directives:
13821
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013822 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
13823 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050013824 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013825 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
13826
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013827 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013828
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013829 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013830
John Roeslerfb2fce12019-07-10 15:45:51 -050013831 As a gRPC message is always made of a gRPC header followed by protocol buffers
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013832 messages, in the previous example the "latitude" of "lo" first PPoint
13833 could be extracted with these equivalent directives:
13834
13835 req.body,ungrpc(48.59),protobuf(1,int32)
13836 req.body,ungrpc(48),protobuf(59.1,int32)
13837 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
13838
13839 Note that the first convert must be "ungrpc", the remaining ones must be
13840 "protobuf" and only the last one may have or not a second argument to
13841 interpret the previous binary sample.
13842
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013843
Christopher Faulet85d79c92016-11-09 16:54:56 +010013844unset-var(<var name>)
13845 Unsets a variable if the input content is defined. The name of the variable
13846 starts with an indication about its scope. The scopes allowed are:
13847 "proc" : the variable is shared with the whole process
13848 "sess" : the variable is shared with the whole session
13849 "txn" : the variable is shared with the transaction (request and
13850 response),
13851 "req" : the variable is shared only during request processing,
13852 "res" : the variable is shared only during response processing.
13853 This prefix is followed by a name. The separator is a '.'. The name may only
13854 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13855
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013856utime(<format>[,<offset>])
13857 Converts an integer supposed to contain a date since epoch to a string
13858 representing this date in UTC time using a format defined by the <format>
13859 string using strftime(3). The purpose is to allow any date format to be used
13860 in logs. An optional <offset> in seconds may be applied to the input date
13861 (positive or negative). See the strftime() man page for the format supported
13862 by your operating system. See also the ltime converter.
13863
13864 Example :
13865
13866 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013867 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013868 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13869
Marcin Deranek9631a282018-04-16 14:30:46 +020013870word(<index>,<delimiters>[,<count>])
13871 Extracts the nth word counting from the beginning (positive index) or from
13872 the end (negative index) considering given delimiters from an input string.
13873 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
13874 Optionally you can specify <count> of words to extract (default: 1).
13875 Value of 0 indicates extraction of all remaining words.
13876
13877 Example :
13878 str(f1_f2_f3__f5),word(4,_) # f5
13879 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
13880 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
13881 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
13882 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013883
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013884wt6([<avalanche>])
13885 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13886 hash function. Optionally, it is possible to apply a full avalanche hash
13887 function to the output if the optional <avalanche> argument equals 1. This
13888 converter uses the same functions as used by the various hash-based load
13889 balancing algorithms, so it will provide exactly the same results. It is
13890 mostly intended for debugging, but can be used as a stick-table entry to
13891 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013892 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
13893 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013894
Willy Tarreau97707872015-01-27 15:12:13 +010013895xor(<value>)
13896 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013897 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013898 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013899 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013900 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013901 "sess" : the variable is shared with the whole session
13902 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013903 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013904 "req" : the variable is shared only during request processing,
13905 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013906 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013907 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013908
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013909xxh32([<seed>])
13910 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13911 variant of the XXHash hash function. This hash supports a seed which defaults
13912 to zero but a different value maybe passed as the <seed> argument. This hash
13913 is known to be very good and very fast so it can be used to hash URLs and/or
13914 URL parameters for use as stick-table keys to collect statistics with a low
13915 collision rate, though care must be taken as the algorithm is not considered
13916 as cryptographically secure.
13917
13918xxh64([<seed>])
13919 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13920 variant of the XXHash hash function. This hash supports a seed which defaults
13921 to zero but a different value maybe passed as the <seed> argument. This hash
13922 is known to be very good and very fast so it can be used to hash URLs and/or
13923 URL parameters for use as stick-table keys to collect statistics with a low
13924 collision rate, though care must be taken as the algorithm is not considered
13925 as cryptographically secure.
13926
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013927
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200139287.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020013929--------------------------------------------
13930
13931A first set of sample fetch methods applies to internal information which does
13932not even relate to any client information. These ones are sometimes used with
13933"monitor-fail" directives to report an internal status to external watchers.
13934The sample fetch methods described in this section are usable anywhere.
13935
13936always_false : boolean
13937 Always returns the boolean "false" value. It may be used with ACLs as a
13938 temporary replacement for another one when adjusting configurations.
13939
13940always_true : boolean
13941 Always returns the boolean "true" value. It may be used with ACLs as a
13942 temporary replacement for another one when adjusting configurations.
13943
13944avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013945 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020013946 divided by the number of active servers. The current backend is used if no
13947 backend is specified. This is very similar to "queue" except that the size of
13948 the farm is considered, in order to give a more accurate measurement of the
13949 time it may take for a new connection to be processed. The main usage is with
13950 ACL to return a sorry page to new users when it becomes certain they will get
13951 a degraded service, or to pass to the backend servers in a header so that
13952 they decide to work in degraded mode or to disable some functions to speed up
13953 the processing a bit. Note that in the event there would not be any active
13954 server anymore, twice the number of queued connections would be considered as
13955 the measured value. This is a fair estimate, as we expect one server to get
13956 back soon anyway, but we still prefer to send new traffic to another backend
13957 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
13958 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010013959
Willy Tarreau74ca5042013-06-11 23:12:07 +020013960be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020013961 Applies to the number of currently established connections on the backend,
13962 possibly including the connection being evaluated. If no backend name is
13963 specified, the current one is used. But it is also possible to check another
13964 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013965 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
13966
13967be_conn_free([<backend>]) : integer
13968 Returns an integer value corresponding to the number of available connections
13969 across available servers in the backend. Queue slots are not included. Backup
13970 servers are also not included, unless all other servers are down. If no
13971 backend name is specified, the current one is used. But it is also possible
13972 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040013973 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
13974 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013975
13976 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
13977 (meaning unlimited), then this fetch clearly does not make sense, in which
13978 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013979
Willy Tarreau74ca5042013-06-11 23:12:07 +020013980be_sess_rate([<backend>]) : integer
13981 Returns an integer value corresponding to the sessions creation rate on the
13982 backend, in number of new sessions per second. This is used with ACLs to
13983 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010013984 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020013985 online dictionary). It can also be useful to add this element to logs using a
13986 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013987
13988 Example :
13989 # Redirect to an error page if the dictionary is requested too often
13990 backend dynamic
13991 mode http
13992 acl being_scanned be_sess_rate gt 100
13993 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010013994
Davor Ocelice9ed2812017-12-25 17:49:28 +010013995bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013996 Returns a binary chain. The input is the hexadecimal representation
13997 of the string.
13998
13999bool(<bool>) : bool
14000 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14001 'false' and '0' are the same. 'true' and '1' are the same.
14002
Willy Tarreau74ca5042013-06-11 23:12:07 +020014003connslots([<backend>]) : integer
14004 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014005 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014006 connections on all servers and the maximum queue size. This is probably only
14007 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014008
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014009 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014010 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014011 usage; see "use_backend" keyword) can be redirected to a different backend.
14012
Willy Tarreau55165fe2009-05-10 12:02:55 +020014013 'connslots' = number of available server connection slots, + number of
14014 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014015
Willy Tarreaua36af912009-10-10 12:02:45 +020014016 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014017 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014018 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014019 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014020 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014021 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014022 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014023
Willy Tarreau55165fe2009-05-10 12:02:55 +020014024 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14025 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014026 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014027 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014028
Willy Tarreau70fe9442018-11-22 16:07:39 +010014029cpu_calls : integer
14030 Returns the number of calls to the task processing the stream or current
14031 request since it was allocated. This number is reset for each new request on
14032 the same connections in case of HTTP keep-alive. This value should usually be
14033 low and stable (around 2 calls for a typically simple request) but may become
14034 high if some processing (compression, caching or analysis) is performed. This
14035 is purely for performance monitoring purposes.
14036
14037cpu_ns_avg : integer
14038 Returns the average number of nanoseconds spent in each call to the task
14039 processing the stream or current request. This number is reset for each new
14040 request on the same connections in case of HTTP keep-alive. This value
14041 indicates the overall cost of processing the request or the connection for
14042 each call. There is no good nor bad value but the time spent in a call
14043 automatically causes latency for other processing (see lat_ns_avg below),
14044 and may affect other connection's apparent response time. Certain operations
14045 like compression, complex regex matching or heavy Lua operations may directly
14046 affect this value, and having it in the logs will make it easier to spot the
14047 faulty processing that needs to be fixed to recover decent performance.
14048 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14049
14050cpu_ns_tot : integer
14051 Returns the total number of nanoseconds spent in each call to the task
14052 processing the stream or current request. This number is reset for each new
14053 request on the same connections in case of HTTP keep-alive. This value
14054 indicates the overall cost of processing the request or the connection for
14055 each call. There is no good nor bad value but the time spent in a call
14056 automatically causes latency for other processing (see lat_ns_avg below),
14057 induces CPU costs on the machine, and may affect other connection's apparent
14058 response time. Certain operations like compression, complex regex matching or
14059 heavy Lua operations may directly affect this value, and having it in the
14060 logs will make it easier to spot the faulty processing that needs to be fixed
14061 to recover decent performance. The value may be artificially high due to a
14062 high cpu_calls count, for example when processing many HTTP chunks, and for
14063 this reason it is often preferred to log cpu_ns_avg instead.
14064
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014065date([<offset>]) : integer
14066 Returns the current date as the epoch (number of seconds since 01/01/1970).
14067 If an offset value is specified, then it is a number of seconds that is added
14068 to the current date before returning the value. This is particularly useful
14069 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014070 It is useful combined with the http_date converter.
14071
14072 Example :
14073
14074 # set an expires header to now+1 hour in every response
14075 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014076
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014077date_us : integer
14078 Return the microseconds part of the date (the "second" part is returned by
14079 date sample). This sample is coherent with the date sample as it is comes
14080 from the same timeval structure.
14081
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014082distcc_body(<token>[,<occ>]) : binary
14083 Parses a distcc message and returns the body associated to occurrence #<occ>
14084 of the token <token>. Occurrences start at 1, and when unspecified, any may
14085 match though in practice only the first one is checked for now. This can be
14086 used to extract file names or arguments in files built using distcc through
14087 haproxy. Please refer to distcc's protocol documentation for the complete
14088 list of supported tokens.
14089
14090distcc_param(<token>[,<occ>]) : integer
14091 Parses a distcc message and returns the parameter associated to occurrence
14092 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14093 any may match though in practice only the first one is checked for now. This
14094 can be used to extract certain information such as the protocol version, the
14095 file size or the argument in files built using distcc through haproxy.
14096 Another use case consists in waiting for the start of the preprocessed file
14097 contents before connecting to the server to avoid keeping idle connections.
14098 Please refer to distcc's protocol documentation for the complete list of
14099 supported tokens.
14100
14101 Example :
14102 # wait up to 20s for the pre-processed file to be uploaded
14103 tcp-request inspect-delay 20s
14104 tcp-request content accept if { distcc_param(DOTI) -m found }
14105 # send large files to the big farm
14106 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14107
Willy Tarreau595ec542013-06-12 21:34:28 +020014108env(<name>) : string
14109 Returns a string containing the value of environment variable <name>. As a
14110 reminder, environment variables are per-process and are sampled when the
14111 process starts. This can be useful to pass some information to a next hop
14112 server, or with ACLs to take specific action when the process is started a
14113 certain way.
14114
14115 Examples :
14116 # Pass the Via header to next hop with the local hostname in it
14117 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14118
14119 # reject cookie-less requests when the STOP environment variable is set
14120 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14121
Willy Tarreau74ca5042013-06-11 23:12:07 +020014122fe_conn([<frontend>]) : integer
14123 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014124 possibly including the connection being evaluated. If no frontend name is
14125 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014126 frontend. It can be used to return a sorry page before hard-blocking, or to
14127 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014128 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014129 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14130 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014131
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014132fe_req_rate([<frontend>]) : integer
14133 Returns an integer value corresponding to the number of HTTP requests per
14134 second sent to a frontend. This number can differ from "fe_sess_rate" in
14135 situations where client-side keep-alive is enabled.
14136
Willy Tarreau74ca5042013-06-11 23:12:07 +020014137fe_sess_rate([<frontend>]) : integer
14138 Returns an integer value corresponding to the sessions creation rate on the
14139 frontend, in number of new sessions per second. This is used with ACLs to
14140 limit the incoming session rate to an acceptable range in order to prevent
14141 abuse of service at the earliest moment, for example when combined with other
14142 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14143 down below the limit. It can also be useful to add this element to logs using
14144 a log-format directive. See also the "rate-limit sessions" directive for use
14145 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014146
14147 Example :
14148 # This frontend limits incoming mails to 10/s with a max of 100
14149 # concurrent connections. We accept any connection below 10/s, and
14150 # force excess clients to wait for 100 ms. Since clients are limited to
14151 # 100 max, there cannot be more than 10 incoming mails per second.
14152 frontend mail
14153 bind :25
14154 mode tcp
14155 maxconn 100
14156 acl too_fast fe_sess_rate ge 10
14157 tcp-request inspect-delay 100ms
14158 tcp-request content accept if ! too_fast
14159 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014160
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014161hostname : string
14162 Returns the system hostname.
14163
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014164int(<integer>) : signed integer
14165 Returns a signed integer.
14166
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014167ipv4(<ipv4>) : ipv4
14168 Returns an ipv4.
14169
14170ipv6(<ipv6>) : ipv6
14171 Returns an ipv6.
14172
Willy Tarreau70fe9442018-11-22 16:07:39 +010014173lat_ns_avg : integer
14174 Returns the average number of nanoseconds spent between the moment the task
14175 handling the stream is woken up and the moment it is effectively called. This
14176 number is reset for each new request on the same connections in case of HTTP
14177 keep-alive. This value indicates the overall latency inflicted to the current
14178 request by all other requests being processed in parallel, and is a direct
14179 indicator of perceived performance due to noisy neighbours. In order to keep
14180 the value low, it is possible to reduce the scheduler's run queue depth using
14181 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14182 once using "tune.maxpollevents", to decrease the stream's nice value using
14183 the "nice" option on the "bind" lines or in the frontend, or to look for
14184 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14185 whose processing needs to be adjusted or fixed. Compression of large buffers
14186 could be a culprit, like heavy regex or long lists of regex.
14187 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14188
14189lat_ns_tot : integer
14190 Returns the total number of nanoseconds spent between the moment the task
14191 handling the stream is woken up and the moment it is effectively called. This
14192 number is reset for each new request on the same connections in case of HTTP
14193 keep-alive. This value indicates the overall latency inflicted to the current
14194 request by all other requests being processed in parallel, and is a direct
14195 indicator of perceived performance due to noisy neighbours. In order to keep
14196 the value low, it is possible to reduce the scheduler's run queue depth using
14197 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14198 once using "tune.maxpollevents", to decrease the stream's nice value using
14199 the "nice" option on the "bind" lines or in the frontend, or to look for
14200 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14201 whose processing needs to be adjusted or fixed. Compression of large buffers
14202 could be a culprit, like heavy regex or long lists of regex. Note: while it
14203 may intuitively seem that the total latency adds to a transfer time, it is
14204 almost never true because while a task waits for the CPU, network buffers
14205 continue to fill up and the next call will process more at once. The value
14206 may be artificially high due to a high cpu_calls count, for example when
14207 processing many HTTP chunks, and for this reason it is often preferred to log
14208 lat_ns_avg instead, which is a more relevant performance indicator.
14209
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014210meth(<method>) : method
14211 Returns a method.
14212
Willy Tarreau0f30d262014-11-24 16:02:05 +010014213nbproc : integer
14214 Returns an integer value corresponding to the number of processes that were
14215 started (it equals the global "nbproc" setting). This is useful for logging
14216 and debugging purposes.
14217
Willy Tarreau74ca5042013-06-11 23:12:07 +020014218nbsrv([<backend>]) : integer
14219 Returns an integer value corresponding to the number of usable servers of
14220 either the current backend or the named backend. This is mostly used with
14221 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014222 switch to an alternate backend when the number of servers is too low to
14223 to handle some load. It is useful to report a failure when combined with
14224 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014225
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014226prio_class : integer
14227 Returns the priority class of the current session for http mode or connection
14228 for tcp mode. The value will be that set by the last call to "http-request
14229 set-priority-class" or "tcp-request content set-priority-class".
14230
14231prio_offset : integer
14232 Returns the priority offset of the current session for http mode or
14233 connection for tcp mode. The value will be that set by the last call to
14234 "http-request set-priority-offset" or "tcp-request content
14235 set-priority-offset".
14236
Willy Tarreau0f30d262014-11-24 16:02:05 +010014237proc : integer
14238 Returns an integer value corresponding to the position of the process calling
14239 the function, between 1 and global.nbproc. This is useful for logging and
14240 debugging purposes.
14241
Willy Tarreau74ca5042013-06-11 23:12:07 +020014242queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014243 Returns the total number of queued connections of the designated backend,
14244 including all the connections in server queues. If no backend name is
14245 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014246 one. This is useful with ACLs or to pass statistics to backend servers. This
14247 can be used to take actions when queuing goes above a known level, generally
14248 indicating a surge of traffic or a massive slowdown on the servers. One
14249 possible action could be to reject new users but still accept old ones. See
14250 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14251
Willy Tarreau84310e22014-02-14 11:59:04 +010014252rand([<range>]) : integer
14253 Returns a random integer value within a range of <range> possible values,
14254 starting at zero. If the range is not specified, it defaults to 2^32, which
14255 gives numbers between 0 and 4294967295. It can be useful to pass some values
14256 needed to take some routing decisions for example, or just for debugging
14257 purposes. This random must not be used for security purposes.
14258
Luca Schimweg8a694b82019-09-10 15:42:52 +020014259uuid([<version>]) : string
14260 Returns a UUID following the RFC4122 standard. If the version is not
14261 specified, a UUID version 4 (fully random) is returned.
14262 Currently, only version 4 is supported.
14263
Willy Tarreau74ca5042013-06-11 23:12:07 +020014264srv_conn([<backend>/]<server>) : integer
14265 Returns an integer value corresponding to the number of currently established
14266 connections on the designated server, possibly including the connection being
14267 evaluated. If <backend> is omitted, then the server is looked up in the
14268 current backend. It can be used to use a specific farm when one server is
14269 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014270 connections with it. See also the "fe_conn", "be_conn", "queue", and
14271 "srv_conn_free" fetch methods.
14272
14273srv_conn_free([<backend>/]<server>) : integer
14274 Returns an integer value corresponding to the number of available connections
14275 on the designated server, possibly including the connection being evaluated.
14276 The value does not include queue slots. If <backend> is omitted, then the
14277 server is looked up in the current backend. It can be used to use a specific
14278 farm when one server is full, or to inform the server about our view of the
14279 number of active connections with it. See also the "be_conn_free" and
14280 "srv_conn" fetch methods.
14281
14282 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14283 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014284
14285srv_is_up([<backend>/]<server>) : boolean
14286 Returns true when the designated server is UP, and false when it is either
14287 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14288 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014289 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014290 availability). Another possible use which is more of a hack consists in
14291 using dummy servers as boolean variables that can be enabled or disabled from
14292 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14293
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014294srv_queue([<backend>/]<server>) : integer
14295 Returns an integer value corresponding to the number of connections currently
14296 pending in the designated server's queue. If <backend> is omitted, then the
14297 server is looked up in the current backend. It can sometimes be used together
14298 with the "use-server" directive to force to use a known faster server when it
14299 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14300 fetch methods.
14301
Willy Tarreau74ca5042013-06-11 23:12:07 +020014302srv_sess_rate([<backend>/]<server>) : integer
14303 Returns an integer corresponding to the sessions creation rate on the
14304 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014305 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014306 used with ACLs but can make sense with logs too. This is used to switch to an
14307 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014308 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014309 overloading servers).
14310
14311 Example :
14312 # Redirect to a separate back
14313 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14314 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14315 use_backend be2 if srv1_full or srv2_full
14316
Willy Tarreau0f30d262014-11-24 16:02:05 +010014317stopping : boolean
14318 Returns TRUE if the process calling the function is currently stopping. This
14319 can be useful for logging, or for relaxing certain checks or helping close
14320 certain connections upon graceful shutdown.
14321
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014322str(<string>) : string
14323 Returns a string.
14324
Willy Tarreau74ca5042013-06-11 23:12:07 +020014325table_avl([<table>]) : integer
14326 Returns the total number of available entries in the current proxy's
14327 stick-table or in the designated stick-table. See also table_cnt.
14328
14329table_cnt([<table>]) : integer
14330 Returns the total number of entries currently in use in the current proxy's
14331 stick-table or in the designated stick-table. See also src_conn_cnt and
14332 table_avl for other entry counting methods.
14333
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014334thread : integer
14335 Returns an integer value corresponding to the position of the thread calling
14336 the function, between 0 and (global.nbthread-1). This is useful for logging
14337 and debugging purposes.
14338
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014339var(<var-name>) : undefined
14340 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014341 sample fetch fails. The name of the variable starts with an indication
14342 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014343 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014344 "sess" : the variable is shared with the whole session
14345 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014346 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014347 "req" : the variable is shared only during request processing,
14348 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014349 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014350 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014351
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200143527.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014353----------------------------------
14354
14355The layer 4 usually describes just the transport layer which in haproxy is
14356closest to the connection, where no content is yet made available. The fetch
14357methods described here are usable as low as the "tcp-request connection" rule
14358sets unless they require some future information. Those generally include
14359TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014360the incoming connection. For retrieving a value from a sticky counters, the
14361counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014362"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14363used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14364can be specified as the first integer argument when using the "sc_" prefix.
14365Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14366table may be specified with the "sc*" form, in which case the currently
14367tracked key will be looked up into this alternate table instead of the table
14368currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014369
Jérôme Magnin35e53a62019-01-16 14:38:37 +010014370bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010014371 Returns the backend connection's HTTP major version encoding, which may be 1
14372 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14373 encoding and not the version present in the request header.
14374
Willy Tarreau74ca5042013-06-11 23:12:07 +020014375be_id : integer
14376 Returns an integer containing the current backend's id. It can be used in
14377 frontends with responses to check which backend processed the request.
14378
Marcin Deranekd2471c22016-12-12 14:08:05 +010014379be_name : string
14380 Returns a string containing the current backend's name. It can be used in
14381 frontends with responses to check which backend processed the request.
14382
Willy Tarreau74ca5042013-06-11 23:12:07 +020014383dst : ip
14384 This is the destination IPv4 address of the connection on the client side,
14385 which is the address the client connected to. It can be useful when running
14386 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14387 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014388 RFC 4291. When the incoming connection passed through address translation or
14389 redirection involving connection tracking, the original destination address
14390 before the redirection will be reported. On Linux systems, the source and
14391 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
14392 is set, because a late response may reopen a timed out connection and switch
14393 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014394
14395dst_conn : integer
14396 Returns an integer value corresponding to the number of currently established
14397 connections on the same socket including the one being evaluated. It is
14398 normally used with ACLs but can as well be used to pass the information to
14399 servers in an HTTP header or in logs. It can be used to either return a sorry
14400 page before hard-blocking, or to use a specific backend to drain new requests
14401 when the socket is considered saturated. This offers the ability to assign
14402 different limits to different listening ports or addresses. See also the
14403 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014404
Willy Tarreau16e01562016-08-09 16:46:18 +020014405dst_is_local : boolean
14406 Returns true if the destination address of the incoming connection is local
14407 to the system, or false if the address doesn't exist on the system, meaning
14408 that it was intercepted in transparent mode. It can be useful to apply
14409 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014410 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014411 be delivered only on this address, or SSH access could be locally redirected.
14412 Please note that the check involves a few system calls, so it's better to do
14413 it only once per connection.
14414
Willy Tarreau74ca5042013-06-11 23:12:07 +020014415dst_port : integer
14416 Returns an integer value corresponding to the destination TCP port of the
14417 connection on the client side, which is the port the client connected to.
14418 This might be used when running in transparent mode, when assigning dynamic
14419 ports to some clients for a whole application session, to stick all users to
14420 a same server, or to pass the destination port information to a server using
14421 an HTTP header.
14422
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014423fc_http_major : integer
14424 Reports the front connection's HTTP major version encoding, which may be 1
14425 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14426 encoding and not on the version present in the request header.
14427
Geoff Simmons7185b782019-08-27 18:31:16 +020014428fc_pp_authority : string
14429 Returns the authority TLV sent by the client in the PROXY protocol header,
14430 if any.
14431
Emeric Brun4f603012017-01-05 15:11:44 +010014432fc_rcvd_proxy : boolean
14433 Returns true if the client initiated the connection with a PROXY protocol
14434 header.
14435
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014436fc_rtt(<unit>) : integer
14437 Returns the Round Trip Time (RTT) measured by the kernel for the client
14438 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14439 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14440 connection is not established, if the connection is not TCP or if the
14441 operating system does not support TCP_INFO, for example Linux kernels before
14442 2.4, the sample fetch fails.
14443
14444fc_rttvar(<unit>) : integer
14445 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14446 client connection. <unit> is facultative, by default the unit is milliseconds.
14447 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14448 server connection is not established, if the connection is not TCP or if the
14449 operating system does not support TCP_INFO, for example Linux kernels before
14450 2.4, the sample fetch fails.
14451
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014452fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014453 Returns the unacked counter measured by the kernel for the client connection.
14454 If the server connection is not established, if the connection is not TCP or
14455 if the operating system does not support TCP_INFO, for example Linux kernels
14456 before 2.4, the sample fetch fails.
14457
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014458fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014459 Returns the sacked counter measured by the kernel for the client connection.
14460 If the server connection is not established, if the connection is not TCP or
14461 if the operating system does not support TCP_INFO, for example Linux kernels
14462 before 2.4, the sample fetch fails.
14463
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014464fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014465 Returns the retransmits counter measured by the kernel for the client
14466 connection. If the server connection is not established, if the connection is
14467 not TCP or if the operating system does not support TCP_INFO, for example
14468 Linux kernels before 2.4, the sample fetch fails.
14469
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014470fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014471 Returns the fack counter measured by the kernel for the client
14472 connection. If the server connection is not established, if the connection is
14473 not TCP or if the operating system does not support TCP_INFO, for example
14474 Linux kernels before 2.4, the sample fetch fails.
14475
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014476fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014477 Returns the lost counter measured by the kernel for the client
14478 connection. If the server connection is not established, if the connection is
14479 not TCP or if the operating system does not support TCP_INFO, for example
14480 Linux kernels before 2.4, the sample fetch fails.
14481
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014482fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014483 Returns the reordering counter measured by the kernel for the client
14484 connection. If the server connection is not established, if the connection is
14485 not TCP or if the operating system does not support TCP_INFO, for example
14486 Linux kernels before 2.4, the sample fetch fails.
14487
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014488fe_defbe : string
14489 Returns a string containing the frontend's default backend name. It can be
14490 used in frontends to check which backend will handle requests by default.
14491
Willy Tarreau74ca5042013-06-11 23:12:07 +020014492fe_id : integer
14493 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014494 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014495 coming via a same frontend to the same server.
14496
Marcin Deranekd2471c22016-12-12 14:08:05 +010014497fe_name : string
14498 Returns a string containing the current frontend's name. It can be used in
14499 backends to check from which frontend it was called, or to stick all users
14500 coming via a same frontend to the same server.
14501
Cyril Bonté62ba8702014-04-22 23:52:25 +020014502sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014503sc0_bytes_in_rate([<table>]) : integer
14504sc1_bytes_in_rate([<table>]) : integer
14505sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014506 Returns the average client-to-server bytes rate from the currently tracked
14507 counters, measured in amount of bytes over the period configured in the
14508 table. See also src_bytes_in_rate.
14509
Cyril Bonté62ba8702014-04-22 23:52:25 +020014510sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014511sc0_bytes_out_rate([<table>]) : integer
14512sc1_bytes_out_rate([<table>]) : integer
14513sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014514 Returns the average server-to-client bytes rate from the currently tracked
14515 counters, measured in amount of bytes over the period configured in the
14516 table. See also src_bytes_out_rate.
14517
Cyril Bonté62ba8702014-04-22 23:52:25 +020014518sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014519sc0_clr_gpc0([<table>]) : integer
14520sc1_clr_gpc0([<table>]) : integer
14521sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014522 Clears the first General Purpose Counter associated to the currently tracked
14523 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014524 stored value is zero, so first invocation will always return zero. This is
14525 typically used as a second ACL in an expression in order to mark a connection
14526 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014527
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014528 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014529 # block if 5 consecutive requests continue to come faster than 10 sess
14530 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014531 acl abuse sc0_http_req_rate gt 10
14532 acl kill sc0_inc_gpc0 gt 5
14533 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014534 tcp-request connection accept if !abuse save
14535 tcp-request connection reject if abuse kill
14536
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014537sc_clr_gpc1(<ctr>[,<table>]) : integer
14538sc0_clr_gpc1([<table>]) : integer
14539sc1_clr_gpc1([<table>]) : integer
14540sc2_clr_gpc1([<table>]) : integer
14541 Clears the second General Purpose Counter associated to the currently tracked
14542 counters, and returns its previous value. Before the first invocation, the
14543 stored value is zero, so first invocation will always return zero. This is
14544 typically used as a second ACL in an expression in order to mark a connection
14545 when a first ACL was verified.
14546
Cyril Bonté62ba8702014-04-22 23:52:25 +020014547sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014548sc0_conn_cnt([<table>]) : integer
14549sc1_conn_cnt([<table>]) : integer
14550sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014551 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014552 counters. See also src_conn_cnt.
14553
Cyril Bonté62ba8702014-04-22 23:52:25 +020014554sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014555sc0_conn_cur([<table>]) : integer
14556sc1_conn_cur([<table>]) : integer
14557sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014558 Returns the current amount of concurrent connections tracking the same
14559 tracked counters. This number is automatically incremented when tracking
14560 begins and decremented when tracking stops. See also src_conn_cur.
14561
Cyril Bonté62ba8702014-04-22 23:52:25 +020014562sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014563sc0_conn_rate([<table>]) : integer
14564sc1_conn_rate([<table>]) : integer
14565sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014566 Returns the average connection rate from the currently tracked counters,
14567 measured in amount of connections over the period configured in the table.
14568 See also src_conn_rate.
14569
Cyril Bonté62ba8702014-04-22 23:52:25 +020014570sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014571sc0_get_gpc0([<table>]) : integer
14572sc1_get_gpc0([<table>]) : integer
14573sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014574 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014575 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014576
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014577sc_get_gpc1(<ctr>[,<table>]) : integer
14578sc0_get_gpc1([<table>]) : integer
14579sc1_get_gpc1([<table>]) : integer
14580sc2_get_gpc1([<table>]) : integer
14581 Returns the value of the second General Purpose Counter associated to the
14582 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14583
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014584sc_get_gpt0(<ctr>[,<table>]) : integer
14585sc0_get_gpt0([<table>]) : integer
14586sc1_get_gpt0([<table>]) : integer
14587sc2_get_gpt0([<table>]) : integer
14588 Returns the value of the first General Purpose Tag associated to the
14589 currently tracked counters. See also src_get_gpt0.
14590
Cyril Bonté62ba8702014-04-22 23:52:25 +020014591sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014592sc0_gpc0_rate([<table>]) : integer
14593sc1_gpc0_rate([<table>]) : integer
14594sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014595 Returns the average increment rate of the first General Purpose Counter
14596 associated to the currently tracked counters. It reports the frequency
14597 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014598 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14599 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14600 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014601
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014602sc_gpc1_rate(<ctr>[,<table>]) : integer
14603sc0_gpc1_rate([<table>]) : integer
14604sc1_gpc1_rate([<table>]) : integer
14605sc2_gpc1_rate([<table>]) : integer
14606 Returns the average increment rate of the second General Purpose Counter
14607 associated to the currently tracked counters. It reports the frequency
14608 which the gpc1 counter was incremented over the configured period. See also
14609 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14610 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14611 be returned, as "gpc1" only holds the event count.
14612
Cyril Bonté62ba8702014-04-22 23:52:25 +020014613sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014614sc0_http_err_cnt([<table>]) : integer
14615sc1_http_err_cnt([<table>]) : integer
14616sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014617 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014618 counters. This includes the both request errors and 4xx error responses.
14619 See also src_http_err_cnt.
14620
Cyril Bonté62ba8702014-04-22 23:52:25 +020014621sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014622sc0_http_err_rate([<table>]) : integer
14623sc1_http_err_rate([<table>]) : integer
14624sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014625 Returns the average rate of HTTP errors from the currently tracked counters,
14626 measured in amount of errors over the period configured in the table. This
14627 includes the both request errors and 4xx error responses. See also
14628 src_http_err_rate.
14629
Cyril Bonté62ba8702014-04-22 23:52:25 +020014630sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014631sc0_http_req_cnt([<table>]) : integer
14632sc1_http_req_cnt([<table>]) : integer
14633sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014634 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014635 counters. This includes every started request, valid or not. See also
14636 src_http_req_cnt.
14637
Cyril Bonté62ba8702014-04-22 23:52:25 +020014638sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014639sc0_http_req_rate([<table>]) : integer
14640sc1_http_req_rate([<table>]) : integer
14641sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014642 Returns the average rate of HTTP requests from the currently tracked
14643 counters, measured in amount of requests over the period configured in
14644 the table. This includes every started request, valid or not. See also
14645 src_http_req_rate.
14646
Cyril Bonté62ba8702014-04-22 23:52:25 +020014647sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014648sc0_inc_gpc0([<table>]) : integer
14649sc1_inc_gpc0([<table>]) : integer
14650sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014651 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014652 tracked counters, and returns its new value. Before the first invocation,
14653 the stored value is zero, so first invocation will increase it to 1 and will
14654 return 1. This is typically used as a second ACL in an expression in order
14655 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014656
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014657 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014658 acl abuse sc0_http_req_rate gt 10
14659 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014660 tcp-request connection reject if abuse kill
14661
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014662sc_inc_gpc1(<ctr>[,<table>]) : integer
14663sc0_inc_gpc1([<table>]) : integer
14664sc1_inc_gpc1([<table>]) : integer
14665sc2_inc_gpc1([<table>]) : integer
14666 Increments the second General Purpose Counter associated to the currently
14667 tracked counters, and returns its new value. Before the first invocation,
14668 the stored value is zero, so first invocation will increase it to 1 and will
14669 return 1. This is typically used as a second ACL in an expression in order
14670 to mark a connection when a first ACL was verified.
14671
Cyril Bonté62ba8702014-04-22 23:52:25 +020014672sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014673sc0_kbytes_in([<table>]) : integer
14674sc1_kbytes_in([<table>]) : integer
14675sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014676 Returns the total amount of client-to-server data from the currently tracked
14677 counters, measured in kilobytes. The test is currently performed on 32-bit
14678 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014679
Cyril Bonté62ba8702014-04-22 23:52:25 +020014680sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014681sc0_kbytes_out([<table>]) : integer
14682sc1_kbytes_out([<table>]) : integer
14683sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014684 Returns the total amount of server-to-client data from the currently tracked
14685 counters, measured in kilobytes. The test is currently performed on 32-bit
14686 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014687
Cyril Bonté62ba8702014-04-22 23:52:25 +020014688sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014689sc0_sess_cnt([<table>]) : integer
14690sc1_sess_cnt([<table>]) : integer
14691sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014692 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014693 into sessions, which means that they were accepted by a "tcp-request
14694 connection" rule, from the currently tracked counters. A backend may count
14695 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014696 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014697 with the client. See also src_sess_cnt.
14698
Cyril Bonté62ba8702014-04-22 23:52:25 +020014699sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014700sc0_sess_rate([<table>]) : integer
14701sc1_sess_rate([<table>]) : integer
14702sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014703 Returns the average session rate from the currently tracked counters,
14704 measured in amount of sessions over the period configured in the table. A
14705 session is a connection that got past the early "tcp-request connection"
14706 rules. A backend may count more sessions than connections because each
14707 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014708 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014709
Cyril Bonté62ba8702014-04-22 23:52:25 +020014710sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014711sc0_tracked([<table>]) : boolean
14712sc1_tracked([<table>]) : boolean
14713sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014714 Returns true if the designated session counter is currently being tracked by
14715 the current session. This can be useful when deciding whether or not we want
14716 to set some values in a header passed to the server.
14717
Cyril Bonté62ba8702014-04-22 23:52:25 +020014718sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014719sc0_trackers([<table>]) : integer
14720sc1_trackers([<table>]) : integer
14721sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014722 Returns the current amount of concurrent connections tracking the same
14723 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014724 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014725 that it does not rely on any stored information but on the table's reference
14726 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014727 may sometimes be more suited for layer7 tracking. It can be used to tell a
14728 server how many concurrent connections there are from a given address for
14729 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014730
Willy Tarreau74ca5042013-06-11 23:12:07 +020014731so_id : integer
14732 Returns an integer containing the current listening socket's id. It is useful
14733 in frontends involving many "bind" lines, or to stick all users coming via a
14734 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014735
Willy Tarreau74ca5042013-06-11 23:12:07 +020014736src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014737 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014738 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14739 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14740 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014741 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14742 directive is used, it can be the address of a client behind another
14743 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014744 "tcp-request connection" which sees the real address. When the incoming
14745 connection passed through address translation or redirection involving
14746 connection tracking, the original destination address before the redirection
14747 will be reported. On Linux systems, the source and destination may seldom
14748 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
14749 response may reopen a timed out connection and switch what is believed to be
14750 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014751
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014752 Example:
14753 # add an HTTP header in requests with the originating address' country
14754 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14755
Willy Tarreau74ca5042013-06-11 23:12:07 +020014756src_bytes_in_rate([<table>]) : integer
14757 Returns the average bytes rate from the incoming connection's source address
14758 in the current proxy's stick-table or in the designated stick-table, measured
14759 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014760 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014761
Willy Tarreau74ca5042013-06-11 23:12:07 +020014762src_bytes_out_rate([<table>]) : integer
14763 Returns the average bytes rate to the incoming connection's source address in
14764 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014765 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014766 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014767
Willy Tarreau74ca5042013-06-11 23:12:07 +020014768src_clr_gpc0([<table>]) : integer
14769 Clears the first General Purpose Counter associated to the incoming
14770 connection's source address in the current proxy's stick-table or in the
14771 designated stick-table, and returns its previous value. If the address is not
14772 found, an entry is created and 0 is returned. This is typically used as a
14773 second ACL in an expression in order to mark a connection when a first ACL
14774 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014775
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014776 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014777 # block if 5 consecutive requests continue to come faster than 10 sess
14778 # per second, and reset the counter as soon as the traffic slows down.
14779 acl abuse src_http_req_rate gt 10
14780 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014781 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014782 tcp-request connection accept if !abuse save
14783 tcp-request connection reject if abuse kill
14784
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014785src_clr_gpc1([<table>]) : integer
14786 Clears the second General Purpose Counter associated to the incoming
14787 connection's source address in the current proxy's stick-table or in the
14788 designated stick-table, and returns its previous value. If the address is not
14789 found, an entry is created and 0 is returned. This is typically used as a
14790 second ACL in an expression in order to mark a connection when a first ACL
14791 was verified.
14792
Willy Tarreau74ca5042013-06-11 23:12:07 +020014793src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014794 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014795 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014796 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014797 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014798
Willy Tarreau74ca5042013-06-11 23:12:07 +020014799src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014800 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014801 current incoming connection's source address in the current proxy's
14802 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014803 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014804
Willy Tarreau74ca5042013-06-11 23:12:07 +020014805src_conn_rate([<table>]) : integer
14806 Returns the average connection rate from the incoming connection's source
14807 address in the current proxy's stick-table or in the designated stick-table,
14808 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014809 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014810
Willy Tarreau74ca5042013-06-11 23:12:07 +020014811src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014812 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014813 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014814 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014815 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014816
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014817src_get_gpc1([<table>]) : integer
14818 Returns the value of the second General Purpose Counter associated to the
14819 incoming connection's source address in the current proxy's stick-table or in
14820 the designated stick-table. If the address is not found, zero is returned.
14821 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14822
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014823src_get_gpt0([<table>]) : integer
14824 Returns the value of the first General Purpose Tag associated to the
14825 incoming connection's source address in the current proxy's stick-table or in
14826 the designated stick-table. If the address is not found, zero is returned.
14827 See also sc/sc0/sc1/sc2_get_gpt0.
14828
Willy Tarreau74ca5042013-06-11 23:12:07 +020014829src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014830 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014831 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014832 stick-table or in the designated stick-table. It reports the frequency
14833 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014834 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14835 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14836 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014837
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014838src_gpc1_rate([<table>]) : integer
14839 Returns the average increment rate of the second General Purpose Counter
14840 associated to the incoming connection's source address in the current proxy's
14841 stick-table or in the designated stick-table. It reports the frequency
14842 which the gpc1 counter was incremented over the configured period. See also
14843 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14844 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14845 be returned, as "gpc1" only holds the event count.
14846
Willy Tarreau74ca5042013-06-11 23:12:07 +020014847src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014848 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014849 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014850 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014851 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014852 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014853
Willy Tarreau74ca5042013-06-11 23:12:07 +020014854src_http_err_rate([<table>]) : integer
14855 Returns the average rate of HTTP errors from the incoming connection's source
14856 address in the current proxy's stick-table or in the designated stick-table,
14857 measured in amount of errors over the period configured in the table. This
14858 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014859 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014860
Willy Tarreau74ca5042013-06-11 23:12:07 +020014861src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014862 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014863 source address in the current proxy's stick-table or in the designated stick-
14864 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014865 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014866
Willy Tarreau74ca5042013-06-11 23:12:07 +020014867src_http_req_rate([<table>]) : integer
14868 Returns the average rate of HTTP requests from the incoming connection's
14869 source address in the current proxy's stick-table or in the designated stick-
14870 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014871 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014872 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014873
Willy Tarreau74ca5042013-06-11 23:12:07 +020014874src_inc_gpc0([<table>]) : integer
14875 Increments the first General Purpose Counter associated to the incoming
14876 connection's source address in the current proxy's stick-table or in the
14877 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014878 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014879 This is typically used as a second ACL in an expression in order to mark a
14880 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014881
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014882 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014883 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014884 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014885 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014886
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014887src_inc_gpc1([<table>]) : integer
14888 Increments the second General Purpose Counter associated to the incoming
14889 connection's source address in the current proxy's stick-table or in the
14890 designated stick-table, and returns its new value. If the address is not
14891 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14892 This is typically used as a second ACL in an expression in order to mark a
14893 connection when a first ACL was verified.
14894
Willy Tarreau16e01562016-08-09 16:46:18 +020014895src_is_local : boolean
14896 Returns true if the source address of the incoming connection is local to the
14897 system, or false if the address doesn't exist on the system, meaning that it
14898 comes from a remote machine. Note that UNIX addresses are considered local.
14899 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014900 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014901 note that the check involves a few system calls, so it's better to do it only
14902 once per connection.
14903
Willy Tarreau74ca5042013-06-11 23:12:07 +020014904src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014905 Returns the total amount of data received from the incoming connection's
14906 source address in the current proxy's stick-table or in the designated
14907 stick-table, measured in kilobytes. If the address is not found, zero is
14908 returned. The test is currently performed on 32-bit integers, which limits
14909 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014910
Willy Tarreau74ca5042013-06-11 23:12:07 +020014911src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014912 Returns the total amount of data sent to the incoming connection's source
14913 address in the current proxy's stick-table or in the designated stick-table,
14914 measured in kilobytes. If the address is not found, zero is returned. The
14915 test is currently performed on 32-bit integers, which limits values to 4
14916 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014917
Willy Tarreau74ca5042013-06-11 23:12:07 +020014918src_port : integer
14919 Returns an integer value corresponding to the TCP source port of the
14920 connection on the client side, which is the port the client connected from.
14921 Usage of this function is very limited as modern protocols do not care much
14922 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014923
Willy Tarreau74ca5042013-06-11 23:12:07 +020014924src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014925 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020014926 connection's source IPv4 address in the current proxy's stick-table or in the
14927 designated stick-table, that were transformed into sessions, which means that
14928 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014929 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014930
Willy Tarreau74ca5042013-06-11 23:12:07 +020014931src_sess_rate([<table>]) : integer
14932 Returns the average session rate from the incoming connection's source
14933 address in the current proxy's stick-table or in the designated stick-table,
14934 measured in amount of sessions over the period configured in the table. A
14935 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014936 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014937
Willy Tarreau74ca5042013-06-11 23:12:07 +020014938src_updt_conn_cnt([<table>]) : integer
14939 Creates or updates the entry associated to the incoming connection's source
14940 address in the current proxy's stick-table or in the designated stick-table.
14941 This table must be configured to store the "conn_cnt" data type, otherwise
14942 the match will be ignored. The current count is incremented by one, and the
14943 expiration timer refreshed. The updated count is returned, so this match
14944 can't return zero. This was used to reject service abusers based on their
14945 source address. Note: it is recommended to use the more complete "track-sc*"
14946 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014947
14948 Example :
14949 # This frontend limits incoming SSH connections to 3 per 10 second for
14950 # each source address, and rejects excess connections until a 10 second
14951 # silence is observed. At most 20 addresses are tracked.
14952 listen ssh
14953 bind :22
14954 mode tcp
14955 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020014956 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020014957 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014958 server local 127.0.0.1:22
14959
Willy Tarreau74ca5042013-06-11 23:12:07 +020014960srv_id : integer
14961 Returns an integer containing the server's id when processing the response.
14962 While it's almost only used with ACLs, it may be used for logging or
14963 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020014964
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200149657.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020014966----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020014967
Willy Tarreau74ca5042013-06-11 23:12:07 +020014968The layer 5 usually describes just the session layer which in haproxy is
14969closest to the session once all the connection handshakes are finished, but
14970when no content is yet made available. The fetch methods described here are
14971usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014972future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020014973
Ben Shillitof25e8e52016-12-02 14:25:37 +00001497451d.all(<prop>[,<prop>*]) : string
14975 Returns values for the properties requested as a string, where values are
14976 separated by the delimiter specified with "51degrees-property-separator".
14977 The device is identified using all the important HTTP headers from the
14978 request. The function can be passed up to five property names, and if a
14979 property name can't be found, the value "NoData" is returned.
14980
14981 Example :
14982 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
14983 # containing the three properties requested using all relevant headers from
14984 # the request.
14985 frontend http-in
14986 bind *:8081
14987 default_backend servers
14988 http-request set-header X-51D-DeviceTypeMobileTablet \
14989 %[51d.all(DeviceType,IsMobile,IsTablet)]
14990
Emeric Brun645ae792014-04-30 14:21:06 +020014991ssl_bc : boolean
14992 Returns true when the back connection was made via an SSL/TLS transport
14993 layer and is locally deciphered. This means the outgoing connection was made
14994 other a server with the "ssl" option.
14995
14996ssl_bc_alg_keysize : integer
14997 Returns the symmetric cipher key size supported in bits when the outgoing
14998 connection was made over an SSL/TLS transport layer.
14999
Olivier Houchard6b77f492018-11-22 18:18:29 +010015000ssl_bc_alpn : string
15001 This extracts the Application Layer Protocol Negotiation field from an
15002 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020015003 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010015004 server. The SSL library must have been built with support for TLS
15005 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15006 not advertised unless the "alpn" keyword on the "server" line specifies a
15007 protocol list. Also, nothing forces the server to pick a protocol from this
15008 list, any other one may be requested. The TLS ALPN extension is meant to
15009 replace the TLS NPN extension. See also "ssl_bc_npn".
15010
Emeric Brun645ae792014-04-30 14:21:06 +020015011ssl_bc_cipher : string
15012 Returns the name of the used cipher when the outgoing connection was made
15013 over an SSL/TLS transport layer.
15014
Patrick Hemmer65674662019-06-04 08:13:03 -040015015ssl_bc_client_random : binary
15016 Returns the client random of the back connection when the incoming connection
15017 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15018 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15019
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015020ssl_bc_is_resumed : boolean
15021 Returns true when the back connection was made over an SSL/TLS transport
15022 layer and the newly created SSL session was resumed using a cached
15023 session or a TLS ticket.
15024
Olivier Houchard6b77f492018-11-22 18:18:29 +010015025ssl_bc_npn : string
15026 This extracts the Next Protocol Negotiation field from an outgoing connection
15027 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020015028 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010015029 built with support for TLS extensions enabled (check haproxy -vv). Note that
15030 the TLS NPN extension is not advertised unless the "npn" keyword on the
15031 "server" line specifies a protocol list. Also, nothing forces the server to
15032 pick a protocol from this list, any other one may be used. Please note that
15033 the TLS NPN extension was replaced with ALPN.
15034
Emeric Brun645ae792014-04-30 14:21:06 +020015035ssl_bc_protocol : string
15036 Returns the name of the used protocol when the outgoing connection was made
15037 over an SSL/TLS transport layer.
15038
Emeric Brunb73a9b02014-04-30 18:49:19 +020015039ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015040 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015041 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15042 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015043
Patrick Hemmer65674662019-06-04 08:13:03 -040015044ssl_bc_server_random : binary
15045 Returns the server random of the back connection when the incoming connection
15046 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15047 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15048
Emeric Brun645ae792014-04-30 14:21:06 +020015049ssl_bc_session_id : binary
15050 Returns the SSL ID of the back connection when the outgoing connection was
15051 made over an SSL/TLS transport layer. It is useful to log if we want to know
15052 if session was reused or not.
15053
Patrick Hemmere0275472018-04-28 19:15:51 -040015054ssl_bc_session_key : binary
15055 Returns the SSL session master key of the back connection when the outgoing
15056 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15057 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15058 BoringSSL.
15059
Emeric Brun645ae792014-04-30 14:21:06 +020015060ssl_bc_use_keysize : integer
15061 Returns the symmetric cipher key size used in bits when the outgoing
15062 connection was made over an SSL/TLS transport layer.
15063
Willy Tarreau74ca5042013-06-11 23:12:07 +020015064ssl_c_ca_err : integer
15065 When the incoming connection was made over an SSL/TLS transport layer,
15066 returns the ID of the first error detected during verification of the client
15067 certificate at depth > 0, or 0 if no error was encountered during this
15068 verification process. Please refer to your SSL library's documentation to
15069 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015070
Willy Tarreau74ca5042013-06-11 23:12:07 +020015071ssl_c_ca_err_depth : integer
15072 When the incoming connection was made over an SSL/TLS transport layer,
15073 returns the depth in the CA chain of the first error detected during the
15074 verification of the client certificate. If no error is encountered, 0 is
15075 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015076
Emeric Brun43e79582014-10-29 19:03:26 +010015077ssl_c_der : binary
15078 Returns the DER formatted certificate presented by the client when the
15079 incoming connection was made over an SSL/TLS transport layer. When used for
15080 an ACL, the value(s) to match against can be passed in hexadecimal form.
15081
Willy Tarreau74ca5042013-06-11 23:12:07 +020015082ssl_c_err : integer
15083 When the incoming connection was made over an SSL/TLS transport layer,
15084 returns the ID of the first error detected during verification at depth 0, or
15085 0 if no error was encountered during this verification process. Please refer
15086 to your SSL library's documentation to find the exhaustive list of error
15087 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015088
Willy Tarreau74ca5042013-06-11 23:12:07 +020015089ssl_c_i_dn([<entry>[,<occ>]]) : string
15090 When the incoming connection was made over an SSL/TLS transport layer,
15091 returns the full distinguished name of the issuer of the certificate
15092 presented by the client when no <entry> is specified, or the value of the
15093 first given entry found from the beginning of the DN. If a positive/negative
15094 occurrence number is specified as the optional second argument, it returns
15095 the value of the nth given entry value from the beginning/end of the DN.
15096 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15097 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020015098
Willy Tarreau74ca5042013-06-11 23:12:07 +020015099ssl_c_key_alg : string
15100 Returns the name of the algorithm used to generate the key of the certificate
15101 presented by the client when the incoming connection was made over an SSL/TLS
15102 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015103
Willy Tarreau74ca5042013-06-11 23:12:07 +020015104ssl_c_notafter : string
15105 Returns the end date presented by the client as a formatted string
15106 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15107 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015108
Willy Tarreau74ca5042013-06-11 23:12:07 +020015109ssl_c_notbefore : string
15110 Returns the start date presented by the client as a formatted string
15111 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15112 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015113
Willy Tarreau74ca5042013-06-11 23:12:07 +020015114ssl_c_s_dn([<entry>[,<occ>]]) : string
15115 When the incoming connection was made over an SSL/TLS transport layer,
15116 returns the full distinguished name of the subject of the certificate
15117 presented by the client when no <entry> is specified, or the value of the
15118 first given entry found from the beginning of the DN. If a positive/negative
15119 occurrence number is specified as the optional second argument, it returns
15120 the value of the nth given entry value from the beginning/end of the DN.
15121 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15122 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015123
Willy Tarreau74ca5042013-06-11 23:12:07 +020015124ssl_c_serial : binary
15125 Returns the serial of the certificate presented by the client when the
15126 incoming connection was made over an SSL/TLS transport layer. When used for
15127 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015128
Willy Tarreau74ca5042013-06-11 23:12:07 +020015129ssl_c_sha1 : binary
15130 Returns the SHA-1 fingerprint of the certificate presented by the client when
15131 the incoming connection was made over an SSL/TLS transport layer. This can be
15132 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015133 Note that the output is binary, so if you want to pass that signature to the
15134 server, you need to encode it in hex or base64, such as in the example below:
15135
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015136 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015137 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015138
Willy Tarreau74ca5042013-06-11 23:12:07 +020015139ssl_c_sig_alg : string
15140 Returns the name of the algorithm used to sign the certificate presented by
15141 the client when the incoming connection was made over an SSL/TLS transport
15142 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015143
Willy Tarreau74ca5042013-06-11 23:12:07 +020015144ssl_c_used : boolean
15145 Returns true if current SSL session uses a client certificate even if current
15146 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015147
Willy Tarreau74ca5042013-06-11 23:12:07 +020015148ssl_c_verify : integer
15149 Returns the verify result error ID when the incoming connection was made over
15150 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15151 refer to your SSL library's documentation for an exhaustive list of error
15152 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015153
Willy Tarreau74ca5042013-06-11 23:12:07 +020015154ssl_c_version : integer
15155 Returns the version of the certificate presented by the client when the
15156 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015157
Emeric Brun43e79582014-10-29 19:03:26 +010015158ssl_f_der : binary
15159 Returns the DER formatted certificate presented by the frontend when the
15160 incoming connection was made over an SSL/TLS transport layer. When used for
15161 an ACL, the value(s) to match against can be passed in hexadecimal form.
15162
Willy Tarreau74ca5042013-06-11 23:12:07 +020015163ssl_f_i_dn([<entry>[,<occ>]]) : string
15164 When the incoming connection was made over an SSL/TLS transport layer,
15165 returns the full distinguished name of the issuer of the certificate
15166 presented by the frontend when no <entry> is specified, or the value of the
15167 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015168 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015169 the value of the nth given entry value from the beginning/end of the DN.
15170 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15171 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020015172
Willy Tarreau74ca5042013-06-11 23:12:07 +020015173ssl_f_key_alg : string
15174 Returns the name of the algorithm used to generate the key of the certificate
15175 presented by the frontend when the incoming connection was made over an
15176 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015177
Willy Tarreau74ca5042013-06-11 23:12:07 +020015178ssl_f_notafter : string
15179 Returns the end date presented by the frontend as a formatted string
15180 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15181 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015182
Willy Tarreau74ca5042013-06-11 23:12:07 +020015183ssl_f_notbefore : string
15184 Returns the start date presented by the frontend as a formatted string
15185 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15186 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015187
Willy Tarreau74ca5042013-06-11 23:12:07 +020015188ssl_f_s_dn([<entry>[,<occ>]]) : string
15189 When the incoming connection was made over an SSL/TLS transport layer,
15190 returns the full distinguished name of the subject of the certificate
15191 presented by the frontend when no <entry> is specified, or the value of the
15192 first given entry found from the beginning of the DN. If a positive/negative
15193 occurrence number is specified as the optional second argument, it returns
15194 the value of the nth given entry value from the beginning/end of the DN.
15195 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15196 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020015197
Willy Tarreau74ca5042013-06-11 23:12:07 +020015198ssl_f_serial : binary
15199 Returns the serial of the certificate presented by the frontend when the
15200 incoming connection was made over an SSL/TLS transport layer. When used for
15201 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015202
Emeric Brun55f4fa82014-04-30 17:11:25 +020015203ssl_f_sha1 : binary
15204 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15205 when the incoming connection was made over an SSL/TLS transport layer. This
15206 can be used to know which certificate was chosen using SNI.
15207
Willy Tarreau74ca5042013-06-11 23:12:07 +020015208ssl_f_sig_alg : string
15209 Returns the name of the algorithm used to sign the certificate presented by
15210 the frontend when the incoming connection was made over an SSL/TLS transport
15211 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015212
Willy Tarreau74ca5042013-06-11 23:12:07 +020015213ssl_f_version : integer
15214 Returns the version of the certificate presented by the frontend when the
15215 incoming connection was made over an SSL/TLS transport layer.
15216
15217ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015218 Returns true when the front connection was made via an SSL/TLS transport
15219 layer and is locally deciphered. This means it has matched a socket declared
15220 with a "bind" line having the "ssl" option.
15221
Willy Tarreau74ca5042013-06-11 23:12:07 +020015222 Example :
15223 # This passes "X-Proto: https" to servers when client connects over SSL
15224 listen http-https
15225 bind :80
15226 bind :443 ssl crt /etc/haproxy.pem
15227 http-request add-header X-Proto https if { ssl_fc }
15228
15229ssl_fc_alg_keysize : integer
15230 Returns the symmetric cipher key size supported in bits when the incoming
15231 connection was made over an SSL/TLS transport layer.
15232
15233ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015234 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015235 incoming connection made via a TLS transport layer and locally deciphered by
15236 haproxy. The result is a string containing the protocol name advertised by
15237 the client. The SSL library must have been built with support for TLS
15238 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15239 not advertised unless the "alpn" keyword on the "bind" line specifies a
15240 protocol list. Also, nothing forces the client to pick a protocol from this
15241 list, any other one may be requested. The TLS ALPN extension is meant to
15242 replace the TLS NPN extension. See also "ssl_fc_npn".
15243
Willy Tarreau74ca5042013-06-11 23:12:07 +020015244ssl_fc_cipher : string
15245 Returns the name of the used cipher when the incoming connection was made
15246 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015247
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015248ssl_fc_cipherlist_bin : binary
15249 Returns the binary form of the client hello cipher list. The maximum returned
15250 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015251 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015252
15253ssl_fc_cipherlist_hex : string
15254 Returns the binary form of the client hello cipher list encoded as
15255 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015256 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015257
15258ssl_fc_cipherlist_str : string
15259 Returns the decoded text form of the client hello cipher list. The maximum
15260 number of ciphers returned is according with the value of
15261 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015262 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015263 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015264
15265ssl_fc_cipherlist_xxh : integer
15266 Returns a xxh64 of the cipher list. This hash can be return only is the value
15267 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015268 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015269
Patrick Hemmer65674662019-06-04 08:13:03 -040015270ssl_fc_client_random : binary
15271 Returns the client random of the front connection when the incoming connection
15272 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15273 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15274
Willy Tarreau74ca5042013-06-11 23:12:07 +020015275ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015276 Returns true if a client certificate is present in an incoming connection over
15277 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015278 Note: on SSL session resumption with Session ID or TLS ticket, client
15279 certificate is not present in the current connection but may be retrieved
15280 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15281 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015282
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015283ssl_fc_has_early : boolean
15284 Returns true if early data were sent, and the handshake didn't happen yet. As
15285 it has security implications, it is useful to be able to refuse those, or
15286 wait until the handshake happened.
15287
Willy Tarreau74ca5042013-06-11 23:12:07 +020015288ssl_fc_has_sni : boolean
15289 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015290 in an incoming connection was made over an SSL/TLS transport layer. Returns
15291 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050015292 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015293 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015294
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015295ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015296 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015297 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15298 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015299
Willy Tarreau74ca5042013-06-11 23:12:07 +020015300ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015301 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015302 made via a TLS transport layer and locally deciphered by haproxy. The result
15303 is a string containing the protocol name advertised by the client. The SSL
15304 library must have been built with support for TLS extensions enabled (check
15305 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15306 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15307 forces the client to pick a protocol from this list, any other one may be
15308 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015309
Willy Tarreau74ca5042013-06-11 23:12:07 +020015310ssl_fc_protocol : string
15311 Returns the name of the used protocol when the incoming connection was made
15312 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015313
Emeric Brunb73a9b02014-04-30 18:49:19 +020015314ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015315 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015316 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15317 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040015318
Patrick Hemmer65674662019-06-04 08:13:03 -040015319ssl_fc_server_random : binary
15320 Returns the server random of the front connection when the incoming connection
15321 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15322 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15323
Willy Tarreau74ca5042013-06-11 23:12:07 +020015324ssl_fc_session_id : binary
15325 Returns the SSL ID of the front connection when the incoming connection was
15326 made over an SSL/TLS transport layer. It is useful to stick a given client to
15327 a server. It is important to note that some browsers refresh their session ID
15328 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020015329
Patrick Hemmere0275472018-04-28 19:15:51 -040015330ssl_fc_session_key : binary
15331 Returns the SSL session master key of the front connection when the incoming
15332 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15333 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15334 BoringSSL.
15335
15336
Willy Tarreau74ca5042013-06-11 23:12:07 +020015337ssl_fc_sni : string
15338 This extracts the Server Name Indication TLS extension (SNI) field from an
15339 incoming connection made via an SSL/TLS transport layer and locally
15340 deciphered by haproxy. The result (when present) typically is a string
15341 matching the HTTPS host name (253 chars or less). The SSL library must have
15342 been built with support for TLS extensions enabled (check haproxy -vv).
15343
15344 This fetch is different from "req_ssl_sni" above in that it applies to the
15345 connection being deciphered by haproxy and not to SSL contents being blindly
15346 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050015347 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020015348 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020015349
Willy Tarreau74ca5042013-06-11 23:12:07 +020015350 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020015351 ssl_fc_sni_end : suffix match
15352 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020015353
Willy Tarreau74ca5042013-06-11 23:12:07 +020015354ssl_fc_use_keysize : integer
15355 Returns the symmetric cipher key size used in bits when the incoming
15356 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015357
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015358
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200153597.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015360------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015361
Willy Tarreau74ca5042013-06-11 23:12:07 +020015362Fetching samples from buffer contents is a bit different from the previous
15363sample fetches above because the sampled data are ephemeral. These data can
15364only be used when they're available and will be lost when they're forwarded.
15365For this reason, samples fetched from buffer contents during a request cannot
15366be used in a response for example. Even while the data are being fetched, they
15367can change. Sometimes it is necessary to set some delays or combine multiple
15368sample fetch methods to ensure that the expected data are complete and usable,
15369for example through TCP request content inspection. Please see the "tcp-request
15370content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015371
Willy Tarreau74ca5042013-06-11 23:12:07 +020015372payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015373 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015374 "stick on", "stick match"), and for "res.payload" when used in the context of
15375 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015376
Willy Tarreau74ca5042013-06-11 23:12:07 +020015377payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15378 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015379 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015380 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015381
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015382req.hdrs : string
15383 Returns the current request headers as string including the last empty line
15384 separating headers from the request body. The last empty line can be used to
15385 detect a truncated header block. This sample fetch is useful for some SPOE
15386 headers analyzers and for advanced logging.
15387
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015388req.hdrs_bin : binary
15389 Returns the current request headers contained in preparsed binary form. This
15390 is useful for offloading some processing with SPOE. Each string is described
15391 by a length followed by the number of bytes indicated in the length. The
15392 length is represented using the variable integer encoding detailed in the
15393 SPOE documentation. The end of the list is marked by a couple of empty header
15394 names and values (length of 0 for both).
15395
15396 *(<str:header-name><str:header-value>)<empty string><empty string>
15397
15398 int: refer to the SPOE documentation for the encoding
15399 str: <int:length><bytes>
15400
Willy Tarreau74ca5042013-06-11 23:12:07 +020015401req.len : integer
15402req_len : integer (deprecated)
15403 Returns an integer value corresponding to the number of bytes present in the
15404 request buffer. This is mostly used in ACL. It is important to understand
15405 that this test does not return false as long as the buffer is changing. This
15406 means that a check with equality to zero will almost always immediately match
15407 at the beginning of the session, while a test for more data will wait for
15408 that data to come in and return false only when haproxy is certain that no
15409 more data will come in. This test was designed to be used with TCP request
15410 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015411
Willy Tarreau74ca5042013-06-11 23:12:07 +020015412req.payload(<offset>,<length>) : binary
15413 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015414 in the request buffer. As a special case, if the <length> argument is zero,
15415 the the whole buffer from <offset> to the end is extracted. This can be used
15416 with ACLs in order to check for the presence of some content in a buffer at
15417 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015418
Willy Tarreau74ca5042013-06-11 23:12:07 +020015419 ACL alternatives :
15420 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015421
Willy Tarreau74ca5042013-06-11 23:12:07 +020015422req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15423 This extracts a binary block whose size is specified at <offset1> for <length>
15424 bytes, and which starts at <offset2> if specified or just after the length in
15425 the request buffer. The <offset2> parameter also supports relative offsets if
15426 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015427
Willy Tarreau74ca5042013-06-11 23:12:07 +020015428 ACL alternatives :
15429 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015430
Willy Tarreau74ca5042013-06-11 23:12:07 +020015431 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015432
Willy Tarreau74ca5042013-06-11 23:12:07 +020015433req.proto_http : boolean
15434req_proto_http : boolean (deprecated)
15435 Returns true when data in the request buffer look like HTTP and correctly
15436 parses as such. It is the same parser as the common HTTP request parser which
15437 is used so there should be no surprises. The test does not match until the
15438 request is complete, failed or timed out. This test may be used to report the
15439 protocol in TCP logs, but the biggest use is to block TCP request analysis
15440 until a complete HTTP request is present in the buffer, for example to track
15441 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015442
Willy Tarreau74ca5042013-06-11 23:12:07 +020015443 Example:
15444 # track request counts per "base" (concatenation of Host+URL)
15445 tcp-request inspect-delay 10s
15446 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015447 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015448
Willy Tarreau74ca5042013-06-11 23:12:07 +020015449req.rdp_cookie([<name>]) : string
15450rdp_cookie([<name>]) : string (deprecated)
15451 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15452 <name>, or any cookie if unspecified. The parser only checks for the first
15453 cookie, as illustrated in the RDP protocol specification. The cookie name is
15454 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15455 contain the user name of the client connecting to the server if properly
15456 configured on the client. The "MSTSHASH" cookie is often used as well for
15457 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015458
Willy Tarreau74ca5042013-06-11 23:12:07 +020015459 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15460 used and thus the distribution of clients to backend servers is not linked to
15461 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15462 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15463 distribution of clients to backend servers than the hash used by "balance
15464 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015465
Willy Tarreau74ca5042013-06-11 23:12:07 +020015466 ACL derivatives :
15467 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015468
Willy Tarreau74ca5042013-06-11 23:12:07 +020015469 Example :
15470 listen tse-farm
15471 bind 0.0.0.0:3389
15472 # wait up to 5s for an RDP cookie in the request
15473 tcp-request inspect-delay 5s
15474 tcp-request content accept if RDP_COOKIE
15475 # apply RDP cookie persistence
15476 persist rdp-cookie
15477 # Persist based on the mstshash cookie
15478 # This is only useful makes sense if
15479 # balance rdp-cookie is not used
15480 stick-table type string size 204800
15481 stick on req.rdp_cookie(mstshash)
15482 server srv1 1.1.1.1:3389
15483 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015484
Willy Tarreau74ca5042013-06-11 23:12:07 +020015485 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15486 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015487
Willy Tarreau74ca5042013-06-11 23:12:07 +020015488req.rdp_cookie_cnt([name]) : integer
15489rdp_cookie_cnt([name]) : integer (deprecated)
15490 Tries to parse the request buffer as RDP protocol, then returns an integer
15491 corresponding to the number of RDP cookies found. If an optional cookie name
15492 is passed, only cookies matching this name are considered. This is mostly
15493 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015494
Willy Tarreau74ca5042013-06-11 23:12:07 +020015495 ACL derivatives :
15496 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015497
Alex Zorin4afdd132018-12-30 13:56:28 +110015498req.ssl_alpn : string
15499 Returns a string containing the values of the Application-Layer Protocol
15500 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
15501 ClientHello message. Note that this only applies to raw contents found in the
15502 request buffer and not to the contents deciphered via an SSL data layer, so
15503 this will not work with "bind" lines having the "ssl" option. This is useful
15504 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020015505 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110015506
15507 Examples :
15508 # Wait for a client hello for at most 5 seconds
15509 tcp-request inspect-delay 5s
15510 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020015511 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110015512 default_backend bk_default
15513
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015514req.ssl_ec_ext : boolean
15515 Returns a boolean identifying if client sent the Supported Elliptic Curves
15516 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015517 message. This can be used to present ECC compatible clients with EC
15518 certificate and to use RSA for all others, on the same IP address. Note that
15519 this only applies to raw contents found in the request buffer and not to
15520 contents deciphered via an SSL data layer, so this will not work with "bind"
15521 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015522
Willy Tarreau74ca5042013-06-11 23:12:07 +020015523req.ssl_hello_type : integer
15524req_ssl_hello_type : integer (deprecated)
15525 Returns an integer value containing the type of the SSL hello message found
15526 in the request buffer if the buffer contains data that parse as a complete
15527 SSL (v3 or superior) client hello message. Note that this only applies to raw
15528 contents found in the request buffer and not to contents deciphered via an
15529 SSL data layer, so this will not work with "bind" lines having the "ssl"
15530 option. This is mostly used in ACL to detect presence of an SSL hello message
15531 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015532
Willy Tarreau74ca5042013-06-11 23:12:07 +020015533req.ssl_sni : string
15534req_ssl_sni : string (deprecated)
15535 Returns a string containing the value of the Server Name TLS extension sent
15536 by a client in a TLS stream passing through the request buffer if the buffer
15537 contains data that parse as a complete SSL (v3 or superior) client hello
15538 message. Note that this only applies to raw contents found in the request
15539 buffer and not to contents deciphered via an SSL data layer, so this will not
15540 work with "bind" lines having the "ssl" option. SNI normally contains the
15541 name of the host the client tries to connect to (for recent browsers). SNI is
15542 useful for allowing or denying access to certain hosts when SSL/TLS is used
15543 by the client. This test was designed to be used with TCP request content
15544 inspection. If content switching is needed, it is recommended to first wait
15545 for a complete client hello (type 1), like in the example below. See also
15546 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015547
Willy Tarreau74ca5042013-06-11 23:12:07 +020015548 ACL derivatives :
15549 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015550
Willy Tarreau74ca5042013-06-11 23:12:07 +020015551 Examples :
15552 # Wait for a client hello for at most 5 seconds
15553 tcp-request inspect-delay 5s
15554 tcp-request content accept if { req_ssl_hello_type 1 }
15555 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15556 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015557
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015558req.ssl_st_ext : integer
15559 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15560 Returns 1 if the client sent SessionTicket TLS Extension
15561 Returns 2 if the client also sent non-zero length TLS SessionTicket
15562 Note that this only applies to raw contents found in the request buffer and
15563 not to contents deciphered via an SSL data layer, so this will not work with
15564 "bind" lines having the "ssl" option. This can for example be used to detect
15565 whether the client sent a SessionTicket or not and stick it accordingly, if
15566 no SessionTicket then stick on SessionID or don't stick as there's no server
15567 side state is there when SessionTickets are in use.
15568
Willy Tarreau74ca5042013-06-11 23:12:07 +020015569req.ssl_ver : integer
15570req_ssl_ver : integer (deprecated)
15571 Returns an integer value containing the version of the SSL/TLS protocol of a
15572 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15573 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15574 composed of the major version multiplied by 65536, added to the minor
15575 version. Note that this only applies to raw contents found in the request
15576 buffer and not to contents deciphered via an SSL data layer, so this will not
15577 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015578 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015579 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015580
Willy Tarreau74ca5042013-06-11 23:12:07 +020015581 ACL derivatives :
15582 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015583
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015584res.len : integer
15585 Returns an integer value corresponding to the number of bytes present in the
15586 response buffer. This is mostly used in ACL. It is important to understand
15587 that this test does not return false as long as the buffer is changing. This
15588 means that a check with equality to zero will almost always immediately match
15589 at the beginning of the session, while a test for more data will wait for
15590 that data to come in and return false only when haproxy is certain that no
15591 more data will come in. This test was designed to be used with TCP response
15592 content inspection.
15593
Willy Tarreau74ca5042013-06-11 23:12:07 +020015594res.payload(<offset>,<length>) : binary
15595 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015596 in the response buffer. As a special case, if the <length> argument is zero,
15597 the the whole buffer from <offset> to the end is extracted. This can be used
15598 with ACLs in order to check for the presence of some content in a buffer at
15599 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015600
Willy Tarreau74ca5042013-06-11 23:12:07 +020015601res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15602 This extracts a binary block whose size is specified at <offset1> for <length>
15603 bytes, and which starts at <offset2> if specified or just after the length in
15604 the response buffer. The <offset2> parameter also supports relative offsets
15605 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015606
Willy Tarreau74ca5042013-06-11 23:12:07 +020015607 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015608
Willy Tarreau971f7b62015-09-29 14:06:59 +020015609res.ssl_hello_type : integer
15610rep_ssl_hello_type : integer (deprecated)
15611 Returns an integer value containing the type of the SSL hello message found
15612 in the response buffer if the buffer contains data that parses as a complete
15613 SSL (v3 or superior) hello message. Note that this only applies to raw
15614 contents found in the response buffer and not to contents deciphered via an
15615 SSL data layer, so this will not work with "server" lines having the "ssl"
15616 option. This is mostly used in ACL to detect presence of an SSL hello message
15617 that is supposed to contain an SSL session ID usable for stickiness.
15618
Willy Tarreau74ca5042013-06-11 23:12:07 +020015619wait_end : boolean
15620 This fetch either returns true when the inspection period is over, or does
15621 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015622 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015623 actions, such as a delayed reject for some special addresses. Since it either
15624 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015625 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015626 "WAIT_END" is always usable without prior declaration. This test was designed
15627 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015628
Willy Tarreau74ca5042013-06-11 23:12:07 +020015629 Examples :
15630 # delay every incoming request by 2 seconds
15631 tcp-request inspect-delay 2s
15632 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015633
Willy Tarreau74ca5042013-06-11 23:12:07 +020015634 # don't immediately tell bad guys they are rejected
15635 tcp-request inspect-delay 10s
15636 acl goodguys src 10.0.0.0/24
15637 acl badguys src 10.0.1.0/24
15638 tcp-request content accept if goodguys
15639 tcp-request content reject if badguys WAIT_END
15640 tcp-request content reject
15641
15642
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200156437.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015644--------------------------------------
15645
15646It is possible to fetch samples from HTTP contents, requests and responses.
15647This application layer is also called layer 7. It is only possible to fetch the
15648data in this section when a full HTTP request or response has been parsed from
15649its respective request or response buffer. This is always the case with all
15650HTTP specific rules and for sections running with "mode http". When using TCP
15651content inspection, it may be necessary to support an inspection delay in order
15652to let the request or response come in first. These fetches may require a bit
15653more CPU resources than the layer 4 ones, but not much since the request and
15654response are indexed.
15655
15656base : string
15657 This returns the concatenation of the first Host header and the path part of
15658 the request, which starts at the first slash and ends before the question
15659 mark. It can be useful in virtual hosted environments to detect URL abuses as
15660 well as to improve shared caches efficiency. Using this with a limited size
15661 stick table also allows one to collect statistics about most commonly
15662 requested objects by host/path. With ACLs it can allow simple content
15663 switching rules involving the host and the path at the same time, such as
15664 "www.example.com/favicon.ico". See also "path" and "uri".
15665
15666 ACL derivatives :
15667 base : exact string match
15668 base_beg : prefix match
15669 base_dir : subdir match
15670 base_dom : domain match
15671 base_end : suffix match
15672 base_len : length match
15673 base_reg : regex match
15674 base_sub : substring match
15675
15676base32 : integer
15677 This returns a 32-bit hash of the value returned by the "base" fetch method
15678 above. This is useful to track per-URL activity on high traffic sites without
15679 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015680 memory. The output type is an unsigned integer. The hash function used is
15681 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15682 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015683
15684base32+src : binary
15685 This returns the concatenation of the base32 fetch above and the src fetch
15686 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15687 depending on the source address family. This can be used to track per-IP,
15688 per-URL counters.
15689
William Lallemand65ad6e12014-01-31 15:08:02 +010015690capture.req.hdr(<idx>) : string
15691 This extracts the content of the header captured by the "capture request
15692 header", idx is the position of the capture keyword in the configuration.
15693 The first entry is an index of 0. See also: "capture request header".
15694
15695capture.req.method : string
15696 This extracts the METHOD of an HTTP request. It can be used in both request
15697 and response. Unlike "method", it can be used in both request and response
15698 because it's allocated.
15699
15700capture.req.uri : string
15701 This extracts the request's URI, which starts at the first slash and ends
15702 before the first space in the request (without the host part). Unlike "path"
15703 and "url", it can be used in both request and response because it's
15704 allocated.
15705
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015706capture.req.ver : string
15707 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15708 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15709 logs because it relies on a persistent flag.
15710
William Lallemand65ad6e12014-01-31 15:08:02 +010015711capture.res.hdr(<idx>) : string
15712 This extracts the content of the header captured by the "capture response
15713 header", idx is the position of the capture keyword in the configuration.
15714 The first entry is an index of 0.
15715 See also: "capture response header"
15716
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015717capture.res.ver : string
15718 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15719 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15720 persistent flag.
15721
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015722req.body : binary
15723 This returns the HTTP request's available body as a block of data. It
15724 requires that the request body has been buffered made available using
15725 "option http-buffer-request". In case of chunked-encoded body, currently only
15726 the first chunk is analyzed.
15727
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015728req.body_param([<name>) : string
15729 This fetch assumes that the body of the POST request is url-encoded. The user
15730 can check if the "content-type" contains the value
15731 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15732 parameter <name> in the body, which ends before '&'. The parameter name is
15733 case-sensitive. If no name is given, any parameter will match, and the first
15734 one will be returned. The result is a string corresponding to the value of the
15735 parameter <name> as presented in the request body (no URL decoding is
15736 performed). Note that the ACL version of this fetch iterates over multiple
15737 parameters and will iteratively report all parameters values if no name is
15738 given.
15739
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015740req.body_len : integer
15741 This returns the length of the HTTP request's available body in bytes. It may
15742 be lower than the advertised length if the body is larger than the buffer. It
15743 requires that the request body has been buffered made available using
15744 "option http-buffer-request".
15745
15746req.body_size : integer
15747 This returns the advertised length of the HTTP request's body in bytes. It
15748 will represent the advertised Content-Length header, or the size of the first
15749 chunk in case of chunked encoding. In order to parse the chunks, it requires
15750 that the request body has been buffered made available using
15751 "option http-buffer-request".
15752
Willy Tarreau74ca5042013-06-11 23:12:07 +020015753req.cook([<name>]) : string
15754cook([<name>]) : string (deprecated)
15755 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15756 header line from the request, and returns its value as string. If no name is
15757 specified, the first cookie value is returned. When used with ACLs, all
15758 matching cookies are evaluated. Spaces around the name and the value are
15759 ignored as requested by the Cookie header specification (RFC6265). The cookie
15760 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15761 well return an empty value if it is present. Use the "found" match to detect
15762 presence. Use the res.cook() variant for response cookies sent by the server.
15763
15764 ACL derivatives :
15765 cook([<name>]) : exact string match
15766 cook_beg([<name>]) : prefix match
15767 cook_dir([<name>]) : subdir match
15768 cook_dom([<name>]) : domain match
15769 cook_end([<name>]) : suffix match
15770 cook_len([<name>]) : length match
15771 cook_reg([<name>]) : regex match
15772 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015773
Willy Tarreau74ca5042013-06-11 23:12:07 +020015774req.cook_cnt([<name>]) : integer
15775cook_cnt([<name>]) : integer (deprecated)
15776 Returns an integer value representing the number of occurrences of the cookie
15777 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015778
Willy Tarreau74ca5042013-06-11 23:12:07 +020015779req.cook_val([<name>]) : integer
15780cook_val([<name>]) : integer (deprecated)
15781 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15782 header line from the request, and converts its value to an integer which is
15783 returned. If no name is specified, the first cookie value is returned. When
15784 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015785
Willy Tarreau74ca5042013-06-11 23:12:07 +020015786cookie([<name>]) : string (deprecated)
15787 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15788 header line from the request, or a "Set-Cookie" header from the response, and
15789 returns its value as a string. A typical use is to get multiple clients
15790 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015791 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015792 multi-peer synchronization and state keeping across restarts. If no name is
15793 specified, the first cookie value is returned. This fetch should not be used
15794 anymore and should be replaced by req.cook() or res.cook() instead as it
15795 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015796
Willy Tarreau74ca5042013-06-11 23:12:07 +020015797hdr([<name>[,<occ>]]) : string
15798 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15799 used on responses. Please refer to these respective fetches for more details.
15800 In case of doubt about the fetch direction, please use the explicit ones.
15801 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015802 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015803
Willy Tarreau74ca5042013-06-11 23:12:07 +020015804req.fhdr(<name>[,<occ>]) : string
15805 This extracts the last occurrence of header <name> in an HTTP request. When
15806 used from an ACL, all occurrences are iterated over until a match is found.
15807 Optionally, a specific occurrence might be specified as a position number.
15808 Positive values indicate a position from the first occurrence, with 1 being
15809 the first one. Negative values indicate positions relative to the last one,
15810 with -1 being the last one. It differs from req.hdr() in that any commas
15811 present in the value are returned and are not used as delimiters. This is
15812 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015813
Willy Tarreau74ca5042013-06-11 23:12:07 +020015814req.fhdr_cnt([<name>]) : integer
15815 Returns an integer value representing the number of occurrences of request
15816 header field name <name>, or the total number of header fields if <name> is
15817 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15818 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015819
Willy Tarreau74ca5042013-06-11 23:12:07 +020015820req.hdr([<name>[,<occ>]]) : string
15821 This extracts the last occurrence of header <name> in an HTTP request. When
15822 used from an ACL, all occurrences are iterated over until a match is found.
15823 Optionally, a specific occurrence might be specified as a position number.
15824 Positive values indicate a position from the first occurrence, with 1 being
15825 the first one. Negative values indicate positions relative to the last one,
15826 with -1 being the last one. A typical use is with the X-Forwarded-For header
15827 once converted to IP, associated with an IP stick-table. The function
15828 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015829 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015830 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015831 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015832
Willy Tarreau74ca5042013-06-11 23:12:07 +020015833 ACL derivatives :
15834 hdr([<name>[,<occ>]]) : exact string match
15835 hdr_beg([<name>[,<occ>]]) : prefix match
15836 hdr_dir([<name>[,<occ>]]) : subdir match
15837 hdr_dom([<name>[,<occ>]]) : domain match
15838 hdr_end([<name>[,<occ>]]) : suffix match
15839 hdr_len([<name>[,<occ>]]) : length match
15840 hdr_reg([<name>[,<occ>]]) : regex match
15841 hdr_sub([<name>[,<occ>]]) : substring match
15842
15843req.hdr_cnt([<name>]) : integer
15844hdr_cnt([<header>]) : integer (deprecated)
15845 Returns an integer value representing the number of occurrences of request
15846 header field name <name>, or the total number of header field values if
15847 <name> is not specified. It is important to remember that one header line may
15848 count as several headers if it has several values. The function considers any
15849 comma as a delimiter for distinct values. If full-line headers are desired
15850 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15851 detect presence, absence or abuse of a specific header, as well as to block
15852 request smuggling attacks by rejecting requests which contain more than one
15853 of certain headers. See "req.hdr" for more information on header matching.
15854
15855req.hdr_ip([<name>[,<occ>]]) : ip
15856hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15857 This extracts the last occurrence of header <name> in an HTTP request,
15858 converts it to an IPv4 or IPv6 address and returns this address. When used
15859 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15860 of every header is checked. Optionally, a specific occurrence might be
15861 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015862 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015863 positions relative to the last one, with -1 being the last one. A typical use
15864 is with the X-Forwarded-For and X-Client-IP headers.
15865
15866req.hdr_val([<name>[,<occ>]]) : integer
15867hdr_val([<name>[,<occ>]]) : integer (deprecated)
15868 This extracts the last occurrence of header <name> in an HTTP request, and
15869 converts it to an integer value. When used with ACLs, all occurrences are
15870 checked, and if <name> is omitted, every value of every header is checked.
15871 Optionally, a specific occurrence might be specified as a position number.
15872 Positive values indicate a position from the first occurrence, with 1 being
15873 the first one. Negative values indicate positions relative to the last one,
15874 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15875
Frédéric Lécailleec891192019-02-26 15:02:35 +010015876
15877
Willy Tarreau74ca5042013-06-11 23:12:07 +020015878http_auth(<userlist>) : boolean
15879 Returns a boolean indicating whether the authentication data received from
15880 the client match a username & password stored in the specified userlist. This
15881 fetch function is not really useful outside of ACLs. Currently only http
15882 basic auth is supported.
15883
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015884http_auth_group(<userlist>) : string
15885 Returns a string corresponding to the user name found in the authentication
15886 data received from the client if both the user name and password are valid
15887 according to the specified userlist. The main purpose is to use it in ACLs
15888 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015889 This fetch function is not really useful outside of ACLs. Currently only http
15890 basic auth is supported.
15891
15892 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015893 http_auth_group(<userlist>) : group ...
15894 Returns true when the user extracted from the request and whose password is
15895 valid according to the specified userlist belongs to at least one of the
15896 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015897
Christopher Fauleta4063562019-08-02 11:51:37 +020015898http_auth_pass : string
15899 Returns the user's password found in the authentication data received from
15900 the client, as supplied in the Authorization header. Not checks are
15901 performed by this sample fetch. Only Basic authentication is supported.
15902
15903http_auth_type : string
15904 Returns the authentication method found in the authentication data received from
15905 the client, as supplied in the Authorization header. Not checks are
15906 performed by this sample fetch. Only Basic authentication is supported.
15907
15908http_auth_user : string
15909 Returns the user name found in the authentication data received from the
15910 client, as supplied in the Authorization header. Not checks are performed by
15911 this sample fetch. Only Basic authentication is supported.
15912
Willy Tarreau74ca5042013-06-11 23:12:07 +020015913http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015914 Returns true when the request being processed is the first one of the
15915 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015916 from some requests when a request is not the first one, or to help grouping
15917 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015918
Willy Tarreau74ca5042013-06-11 23:12:07 +020015919method : integer + string
15920 Returns an integer value corresponding to the method in the HTTP request. For
15921 example, "GET" equals 1 (check sources to establish the matching). Value 9
15922 means "other method" and may be converted to a string extracted from the
15923 stream. This should not be used directly as a sample, this is only meant to
15924 be used from ACLs, which transparently convert methods from patterns to these
15925 integer + string values. Some predefined ACL already check for most common
15926 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015927
Willy Tarreau74ca5042013-06-11 23:12:07 +020015928 ACL derivatives :
15929 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015930
Willy Tarreau74ca5042013-06-11 23:12:07 +020015931 Example :
15932 # only accept GET and HEAD requests
15933 acl valid_method method GET HEAD
15934 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020015935
Willy Tarreau74ca5042013-06-11 23:12:07 +020015936path : string
15937 This extracts the request's URL path, which starts at the first slash and
15938 ends before the question mark (without the host part). A typical use is with
15939 prefetch-capable caches, and with portals which need to aggregate multiple
15940 information from databases and keep them in caches. Note that with outgoing
15941 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010015942 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020015943 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015944
Willy Tarreau74ca5042013-06-11 23:12:07 +020015945 ACL derivatives :
15946 path : exact string match
15947 path_beg : prefix match
15948 path_dir : subdir match
15949 path_dom : domain match
15950 path_end : suffix match
15951 path_len : length match
15952 path_reg : regex match
15953 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015954
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015955query : string
15956 This extracts the request's query string, which starts after the first
15957 question mark. If no question mark is present, this fetch returns nothing. If
15958 a question mark is present but nothing follows, it returns an empty string.
15959 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010015960 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015961 which stops before the question mark.
15962
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015963req.hdr_names([<delim>]) : string
15964 This builds a string made from the concatenation of all header names as they
15965 appear in the request when the rule is evaluated. The default delimiter is
15966 the comma (',') but it may be overridden as an optional argument <delim>. In
15967 this case, only the first character of <delim> is considered.
15968
Willy Tarreau74ca5042013-06-11 23:12:07 +020015969req.ver : string
15970req_ver : string (deprecated)
15971 Returns the version string from the HTTP request, for example "1.1". This can
15972 be useful for logs, but is mostly there for ACL. Some predefined ACL already
15973 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015974
Willy Tarreau74ca5042013-06-11 23:12:07 +020015975 ACL derivatives :
15976 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020015977
Willy Tarreau74ca5042013-06-11 23:12:07 +020015978res.comp : boolean
15979 Returns the boolean "true" value if the response has been compressed by
15980 HAProxy, otherwise returns boolean "false". This may be used to add
15981 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015982
Willy Tarreau74ca5042013-06-11 23:12:07 +020015983res.comp_algo : string
15984 Returns a string containing the name of the algorithm used if the response
15985 was compressed by HAProxy, for example : "deflate". This may be used to add
15986 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015987
Willy Tarreau74ca5042013-06-11 23:12:07 +020015988res.cook([<name>]) : string
15989scook([<name>]) : string (deprecated)
15990 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15991 header line from the response, and returns its value as string. If no name is
15992 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015993
Willy Tarreau74ca5042013-06-11 23:12:07 +020015994 ACL derivatives :
15995 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015996
Willy Tarreau74ca5042013-06-11 23:12:07 +020015997res.cook_cnt([<name>]) : integer
15998scook_cnt([<name>]) : integer (deprecated)
15999 Returns an integer value representing the number of occurrences of the cookie
16000 <name> in the response, or all cookies if <name> is not specified. This is
16001 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016002
Willy Tarreau74ca5042013-06-11 23:12:07 +020016003res.cook_val([<name>]) : integer
16004scook_val([<name>]) : integer (deprecated)
16005 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16006 header line from the response, and converts its value to an integer which is
16007 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016008
Willy Tarreau74ca5042013-06-11 23:12:07 +020016009res.fhdr([<name>[,<occ>]]) : string
16010 This extracts the last occurrence of header <name> in an HTTP response, or of
16011 the last header if no <name> is specified. Optionally, a specific occurrence
16012 might be specified as a position number. Positive values indicate a position
16013 from the first occurrence, with 1 being the first one. Negative values
16014 indicate positions relative to the last one, with -1 being the last one. It
16015 differs from res.hdr() in that any commas present in the value are returned
16016 and are not used as delimiters. If this is not desired, the res.hdr() fetch
16017 should be used instead. This is sometimes useful with headers such as Date or
16018 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016019
Willy Tarreau74ca5042013-06-11 23:12:07 +020016020res.fhdr_cnt([<name>]) : integer
16021 Returns an integer value representing the number of occurrences of response
16022 header field name <name>, or the total number of header fields if <name> is
16023 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
16024 the number of full line headers and does not stop on commas. If this is not
16025 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016026
Willy Tarreau74ca5042013-06-11 23:12:07 +020016027res.hdr([<name>[,<occ>]]) : string
16028shdr([<name>[,<occ>]]) : string (deprecated)
16029 This extracts the last occurrence of header <name> in an HTTP response, or of
16030 the last header if no <name> is specified. Optionally, a specific occurrence
16031 might be specified as a position number. Positive values indicate a position
16032 from the first occurrence, with 1 being the first one. Negative values
16033 indicate positions relative to the last one, with -1 being the last one. This
16034 can be useful to learn some data into a stick-table. The function considers
16035 any comma as a delimiter for distinct values. If this is not desired, the
16036 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016037
Willy Tarreau74ca5042013-06-11 23:12:07 +020016038 ACL derivatives :
16039 shdr([<name>[,<occ>]]) : exact string match
16040 shdr_beg([<name>[,<occ>]]) : prefix match
16041 shdr_dir([<name>[,<occ>]]) : subdir match
16042 shdr_dom([<name>[,<occ>]]) : domain match
16043 shdr_end([<name>[,<occ>]]) : suffix match
16044 shdr_len([<name>[,<occ>]]) : length match
16045 shdr_reg([<name>[,<occ>]]) : regex match
16046 shdr_sub([<name>[,<occ>]]) : substring match
16047
16048res.hdr_cnt([<name>]) : integer
16049shdr_cnt([<name>]) : integer (deprecated)
16050 Returns an integer value representing the number of occurrences of response
16051 header field name <name>, or the total number of header fields if <name> is
16052 not specified. The function considers any comma as a delimiter for distinct
16053 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16054 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016055
Willy Tarreau74ca5042013-06-11 23:12:07 +020016056res.hdr_ip([<name>[,<occ>]]) : ip
16057shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16058 This extracts the last occurrence of header <name> in an HTTP response,
16059 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16060 specific occurrence might be specified as a position number. Positive values
16061 indicate a position from the first occurrence, with 1 being the first one.
16062 Negative values indicate positions relative to the last one, with -1 being
16063 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016064
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016065res.hdr_names([<delim>]) : string
16066 This builds a string made from the concatenation of all header names as they
16067 appear in the response when the rule is evaluated. The default delimiter is
16068 the comma (',') but it may be overridden as an optional argument <delim>. In
16069 this case, only the first character of <delim> is considered.
16070
Willy Tarreau74ca5042013-06-11 23:12:07 +020016071res.hdr_val([<name>[,<occ>]]) : integer
16072shdr_val([<name>[,<occ>]]) : integer (deprecated)
16073 This extracts the last occurrence of header <name> in an HTTP response, and
16074 converts it to an integer value. Optionally, a specific occurrence might be
16075 specified as a position number. Positive values indicate a position from the
16076 first occurrence, with 1 being the first one. Negative values indicate
16077 positions relative to the last one, with -1 being the last one. This can be
16078 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016079
Willy Tarreau74ca5042013-06-11 23:12:07 +020016080res.ver : string
16081resp_ver : string (deprecated)
16082 Returns the version string from the HTTP response, for example "1.1". This
16083 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016084
Willy Tarreau74ca5042013-06-11 23:12:07 +020016085 ACL derivatives :
16086 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016087
Willy Tarreau74ca5042013-06-11 23:12:07 +020016088set-cookie([<name>]) : string (deprecated)
16089 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16090 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016091 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016092 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016093
Willy Tarreau74ca5042013-06-11 23:12:07 +020016094 This fetch function is deprecated and has been superseded by the "res.cook"
16095 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016096
Willy Tarreau74ca5042013-06-11 23:12:07 +020016097status : integer
16098 Returns an integer containing the HTTP status code in the HTTP response, for
16099 example, 302. It is mostly used within ACLs and integer ranges, for example,
16100 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016101
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016102unique-id : string
16103 Returns the unique-id attached to the request. The directive
16104 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16105 fails. Note that the unique-id is usually used with HTTP requests, however this
16106 sample fetch can be used with other protocols. Obviously, if it is used with
16107 other protocols than HTTP, the unique-id-format directive must not contain
16108 HTTP parts. See: unique-id-format and unique-id-header
16109
Willy Tarreau74ca5042013-06-11 23:12:07 +020016110url : string
16111 This extracts the request's URL as presented in the request. A typical use is
16112 with prefetch-capable caches, and with portals which need to aggregate
16113 multiple information from databases and keep them in caches. With ACLs, using
16114 "path" is preferred over using "url", because clients may send a full URL as
16115 is normally done with proxies. The only real use is to match "*" which does
16116 not match in "path", and for which there is already a predefined ACL. See
16117 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016118
Willy Tarreau74ca5042013-06-11 23:12:07 +020016119 ACL derivatives :
16120 url : exact string match
16121 url_beg : prefix match
16122 url_dir : subdir match
16123 url_dom : domain match
16124 url_end : suffix match
16125 url_len : length match
16126 url_reg : regex match
16127 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016128
Willy Tarreau74ca5042013-06-11 23:12:07 +020016129url_ip : ip
16130 This extracts the IP address from the request's URL when the host part is
16131 presented as an IP address. Its use is very limited. For instance, a
16132 monitoring system might use this field as an alternative for the source IP in
16133 order to test what path a given source address would follow, or to force an
16134 entry in a table for a given source address. With ACLs it can be used to
16135 restrict access to certain systems through a proxy, for example when combined
16136 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016137
Willy Tarreau74ca5042013-06-11 23:12:07 +020016138url_port : integer
16139 This extracts the port part from the request's URL. Note that if the port is
16140 not specified in the request, port 80 is assumed. With ACLs it can be used to
16141 restrict access to certain systems through a proxy, for example when combined
16142 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016143
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016144urlp([<name>[,<delim>]]) : string
16145url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016146 This extracts the first occurrence of the parameter <name> in the query
16147 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016148 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16149 any parameter will match, and the first one will be returned. The result is
16150 a string corresponding to the value of the parameter <name> as presented in
16151 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016152 stickiness based on a client ID, to extract an application cookie passed as a
16153 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016154 this fetch iterates over multiple parameters and will iteratively report all
16155 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016156
Willy Tarreau74ca5042013-06-11 23:12:07 +020016157 ACL derivatives :
16158 urlp(<name>[,<delim>]) : exact string match
16159 urlp_beg(<name>[,<delim>]) : prefix match
16160 urlp_dir(<name>[,<delim>]) : subdir match
16161 urlp_dom(<name>[,<delim>]) : domain match
16162 urlp_end(<name>[,<delim>]) : suffix match
16163 urlp_len(<name>[,<delim>]) : length match
16164 urlp_reg(<name>[,<delim>]) : regex match
16165 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016166
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016167
Willy Tarreau74ca5042013-06-11 23:12:07 +020016168 Example :
16169 # match http://example.com/foo?PHPSESSIONID=some_id
16170 stick on urlp(PHPSESSIONID)
16171 # match http://example.com/foo;JSESSIONID=some_id
16172 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016173
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016174urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016175 See "urlp" above. This one extracts the URL parameter <name> in the request
16176 and converts it to an integer value. This can be used for session stickiness
16177 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016178
Dragan Dosen0070cd52016-06-16 12:19:49 +020016179url32 : integer
16180 This returns a 32-bit hash of the value obtained by concatenating the first
16181 Host header and the whole URL including parameters (not only the path part of
16182 the request, as in the "base32" fetch above). This is useful to track per-URL
16183 activity. A shorter hash is stored, saving a lot of memory. The output type
16184 is an unsigned integer.
16185
16186url32+src : binary
16187 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16188 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16189 the source address family. This can be used to track per-IP, per-URL counters.
16190
Willy Tarreau198a7442008-01-17 12:05:32 +010016191
Willy Tarreau74ca5042013-06-11 23:12:07 +0200161927.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016193---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016194
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016195Some predefined ACLs are hard-coded so that they do not have to be declared in
16196every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016197order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016198
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016199ACL name Equivalent to Usage
16200---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016201FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016202HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016203HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16204HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016205HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16206HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16207HTTP_URL_SLASH url_beg / match URL beginning with "/"
16208HTTP_URL_STAR url * match URL equal to "*"
16209LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016210METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016211METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016212METH_GET method GET HEAD match HTTP GET or HEAD method
16213METH_HEAD method HEAD match HTTP HEAD method
16214METH_OPTIONS method OPTIONS match HTTP OPTIONS method
16215METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016216METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016217METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020016218RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016219REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016220TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016221WAIT_END wait_end wait for end of content analysis
16222---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016223
Willy Tarreaub937b7e2010-01-12 15:27:54 +010016224
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162258. Logging
16226----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010016227
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016228One of HAProxy's strong points certainly lies is its precise logs. It probably
16229provides the finest level of information available for such a product, which is
16230very important for troubleshooting complex environments. Standard information
16231provided in logs include client ports, TCP/HTTP state timers, precise session
16232state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016233to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016234headers.
16235
16236In order to improve administrators reactivity, it offers a great transparency
16237about encountered problems, both internal and external, and it is possible to
16238send logs to different sources at the same time with different level filters :
16239
16240 - global process-level logs (system errors, start/stop, etc..)
16241 - per-instance system and internal errors (lack of resource, bugs, ...)
16242 - per-instance external troubles (servers up/down, max connections)
16243 - per-instance activity (client connections), either at the establishment or
16244 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016245 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016246 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016247
16248The ability to distribute different levels of logs to different log servers
16249allow several production teams to interact and to fix their problems as soon
16250as possible. For example, the system team might monitor system-wide errors,
16251while the application team might be monitoring the up/down for their servers in
16252real time, and the security team might analyze the activity logs with one hour
16253delay.
16254
16255
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162568.1. Log levels
16257---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016258
Simon Hormandf791f52011-05-29 15:01:10 +090016259TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016260source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016261HTTP request, HTTP return code, number of bytes transmitted, conditions
16262in which the session ended, and even exchanged cookies values. For example
16263track a particular user's problems. All messages may be sent to up to two
16264syslog servers. Check the "log" keyword in section 4.2 for more information
16265about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016266
16267
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162688.2. Log formats
16269----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016270
William Lallemand48940402012-01-30 16:47:22 +010016271HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016272and will be detailed in the following sections. A few of them may vary
16273slightly with the configuration, due to indicators specific to certain
16274options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016275
16276 - the default format, which is very basic and very rarely used. It only
16277 provides very basic information about the incoming connection at the moment
16278 it is accepted : source IP:port, destination IP:port, and frontend-name.
16279 This mode will eventually disappear so it will not be described to great
16280 extents.
16281
16282 - the TCP format, which is more advanced. This format is enabled when "option
16283 tcplog" is set on the frontend. HAProxy will then usually wait for the
16284 connection to terminate before logging. This format provides much richer
16285 information, such as timers, connection counts, queue size, etc... This
16286 format is recommended for pure TCP proxies.
16287
16288 - the HTTP format, which is the most advanced for HTTP proxying. This format
16289 is enabled when "option httplog" is set on the frontend. It provides the
16290 same information as the TCP format with some HTTP-specific fields such as
16291 the request, the status code, and captures of headers and cookies. This
16292 format is recommended for HTTP proxies.
16293
Emeric Brun3a058f32009-06-30 18:26:00 +020016294 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16295 fields arranged in the same order as the CLF format. In this mode, all
16296 timers, captures, flags, etc... appear one per field after the end of the
16297 common fields, in the same order they appear in the standard HTTP format.
16298
William Lallemand48940402012-01-30 16:47:22 +010016299 - the custom log format, allows you to make your own log line.
16300
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016301Next sections will go deeper into details for each of these formats. Format
16302specification will be performed on a "field" basis. Unless stated otherwise, a
16303field is a portion of text delimited by any number of spaces. Since syslog
16304servers are susceptible of inserting fields at the beginning of a line, it is
16305always assumed that the first field is the one containing the process name and
16306identifier.
16307
16308Note : Since log lines may be quite long, the log examples in sections below
16309 might be broken into multiple lines. The example log lines will be
16310 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16311 broken into multiple lines, each non-final line will end with a
16312 backslash ('\') and the next line will start indented by two characters.
16313
16314
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163158.2.1. Default log format
16316-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016317
16318This format is used when no specific option is set. The log is emitted as soon
16319as the connection is accepted. One should note that this currently is the only
16320format which logs the request's destination IP and ports.
16321
16322 Example :
16323 listen www
16324 mode http
16325 log global
16326 server srv1 127.0.0.1:8000
16327
16328 >>> Feb 6 12:12:09 localhost \
16329 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16330 (www/HTTP)
16331
16332 Field Format Extract from the example above
16333 1 process_name '[' pid ']:' haproxy[14385]:
16334 2 'Connect from' Connect from
16335 3 source_ip ':' source_port 10.0.1.2:33312
16336 4 'to' to
16337 5 destination_ip ':' destination_port 10.0.3.31:8012
16338 6 '(' frontend_name '/' mode ')' (www/HTTP)
16339
16340Detailed fields description :
16341 - "source_ip" is the IP address of the client which initiated the connection.
16342 - "source_port" is the TCP port of the client which initiated the connection.
16343 - "destination_ip" is the IP address the client connected to.
16344 - "destination_port" is the TCP port the client connected to.
16345 - "frontend_name" is the name of the frontend (or listener) which received
16346 and processed the connection.
16347 - "mode is the mode the frontend is operating (TCP or HTTP).
16348
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016349In case of a UNIX socket, the source and destination addresses are marked as
16350"unix:" and the ports reflect the internal ID of the socket which accepted the
16351connection (the same ID as reported in the stats).
16352
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016353It is advised not to use this deprecated format for newer installations as it
16354will eventually disappear.
16355
16356
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163578.2.2. TCP log format
16358---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016359
16360The TCP format is used when "option tcplog" is specified in the frontend, and
16361is the recommended format for pure TCP proxies. It provides a lot of precious
16362information for troubleshooting. Since this format includes timers and byte
16363counts, the log is normally emitted at the end of the session. It can be
16364emitted earlier if "option logasap" is specified, which makes sense in most
16365environments with long sessions such as remote terminals. Sessions which match
16366the "monitor" rules are never logged. It is also possible not to emit logs for
16367sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016368specifying "option dontlognull" in the frontend. Successful connections will
16369not be logged if "option dontlog-normal" is specified in the frontend. A few
16370fields may slightly vary depending on some configuration options, those are
16371marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016372
16373 Example :
16374 frontend fnt
16375 mode tcp
16376 option tcplog
16377 log global
16378 default_backend bck
16379
16380 backend bck
16381 server srv1 127.0.0.1:8000
16382
16383 >>> Feb 6 12:12:56 localhost \
16384 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
16385 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
16386
16387 Field Format Extract from the example above
16388 1 process_name '[' pid ']:' haproxy[14387]:
16389 2 client_ip ':' client_port 10.0.1.2:33313
16390 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
16391 4 frontend_name fnt
16392 5 backend_name '/' server_name bck/srv1
16393 6 Tw '/' Tc '/' Tt* 0/0/5007
16394 7 bytes_read* 212
16395 8 termination_state --
16396 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16397 10 srv_queue '/' backend_queue 0/0
16398
16399Detailed fields description :
16400 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016401 connection to haproxy. If the connection was accepted on a UNIX socket
16402 instead, the IP address would be replaced with the word "unix". Note that
16403 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016404 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016405 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016406 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016407
16408 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016409 If the connection was accepted on a UNIX socket instead, the port would be
16410 replaced with the ID of the accepting socket, which is also reported in the
16411 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016412
16413 - "accept_date" is the exact date when the connection was received by haproxy
16414 (which might be very slightly different from the date observed on the
16415 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016416 the same date which may appear in any upstream firewall's log. When used in
16417 HTTP mode, the accept_date field will be reset to the first moment the
16418 connection is ready to receive a new request (end of previous response for
16419 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016420
16421 - "frontend_name" is the name of the frontend (or listener) which received
16422 and processed the connection.
16423
16424 - "backend_name" is the name of the backend (or listener) which was selected
16425 to manage the connection to the server. This will be the same as the
16426 frontend if no switching rule has been applied, which is common for TCP
16427 applications.
16428
16429 - "server_name" is the name of the last server to which the connection was
16430 sent, which might differ from the first one if there were connection errors
16431 and a redispatch occurred. Note that this server belongs to the backend
16432 which processed the request. If the connection was aborted before reaching
16433 a server, "<NOSRV>" is indicated instead of a server name.
16434
16435 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16436 It can be "-1" if the connection was aborted before reaching the queue.
16437 See "Timers" below for more details.
16438
16439 - "Tc" is the total time in milliseconds spent waiting for the connection to
16440 establish to the final server, including retries. It can be "-1" if the
16441 connection was aborted before a connection could be established. See
16442 "Timers" below for more details.
16443
16444 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016445 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016446 "option logasap" was specified, then the time counting stops at the moment
16447 the log is emitted. In this case, a '+' sign is prepended before the value,
16448 indicating that the final one will be larger. See "Timers" below for more
16449 details.
16450
16451 - "bytes_read" is the total number of bytes transmitted from the server to
16452 the client when the log is emitted. If "option logasap" is specified, the
16453 this value will be prefixed with a '+' sign indicating that the final one
16454 may be larger. Please note that this value is a 64-bit counter, so log
16455 analysis tools must be able to handle it without overflowing.
16456
16457 - "termination_state" is the condition the session was in when the session
16458 ended. This indicates the session state, which side caused the end of
16459 session to happen, and for what reason (timeout, error, ...). The normal
16460 flags should be "--", indicating the session was closed by either end with
16461 no data remaining in buffers. See below "Session state at disconnection"
16462 for more details.
16463
16464 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016465 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016466 limits have been reached. For instance, if actconn is close to 512 when
16467 multiple connection errors occur, chances are high that the system limits
16468 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016469 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016470
16471 - "feconn" is the total number of concurrent connections on the frontend when
16472 the session was logged. It is useful to estimate the amount of resource
16473 required to sustain high loads, and to detect when the frontend's "maxconn"
16474 has been reached. Most often when this value increases by huge jumps, it is
16475 because there is congestion on the backend servers, but sometimes it can be
16476 caused by a denial of service attack.
16477
16478 - "beconn" is the total number of concurrent connections handled by the
16479 backend when the session was logged. It includes the total number of
16480 concurrent connections active on servers as well as the number of
16481 connections pending in queues. It is useful to estimate the amount of
16482 additional servers needed to support high loads for a given application.
16483 Most often when this value increases by huge jumps, it is because there is
16484 congestion on the backend servers, but sometimes it can be caused by a
16485 denial of service attack.
16486
16487 - "srv_conn" is the total number of concurrent connections still active on
16488 the server when the session was logged. It can never exceed the server's
16489 configured "maxconn" parameter. If this value is very often close or equal
16490 to the server's "maxconn", it means that traffic regulation is involved a
16491 lot, meaning that either the server's maxconn value is too low, or that
16492 there aren't enough servers to process the load with an optimal response
16493 time. When only one of the server's "srv_conn" is high, it usually means
16494 that this server has some trouble causing the connections to take longer to
16495 be processed than on other servers.
16496
16497 - "retries" is the number of connection retries experienced by this session
16498 when trying to connect to the server. It must normally be zero, unless a
16499 server is being stopped at the same moment the connection was attempted.
16500 Frequent retries generally indicate either a network problem between
16501 haproxy and the server, or a misconfigured system backlog on the server
16502 preventing new connections from being queued. This field may optionally be
16503 prefixed with a '+' sign, indicating that the session has experienced a
16504 redispatch after the maximal retry count has been reached on the initial
16505 server. In this case, the server name appearing in the log is the one the
16506 connection was redispatched to, and not the first one, though both may
16507 sometimes be the same in case of hashing for instance. So as a general rule
16508 of thumb, when a '+' is present in front of the retry count, this count
16509 should not be attributed to the logged server.
16510
16511 - "srv_queue" is the total number of requests which were processed before
16512 this one in the server queue. It is zero when the request has not gone
16513 through the server queue. It makes it possible to estimate the approximate
16514 server's response time by dividing the time spent in queue by the number of
16515 requests in the queue. It is worth noting that if a session experiences a
16516 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016517 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016518 backend queue unless a redispatch occurs.
16519
16520 - "backend_queue" is the total number of requests which were processed before
16521 this one in the backend's global queue. It is zero when the request has not
16522 gone through the global queue. It makes it possible to estimate the average
16523 queue length, which easily translates into a number of missing servers when
16524 divided by a server's "maxconn" parameter. It is worth noting that if a
16525 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016526 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016527 through both the server queue and the backend queue unless a redispatch
16528 occurs.
16529
16530
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165318.2.3. HTTP log format
16532----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016533
16534The HTTP format is the most complete and the best suited for HTTP proxies. It
16535is enabled by when "option httplog" is specified in the frontend. It provides
16536the same level of information as the TCP format with additional features which
16537are specific to the HTTP protocol. Just like the TCP format, the log is usually
16538emitted at the end of the session, unless "option logasap" is specified, which
16539generally only makes sense for download sites. A session which matches the
16540"monitor" rules will never logged. It is also possible not to log sessions for
16541which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016542frontend. Successful connections will not be logged if "option dontlog-normal"
16543is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016544
16545Most fields are shared with the TCP log, some being different. A few fields may
16546slightly vary depending on some configuration options. Those ones are marked
16547with a star ('*') after the field name below.
16548
16549 Example :
16550 frontend http-in
16551 mode http
16552 option httplog
16553 log global
16554 default_backend bck
16555
16556 backend static
16557 server srv1 127.0.0.1:8000
16558
16559 >>> Feb 6 12:14:14 localhost \
16560 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16561 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 +010016562 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016563
16564 Field Format Extract from the example above
16565 1 process_name '[' pid ']:' haproxy[14389]:
16566 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016567 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016568 4 frontend_name http-in
16569 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016570 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016571 7 status_code 200
16572 8 bytes_read* 2750
16573 9 captured_request_cookie -
16574 10 captured_response_cookie -
16575 11 termination_state ----
16576 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16577 13 srv_queue '/' backend_queue 0/0
16578 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16579 15 '{' captured_response_headers* '}' {}
16580 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016581
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016582Detailed fields description :
16583 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016584 connection to haproxy. If the connection was accepted on a UNIX socket
16585 instead, the IP address would be replaced with the word "unix". Note that
16586 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016587 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016588 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016589 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016590
16591 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016592 If the connection was accepted on a UNIX socket instead, the port would be
16593 replaced with the ID of the accepting socket, which is also reported in the
16594 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016595
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016596 - "request_date" is the exact date when the first byte of the HTTP request
16597 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016598
16599 - "frontend_name" is the name of the frontend (or listener) which received
16600 and processed the connection.
16601
16602 - "backend_name" is the name of the backend (or listener) which was selected
16603 to manage the connection to the server. This will be the same as the
16604 frontend if no switching rule has been applied.
16605
16606 - "server_name" is the name of the last server to which the connection was
16607 sent, which might differ from the first one if there were connection errors
16608 and a redispatch occurred. Note that this server belongs to the backend
16609 which processed the request. If the request was aborted before reaching a
16610 server, "<NOSRV>" is indicated instead of a server name. If the request was
16611 intercepted by the stats subsystem, "<STATS>" is indicated instead.
16612
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016613 - "TR" is the total time in milliseconds spent waiting for a full HTTP
16614 request from the client (not counting body) after the first byte was
16615 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050016616 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016617 always be very small because a request generally fits in one single packet.
16618 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020016619 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
16620 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016621
16622 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16623 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020016624 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016625
16626 - "Tc" is the total time in milliseconds spent waiting for the connection to
16627 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020016628 request was aborted before a connection could be established. See section
16629 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016630
16631 - "Tr" is the total time in milliseconds spent waiting for the server to send
16632 a full HTTP response, not counting data. It can be "-1" if the request was
16633 aborted before a complete response could be received. It generally matches
16634 the server's processing time for the request, though it may be altered by
16635 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020016636 "GET" requests generally indicate an overloaded server. See section 8.4
16637 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016638
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016639 - "Ta" is the time the request remained active in haproxy, which is the total
16640 time in milliseconds elapsed between the first byte of the request was
16641 received and the last byte of response was sent. It covers all possible
16642 processing except the handshake (see Th) and idle time (see Ti). There is
16643 one exception, if "option logasap" was specified, then the time counting
16644 stops at the moment the log is emitted. In this case, a '+' sign is
16645 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020016646 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016647
16648 - "status_code" is the HTTP status code returned to the client. This status
16649 is generally set by the server, but it might also be set by haproxy when
16650 the server cannot be reached or when its response is blocked by haproxy.
16651
16652 - "bytes_read" is the total number of bytes transmitted to the client when
16653 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050016654 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016655 the final one may be larger. Please note that this value is a 64-bit
16656 counter, so log analysis tools must be able to handle it without
16657 overflowing.
16658
16659 - "captured_request_cookie" is an optional "name=value" entry indicating that
16660 the client had this cookie in the request. The cookie name and its maximum
16661 length are defined by the "capture cookie" statement in the frontend
16662 configuration. The field is a single dash ('-') when the option is not
16663 set. Only one cookie may be captured, it is generally used to track session
16664 ID exchanges between a client and a server to detect session crossing
16665 between clients due to application bugs. For more details, please consult
16666 the section "Capturing HTTP headers and cookies" below.
16667
16668 - "captured_response_cookie" is an optional "name=value" entry indicating
16669 that the server has returned a cookie with its response. The cookie name
16670 and its maximum length are defined by the "capture cookie" statement in the
16671 frontend configuration. The field is a single dash ('-') when the option is
16672 not set. Only one cookie may be captured, it is generally used to track
16673 session ID exchanges between a client and a server to detect session
16674 crossing between clients due to application bugs. For more details, please
16675 consult the section "Capturing HTTP headers and cookies" below.
16676
16677 - "termination_state" is the condition the session was in when the session
16678 ended. This indicates the session state, which side caused the end of
16679 session to happen, for what reason (timeout, error, ...), just like in TCP
16680 logs, and information about persistence operations on cookies in the last
16681 two characters. The normal flags should begin with "--", indicating the
16682 session was closed by either end with no data remaining in buffers. See
16683 below "Session state at disconnection" for more details.
16684
16685 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016686 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016687 limits have been reached. For instance, if actconn is close to 512 or 1024
16688 when multiple connection errors occur, chances are high that the system
16689 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016690 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016691 system.
16692
16693 - "feconn" is the total number of concurrent connections on the frontend when
16694 the session was logged. It is useful to estimate the amount of resource
16695 required to sustain high loads, and to detect when the frontend's "maxconn"
16696 has been reached. Most often when this value increases by huge jumps, it is
16697 because there is congestion on the backend servers, but sometimes it can be
16698 caused by a denial of service attack.
16699
16700 - "beconn" is the total number of concurrent connections handled by the
16701 backend when the session was logged. It includes the total number of
16702 concurrent connections active on servers as well as the number of
16703 connections pending in queues. It is useful to estimate the amount of
16704 additional servers needed to support high loads for a given application.
16705 Most often when this value increases by huge jumps, it is because there is
16706 congestion on the backend servers, but sometimes it can be caused by a
16707 denial of service attack.
16708
16709 - "srv_conn" is the total number of concurrent connections still active on
16710 the server when the session was logged. It can never exceed the server's
16711 configured "maxconn" parameter. If this value is very often close or equal
16712 to the server's "maxconn", it means that traffic regulation is involved a
16713 lot, meaning that either the server's maxconn value is too low, or that
16714 there aren't enough servers to process the load with an optimal response
16715 time. When only one of the server's "srv_conn" is high, it usually means
16716 that this server has some trouble causing the requests to take longer to be
16717 processed than on other servers.
16718
16719 - "retries" is the number of connection retries experienced by this session
16720 when trying to connect to the server. It must normally be zero, unless a
16721 server is being stopped at the same moment the connection was attempted.
16722 Frequent retries generally indicate either a network problem between
16723 haproxy and the server, or a misconfigured system backlog on the server
16724 preventing new connections from being queued. This field may optionally be
16725 prefixed with a '+' sign, indicating that the session has experienced a
16726 redispatch after the maximal retry count has been reached on the initial
16727 server. In this case, the server name appearing in the log is the one the
16728 connection was redispatched to, and not the first one, though both may
16729 sometimes be the same in case of hashing for instance. So as a general rule
16730 of thumb, when a '+' is present in front of the retry count, this count
16731 should not be attributed to the logged server.
16732
16733 - "srv_queue" is the total number of requests which were processed before
16734 this one in the server queue. It is zero when the request has not gone
16735 through the server queue. It makes it possible to estimate the approximate
16736 server's response time by dividing the time spent in queue by the number of
16737 requests in the queue. It is worth noting that if a session experiences a
16738 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016739 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016740 backend queue unless a redispatch occurs.
16741
16742 - "backend_queue" is the total number of requests which were processed before
16743 this one in the backend's global queue. It is zero when the request has not
16744 gone through the global queue. It makes it possible to estimate the average
16745 queue length, which easily translates into a number of missing servers when
16746 divided by a server's "maxconn" parameter. It is worth noting that if a
16747 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016748 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016749 through both the server queue and the backend queue unless a redispatch
16750 occurs.
16751
16752 - "captured_request_headers" is a list of headers captured in the request due
16753 to the presence of the "capture request header" statement in the frontend.
16754 Multiple headers can be captured, they will be delimited by a vertical bar
16755 ('|'). When no capture is enabled, the braces do not appear, causing a
16756 shift of remaining fields. It is important to note that this field may
16757 contain spaces, and that using it requires a smarter log parser than when
16758 it's not used. Please consult the section "Capturing HTTP headers and
16759 cookies" below for more details.
16760
16761 - "captured_response_headers" is a list of headers captured in the response
16762 due to the presence of the "capture response header" statement in the
16763 frontend. Multiple headers can be captured, they will be delimited by a
16764 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16765 causing a shift of remaining fields. It is important to note that this
16766 field may contain spaces, and that using it requires a smarter log parser
16767 than when it's not used. Please consult the section "Capturing HTTP headers
16768 and cookies" below for more details.
16769
16770 - "http_request" is the complete HTTP request line, including the method,
16771 request and HTTP version string. Non-printable characters are encoded (see
16772 below the section "Non-printable characters"). This is always the last
16773 field, and it is always delimited by quotes and is the only one which can
16774 contain quotes. If new fields are added to the log format, they will be
16775 added before this field. This field might be truncated if the request is
16776 huge and does not fit in the standard syslog buffer (1024 characters). This
16777 is the reason why this field must always remain the last one.
16778
16779
Cyril Bontédc4d9032012-04-08 21:57:39 +0200167808.2.4. Custom log format
16781------------------------
William Lallemand48940402012-01-30 16:47:22 +010016782
Willy Tarreau2beef582012-12-20 17:22:52 +010016783The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016784mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016785
Davor Ocelice9ed2812017-12-25 17:49:28 +010016786HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016787Variables can take arguments using braces ('{}'), and multiple arguments are
16788separated by commas within the braces. Flags may be added or removed by
16789prefixing them with a '+' or '-' sign.
16790
16791Special variable "%o" may be used to propagate its flags to all other
16792variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016793("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016794
Willy Tarreauc8368452012-12-21 00:09:23 +010016795If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016796as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016797less common information such as the client's SSL certificate's DN, or to log
16798the key that would be used to store an entry into a stick table.
16799
William Lallemand48940402012-01-30 16:47:22 +010016800Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016801In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016802in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016803
Dragan Dosen835b9212016-02-12 13:23:03 +010016804Note: when using the RFC5424 syslog message format, the characters '"',
16805'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16806https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16807such cases, the use of the flag "E" should be considered.
16808
William Lallemand48940402012-01-30 16:47:22 +010016809Flags are :
16810 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016811 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016812 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16813 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016814
16815 Example:
16816
16817 log-format %T\ %t\ Some\ Text
16818 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16819
Dragan Dosen835b9212016-02-12 13:23:03 +010016820 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16821
William Lallemand48940402012-01-30 16:47:22 +010016822At the moment, the default HTTP format is defined this way :
16823
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016824 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16825 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016826
William Lallemandbddd4fd2012-02-27 11:23:10 +010016827the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016828
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016829 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16830 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16831 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016832
William Lallemandbddd4fd2012-02-27 11:23:10 +010016833and the default TCP format is defined this way :
16834
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016835 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16836 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016837
William Lallemand48940402012-01-30 16:47:22 +010016838Please refer to the table below for currently defined variables :
16839
William Lallemandbddd4fd2012-02-27 11:23:10 +010016840 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016841 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016842 +---+------+-----------------------------------------------+-------------+
16843 | | %o | special variable, apply flags on all next var | |
16844 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016845 | | %B | bytes_read (from server to client) | numeric |
16846 | H | %CC | captured_request_cookie | string |
16847 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016848 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016849 | H | %HM | HTTP method (ex: POST) | string |
16850 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016851 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016852 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16853 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016854 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016855 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016856 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016857 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016858 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016859 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016860 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016861 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16862 | H | %Ti | idle time before the HTTP request | numeric |
16863 | H | %Tq | Th + Ti + TR | numeric |
16864 | H | %TR | time to receive the full request from 1st byte| numeric |
16865 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016866 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016867 | | %Tt | Tt | numeric |
16868 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016869 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016870 | | %ac | actconn | numeric |
16871 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016872 | | %bc | beconn (backend concurrent connections) | numeric |
16873 | | %bi | backend_source_ip (connecting address) | IP |
16874 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016875 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016876 | | %ci | client_ip (accepted address) | IP |
16877 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016878 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016879 | | %fc | feconn (frontend concurrent connections) | numeric |
16880 | | %fi | frontend_ip (accepting address) | IP |
16881 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016882 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016883 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016884 | | %hr | captured_request_headers default style | string |
16885 | | %hrl | captured_request_headers CLF style | string list |
16886 | | %hs | captured_response_headers default style | string |
16887 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016888 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016889 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016890 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016891 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016892 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016893 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016894 | | %sc | srv_conn (server concurrent connections) | numeric |
16895 | | %si | server_IP (target address) | IP |
16896 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016897 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016898 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16899 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016900 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016901 | H | %tr | date_time of HTTP request | date |
16902 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020016903 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016904 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016905 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016906 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016907
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016908 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016909
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016910
169118.2.5. Error log format
16912-----------------------
16913
16914When an incoming connection fails due to an SSL handshake or an invalid PROXY
16915protocol header, haproxy will log the event using a shorter, fixed line format.
16916By default, logs are emitted at the LOG_INFO level, unless the option
16917"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016918will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016919logged if the "dontlognull" option is set.
16920
16921The format looks like this :
16922
16923 >>> Dec 3 18:27:14 localhost \
16924 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
16925 Connection error during SSL handshake
16926
16927 Field Format Extract from the example above
16928 1 process_name '[' pid ']:' haproxy[6103]:
16929 2 client_ip ':' client_port 127.0.0.1:56059
16930 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
16931 4 frontend_name "/" bind_name ":" frt/f1:
16932 5 message Connection error during SSL handshake
16933
16934These fields just provide minimal information to help debugging connection
16935failures.
16936
16937
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169388.3. Advanced logging options
16939-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016940
16941Some advanced logging options are often looked for but are not easy to find out
16942just by looking at the various options. Here is an entry point for the few
16943options which can enable better logging. Please refer to the keywords reference
16944for more information about their usage.
16945
16946
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169478.3.1. Disabling logging of external tests
16948------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016949
16950It is quite common to have some monitoring tools perform health checks on
16951haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
16952commercial load-balancer, and sometimes it will simply be a more complete
16953monitoring system such as Nagios. When the tests are very frequent, users often
16954ask how to disable logging for those checks. There are three possibilities :
16955
16956 - if connections come from everywhere and are just TCP probes, it is often
16957 desired to simply disable logging of connections without data exchange, by
16958 setting "option dontlognull" in the frontend. It also disables logging of
16959 port scans, which may or may not be desired.
16960
16961 - if the connection come from a known source network, use "monitor-net" to
16962 declare this network as monitoring only. Any host in this network will then
16963 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016964 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016965 such as other load-balancers.
16966
16967 - if the tests are performed on a known URI, use "monitor-uri" to declare
16968 this URI as dedicated to monitoring. Any host sending this request will
16969 only get the result of a health-check, and the request will not be logged.
16970
16971
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169728.3.2. Logging before waiting for the session to terminate
16973----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016974
16975The problem with logging at end of connection is that you have no clue about
16976what is happening during very long sessions, such as remote terminal sessions
16977or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010016978"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016979just before data transfer begins. This means that in case of TCP, it will still
16980log the connection status to the server, and in case of HTTP, it will log just
16981after processing the server headers. In this case, the number of bytes reported
16982is the number of header bytes sent to the client. In order to avoid confusion
16983with normal logs, the total time field and the number of bytes are prefixed
16984with a '+' sign which means that real numbers are certainly larger.
16985
16986
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169878.3.3. Raising log level upon errors
16988------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016989
16990Sometimes it is more convenient to separate normal traffic from errors logs,
16991for instance in order to ease error monitoring from log files. When the option
16992"log-separate-errors" is used, connections which experience errors, timeouts,
16993retries, redispatches or HTTP status codes 5xx will see their syslog level
16994raised from "info" to "err". This will help a syslog daemon store the log in
16995a separate file. It is very important to keep the errors in the normal traffic
16996file too, so that log ordering is not altered. You should also be careful if
16997you already have configured your syslog daemon to store all logs higher than
16998"notice" in an "admin" file, because the "err" level is higher than "notice".
16999
17000
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170018.3.4. Disabling logging of successful connections
17002--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017003
17004Although this may sound strange at first, some large sites have to deal with
17005multiple thousands of logs per second and are experiencing difficulties keeping
17006them intact for a long time or detecting errors within them. If the option
17007"dontlog-normal" is set on the frontend, all normal connections will not be
17008logged. In this regard, a normal connection is defined as one without any
17009error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
17010and a response with a status 5xx is not considered normal and will be logged
17011too. Of course, doing is is really discouraged as it will remove most of the
17012useful information from the logs. Do this only if you have no other
17013alternative.
17014
17015
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170168.4. Timing events
17017------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017018
17019Timers provide a great help in troubleshooting network problems. All values are
17020reported in milliseconds (ms). These timers should be used in conjunction with
17021the session termination flags. In TCP mode with "option tcplog" set on the
17022frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017023mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
17024addition, three other measures are provided, "Th", "Ti", and "Tq".
17025
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010017026Timings events in HTTP mode:
17027
17028 first request 2nd request
17029 |<-------------------------------->|<-------------- ...
17030 t tr t tr ...
17031 ---|----|----|----|----|----|----|----|----|--
17032 : Th Ti TR Tw Tc Tr Td : Ti ...
17033 :<---- Tq ---->: :
17034 :<-------------- Tt -------------->:
17035 :<--------- Ta --------->:
17036
17037Timings events in TCP mode:
17038
17039 TCP session
17040 |<----------------->|
17041 t t
17042 ---|----|----|----|----|---
17043 | Th Tw Tc Td |
17044 |<------ Tt ------->|
17045
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017046 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017047 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017048 only happen once during the whole connection's lifetime. A large time here
17049 may indicate that the client only pre-established the connection without
17050 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017051 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017052 SSL handshake was very expensive to compute. Please note that this time is
17053 reported only before the first request, so it is safe to average it over
17054 all request to calculate the amortized value. The second and subsequent
17055 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017056
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017057 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17058 counts between the end of the handshakes and the first byte of the HTTP
17059 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017060 to count after the end of the transmission the previous response. When a
17061 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17062 after the previous request. Some browsers pre-establish connections to a
17063 server in order to reduce the latency of a future request, and keep them
17064 pending until they need it. This delay will be reported as the idle time. A
17065 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017066
17067 - TR: total time to get the client request (HTTP mode only). It's the time
17068 elapsed between the first bytes received and the moment the proxy received
17069 the empty line marking the end of the HTTP headers. The value "-1"
17070 indicates that the end of headers has never been seen. This happens when
17071 the client closes prematurely or times out. This time is usually very short
17072 since most requests fit in a single packet. A large time may indicate a
17073 request typed by hand during a test.
17074
17075 - Tq: total time to get the client request from the accept date or since the
17076 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017077 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 +020017078 returns -1 as well. This timer used to be very useful before the arrival of
17079 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17080 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17081 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017082
17083 - Tw: total time spent in the queues waiting for a connection slot. It
17084 accounts for backend queue as well as the server queues, and depends on the
17085 queue size, and the time needed for the server to complete previous
17086 requests. The value "-1" means that the request was killed before reaching
17087 the queue, which is generally what happens with invalid or denied requests.
17088
17089 - Tc: total time to establish the TCP connection to the server. It's the time
17090 elapsed between the moment the proxy sent the connection request, and the
17091 moment it was acknowledged by the server, or between the TCP SYN packet and
17092 the matching SYN/ACK packet in return. The value "-1" means that the
17093 connection never established.
17094
17095 - Tr: server response time (HTTP mode only). It's the time elapsed between
17096 the moment the TCP connection was established to the server and the moment
17097 the server sent its complete response headers. It purely shows its request
17098 processing time, without the network overhead due to the data transmission.
17099 It is worth noting that when the client has data to send to the server, for
17100 instance during a POST request, the time already runs, and this can distort
17101 apparent response time. For this reason, it's generally wise not to trust
17102 too much this field for POST requests initiated from clients behind an
17103 untrusted network. A value of "-1" here means that the last the response
17104 header (empty line) was never seen, most likely because the server timeout
17105 stroke before the server managed to process the request.
17106
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017107 - Ta: total active time for the HTTP request, between the moment the proxy
17108 received the first byte of the request header and the emission of the last
17109 byte of the response body. The exception is when the "logasap" option is
17110 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17111 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17112 by subtracting other timers when valid :
17113
17114 Td = Ta - (TR + Tw + Tc + Tr)
17115
17116 Timers with "-1" values have to be excluded from this equation. Note that
17117 "Ta" can never be negative.
17118
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017119 - Tt: total session duration time, between the moment the proxy accepted it
17120 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017121 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17122 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017123 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017124
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017125 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017126
17127 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017128 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17129 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017130
17131These timers provide precious indications on trouble causes. Since the TCP
17132protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17133that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017134due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17135"Tt" is close to a timeout value specified in the configuration, it often means
17136that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017137
17138Most common cases :
17139
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017140 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17141 the client and the proxy. This is very rare on local networks but might
17142 happen when clients are on far remote networks and send large requests. It
17143 may happen that values larger than usual appear here without any network
17144 cause. Sometimes, during an attack or just after a resource starvation has
17145 ended, haproxy may accept thousands of connections in a few milliseconds.
17146 The time spent accepting these connections will inevitably slightly delay
17147 processing of other connections, and it can happen that request times in the
17148 order of a few tens of milliseconds are measured after a few thousands of
17149 new connections have been accepted at once. Using one of the keep-alive
17150 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017151 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017152
17153 - If "Tc" is close to 3000, a packet has probably been lost between the
17154 server and the proxy during the server connection phase. This value should
17155 always be very low, such as 1 ms on local networks and less than a few tens
17156 of ms on remote networks.
17157
Willy Tarreau55165fe2009-05-10 12:02:55 +020017158 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17159 to be the average majored by 3000, there are probably some packets lost
17160 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017161
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017162 - If "Ta" is large even for small byte counts, it generally is because
17163 neither the client nor the server decides to close the connection while
17164 haproxy is running in tunnel mode and both have agreed on a keep-alive
17165 connection mode. In order to solve this issue, it will be needed to specify
17166 one of the HTTP options to manipulate keep-alive or close options on either
17167 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17168 important when connection regulation is used with the "maxconn" option on
17169 the servers, since no new connection will be sent to the server until
17170 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017171
17172Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17173
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017174 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017175 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017176 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017177
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017178 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017179 or it aborted too early. Check the session termination flags
17180 then "timeout http-request" and "timeout client" settings.
17181
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017182 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017183 servers were out of order, because the request was invalid
17184 or forbidden by ACL rules. Check the session termination
17185 flags.
17186
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017187 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17188 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017189 Check the session termination flags, then check the
17190 "timeout connect" setting. Note that the tarpit action might
17191 return similar-looking patterns, with "Tw" equal to the time
17192 the client connection was maintained open.
17193
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017194 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017195 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017196 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017197 termination flags, then check the "timeout server" setting.
17198
17199
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172008.5. Session state at disconnection
17201-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017202
17203TCP and HTTP logs provide a session termination indicator in the
17204"termination_state" field, just before the number of active connections. It is
172052-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17206each of which has a special meaning :
17207
17208 - On the first character, a code reporting the first event which caused the
17209 session to terminate :
17210
17211 C : the TCP session was unexpectedly aborted by the client.
17212
17213 S : the TCP session was unexpectedly aborted by the server, or the
17214 server explicitly refused it.
17215
17216 P : the session was prematurely aborted by the proxy, because of a
17217 connection limit enforcement, because a DENY filter was matched,
17218 because of a security check which detected and blocked a dangerous
17219 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010017220 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020017221
17222 L : the session was locally processed by haproxy and was not passed to
17223 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017224
17225 R : a resource on the proxy has been exhausted (memory, sockets, source
17226 ports, ...). Usually, this appears during the connection phase, and
17227 system logs should contain a copy of the precise error. If this
17228 happens, it must be considered as a very serious anomaly which
17229 should be fixed as soon as possible by any means.
17230
17231 I : an internal error was identified by the proxy during a self-check.
17232 This should NEVER happen, and you are encouraged to report any log
17233 containing this, because this would almost certainly be a bug. It
17234 would be wise to preventively restart the process after such an
17235 event too, in case it would be caused by memory corruption.
17236
Simon Horman752dc4a2011-06-21 14:34:59 +090017237 D : the session was killed by haproxy because the server was detected
17238 as down and was configured to kill all connections when going down.
17239
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070017240 U : the session was killed by haproxy on this backup server because an
17241 active server was detected as up and was configured to kill all
17242 backup connections when going up.
17243
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017244 K : the session was actively killed by an admin operating on haproxy.
17245
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017246 c : the client-side timeout expired while waiting for the client to
17247 send or receive data.
17248
17249 s : the server-side timeout expired while waiting for the server to
17250 send or receive data.
17251
17252 - : normal session completion, both the client and the server closed
17253 with nothing left in the buffers.
17254
17255 - on the second character, the TCP or HTTP session state when it was closed :
17256
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017257 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017258 (HTTP mode only). Nothing was sent to any server.
17259
17260 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17261 only happen when servers have a 'maxconn' parameter set. It can
17262 also happen in the global queue after a redispatch consecutive to
17263 a failed attempt to connect to a dying server. If no redispatch is
17264 reported, then no connection attempt was made to any server.
17265
17266 C : the proxy was waiting for the CONNECTION to establish on the
17267 server. The server might at most have noticed a connection attempt.
17268
17269 H : the proxy was waiting for complete, valid response HEADERS from the
17270 server (HTTP only).
17271
17272 D : the session was in the DATA phase.
17273
17274 L : the proxy was still transmitting LAST data to the client while the
17275 server had already finished. This one is very rare as it can only
17276 happen when the client dies while receiving the last packets.
17277
17278 T : the request was tarpitted. It has been held open with the client
17279 during the whole "timeout tarpit" duration or until the client
17280 closed, both of which will be reported in the "Tw" timer.
17281
17282 - : normal session completion after end of data transfer.
17283
17284 - the third character tells whether the persistence cookie was provided by
17285 the client (only in HTTP mode) :
17286
17287 N : the client provided NO cookie. This is usually the case for new
17288 visitors, so counting the number of occurrences of this flag in the
17289 logs generally indicate a valid trend for the site frequentation.
17290
17291 I : the client provided an INVALID cookie matching no known server.
17292 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017293 cookies between HTTP/HTTPS sites, persistence conditionally
17294 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017295
17296 D : the client provided a cookie designating a server which was DOWN,
17297 so either "option persist" was used and the client was sent to
17298 this server, or it was not set and the client was redispatched to
17299 another server.
17300
Willy Tarreau996a92c2010-10-13 19:30:47 +020017301 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017302 server.
17303
Willy Tarreau996a92c2010-10-13 19:30:47 +020017304 E : the client provided a valid cookie, but with a last date which was
17305 older than what is allowed by the "maxidle" cookie parameter, so
17306 the cookie is consider EXPIRED and is ignored. The request will be
17307 redispatched just as if there was no cookie.
17308
17309 O : the client provided a valid cookie, but with a first date which was
17310 older than what is allowed by the "maxlife" cookie parameter, so
17311 the cookie is consider too OLD and is ignored. The request will be
17312 redispatched just as if there was no cookie.
17313
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017314 U : a cookie was present but was not used to select the server because
17315 some other server selection mechanism was used instead (typically a
17316 "use-server" rule).
17317
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017318 - : does not apply (no cookie set in configuration).
17319
17320 - the last character reports what operations were performed on the persistence
17321 cookie returned by the server (only in HTTP mode) :
17322
17323 N : NO cookie was provided by the server, and none was inserted either.
17324
17325 I : no cookie was provided by the server, and the proxy INSERTED one.
17326 Note that in "cookie insert" mode, if the server provides a cookie,
17327 it will still be overwritten and reported as "I" here.
17328
Willy Tarreau996a92c2010-10-13 19:30:47 +020017329 U : the proxy UPDATED the last date in the cookie that was presented by
17330 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017331 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017332 date indicated in the cookie. If any other change happens, such as
17333 a redispatch, then the cookie will be marked as inserted instead.
17334
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017335 P : a cookie was PROVIDED by the server and transmitted as-is.
17336
17337 R : the cookie provided by the server was REWRITTEN by the proxy, which
17338 happens in "cookie rewrite" or "cookie prefix" modes.
17339
17340 D : the cookie provided by the server was DELETED by the proxy.
17341
17342 - : does not apply (no cookie set in configuration).
17343
Willy Tarreau996a92c2010-10-13 19:30:47 +020017344The combination of the two first flags gives a lot of information about what
17345was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017346helpful to detect server saturation, network troubles, local system resource
17347starvation, attacks, etc...
17348
17349The most common termination flags combinations are indicated below. They are
17350alphabetically sorted, with the lowercase set just after the upper case for
17351easier finding and understanding.
17352
17353 Flags Reason
17354
17355 -- Normal termination.
17356
17357 CC The client aborted before the connection could be established to the
17358 server. This can happen when haproxy tries to connect to a recently
17359 dead (or unchecked) server, and the client aborts while haproxy is
17360 waiting for the server to respond or for "timeout connect" to expire.
17361
17362 CD The client unexpectedly aborted during data transfer. This can be
17363 caused by a browser crash, by an intermediate equipment between the
17364 client and haproxy which decided to actively break the connection,
17365 by network routing issues between the client and haproxy, or by a
17366 keep-alive session between the server and the client terminated first
17367 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010017368
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017369 cD The client did not send nor acknowledge any data for as long as the
17370 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020017371 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017372
17373 CH The client aborted while waiting for the server to start responding.
17374 It might be the server taking too long to respond or the client
17375 clicking the 'Stop' button too fast.
17376
17377 cH The "timeout client" stroke while waiting for client data during a
17378 POST request. This is sometimes caused by too large TCP MSS values
17379 for PPPoE networks which cannot transport full-sized packets. It can
17380 also happen when client timeout is smaller than server timeout and
17381 the server takes too long to respond.
17382
17383 CQ The client aborted while its session was queued, waiting for a server
17384 with enough empty slots to accept it. It might be that either all the
17385 servers were saturated or that the assigned server was taking too
17386 long a time to respond.
17387
17388 CR The client aborted before sending a full HTTP request. Most likely
17389 the request was typed by hand using a telnet client, and aborted
17390 too early. The HTTP status code is likely a 400 here. Sometimes this
17391 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017392 and the client. "option http-ignore-probes" can be used to ignore
17393 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017394
17395 cR The "timeout http-request" stroke before the client sent a full HTTP
17396 request. This is sometimes caused by too large TCP MSS values on the
17397 client side for PPPoE networks which cannot transport full-sized
17398 packets, or by clients sending requests by hand and not typing fast
17399 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017400 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017401 some browsers started to implement a "pre-connect" feature consisting
17402 in speculatively connecting to some recently visited web sites just
17403 in case the user would like to visit them. This results in many
17404 connections being established to web sites, which end up in 408
17405 Request Timeout if the timeout strikes first, or 400 Bad Request when
17406 the browser decides to close them first. These ones pollute the log
17407 and feed the error counters. Some versions of some browsers have even
17408 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017409 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017410 http-ignore-probes" in the frontend, resulting in connections with
17411 zero data transfer to be totally ignored. This will definitely hide
17412 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017413
17414 CT The client aborted while its session was tarpitted. It is important to
17415 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017416 wrong tarpit rules have been written. If a lot of them happen, it
17417 might make sense to lower the "timeout tarpit" value to something
17418 closer to the average reported "Tw" timer, in order not to consume
17419 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017420
Willy Tarreau570f2212013-06-10 16:42:09 +020017421 LR The request was intercepted and locally handled by haproxy. Generally
17422 it means that this was a redirect or a stats request.
17423
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017424 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017425 the TCP connection (the proxy received a TCP RST or an ICMP message
17426 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017427 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017428 or no ARP response on local network). When this happens in HTTP mode,
17429 the status code is likely a 502 or 503 here.
17430
17431 sC The "timeout connect" stroke before a connection to the server could
17432 complete. When this happens in HTTP mode, the status code is likely a
17433 503 or 504 here.
17434
17435 SD The connection to the server died with an error during the data
17436 transfer. This usually means that haproxy has received an RST from
17437 the server or an ICMP message from an intermediate equipment while
17438 exchanging data with the server. This can be caused by a server crash
17439 or by a network issue on an intermediate equipment.
17440
17441 sD The server did not send nor acknowledge any data for as long as the
17442 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017443 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017444 load-balancers, ...), as well as keep-alive sessions maintained
17445 between the client and the server expiring first on haproxy.
17446
17447 SH The server aborted before sending its full HTTP response headers, or
17448 it crashed while processing the request. Since a server aborting at
17449 this moment is very rare, it would be wise to inspect its logs to
17450 control whether it crashed and why. The logged request may indicate a
17451 small set of faulty requests, demonstrating bugs in the application.
17452 Sometimes this might also be caused by an IDS killing the connection
17453 between haproxy and the server.
17454
17455 sH The "timeout server" stroke before the server could return its
17456 response headers. This is the most common anomaly, indicating too
17457 long transactions, probably caused by server or database saturation.
17458 The immediate workaround consists in increasing the "timeout server"
17459 setting, but it is important to keep in mind that the user experience
17460 will suffer from these long response times. The only long term
17461 solution is to fix the application.
17462
17463 sQ The session spent too much time in queue and has been expired. See
17464 the "timeout queue" and "timeout connect" settings to find out how to
17465 fix this if it happens too often. If it often happens massively in
17466 short periods, it may indicate general problems on the affected
17467 servers due to I/O or database congestion, or saturation caused by
17468 external attacks.
17469
17470 PC The proxy refused to establish a connection to the server because the
17471 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017472 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017473 so that it does not happen anymore. This status is very rare and
17474 might happen when the global "ulimit-n" parameter is forced by hand.
17475
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017476 PD The proxy blocked an incorrectly formatted chunked encoded message in
17477 a request or a response, after the server has emitted its headers. In
17478 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017479 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017480 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017481
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017482 PH The proxy blocked the server's response, because it was invalid,
17483 incomplete, dangerous (cache control), or matched a security filter.
17484 In any case, an HTTP 502 error is sent to the client. One possible
17485 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017486 containing unauthorized characters. It is also possible but quite
17487 rare, that the proxy blocked a chunked-encoding request from the
17488 client due to an invalid syntax, before the server responded. In this
17489 case, an HTTP 400 error is sent to the client and reported in the
17490 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017491
17492 PR The proxy blocked the client's HTTP request, either because of an
17493 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17494 the client, or because a deny filter matched, in which case it
17495 returned an HTTP 403 error.
17496
17497 PT The proxy blocked the client's request and has tarpitted its
17498 connection before returning it a 500 server error. Nothing was sent
17499 to the server. The connection was maintained open for as long as
17500 reported by the "Tw" timer field.
17501
17502 RC A local resource has been exhausted (memory, sockets, source ports)
17503 preventing the connection to the server from establishing. The error
17504 logs will tell precisely what was missing. This is very rare and can
17505 only be solved by proper system tuning.
17506
Willy Tarreau996a92c2010-10-13 19:30:47 +020017507The combination of the two last flags gives a lot of information about how
17508persistence was handled by the client, the server and by haproxy. This is very
17509important to troubleshoot disconnections, when users complain they have to
17510re-authenticate. The commonly encountered flags are :
17511
17512 -- Persistence cookie is not enabled.
17513
17514 NN No cookie was provided by the client, none was inserted in the
17515 response. For instance, this can be in insert mode with "postonly"
17516 set on a GET request.
17517
17518 II A cookie designating an invalid server was provided by the client,
17519 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017520 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017521 value can be presented by a client when no other server knows it.
17522
17523 NI No cookie was provided by the client, one was inserted in the
17524 response. This typically happens for first requests from every user
17525 in "insert" mode, which makes it an easy way to count real users.
17526
17527 VN A cookie was provided by the client, none was inserted in the
17528 response. This happens for most responses for which the client has
17529 already got a cookie.
17530
17531 VU A cookie was provided by the client, with a last visit date which is
17532 not completely up-to-date, so an updated cookie was provided in
17533 response. This can also happen if there was no date at all, or if
17534 there was a date but the "maxidle" parameter was not set, so that the
17535 cookie can be switched to unlimited time.
17536
17537 EI A cookie was provided by the client, with a last visit date which is
17538 too old for the "maxidle" parameter, so the cookie was ignored and a
17539 new cookie was inserted in the response.
17540
17541 OI A cookie was provided by the client, with a first visit date which is
17542 too old for the "maxlife" parameter, so the cookie was ignored and a
17543 new cookie was inserted in the response.
17544
17545 DI The server designated by the cookie was down, a new server was
17546 selected and a new cookie was emitted in the response.
17547
17548 VI The server designated by the cookie was not marked dead but could not
17549 be reached. A redispatch happened and selected another one, which was
17550 then advertised in the response.
17551
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017552
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175538.6. Non-printable characters
17554-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017555
17556In order not to cause trouble to log analysis tools or terminals during log
17557consulting, non-printable characters are not sent as-is into log files, but are
17558converted to the two-digits hexadecimal representation of their ASCII code,
17559prefixed by the character '#'. The only characters that can be logged without
17560being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17561escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17562is the same for the character '"' which becomes "#22", as well as '{', '|' and
17563'}' when logging headers.
17564
17565Note that the space character (' ') is not encoded in headers, which can cause
17566issues for tools relying on space count to locate fields. A typical header
17567containing spaces is "User-Agent".
17568
17569Last, it has been observed that some syslog daemons such as syslog-ng escape
17570the quote ('"') with a backslash ('\'). The reverse operation can safely be
17571performed since no quote may appear anywhere else in the logs.
17572
17573
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175748.7. Capturing HTTP cookies
17575---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017576
17577Cookie capture simplifies the tracking a complete user session. This can be
17578achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017579section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017580cookie will simultaneously be checked in the request ("Cookie:" header) and in
17581the response ("Set-Cookie:" header). The respective values will be reported in
17582the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017583locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017584not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17585user switches to a new session for example, because the server will reassign it
17586a new cookie. It is also possible to detect if a server unexpectedly sets a
17587wrong cookie to a client, leading to session crossing.
17588
17589 Examples :
17590 # capture the first cookie whose name starts with "ASPSESSION"
17591 capture cookie ASPSESSION len 32
17592
17593 # capture the first cookie whose name is exactly "vgnvisitor"
17594 capture cookie vgnvisitor= len 32
17595
17596
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175978.8. Capturing HTTP headers
17598---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017599
17600Header captures are useful to track unique request identifiers set by an upper
17601proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
17602the response, one can search for information about the response length, how the
17603server asked the cache to behave, or an object location during a redirection.
17604
17605Header captures are performed using the "capture request header" and "capture
17606response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017607section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017608
17609It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017610time. Non-existent headers are logged as empty strings, and if one header
17611appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017612are grouped within braces '{' and '}' in the same order as they were declared,
17613and delimited with a vertical bar '|' without any space. Response headers
17614follow the same representation, but are displayed after a space following the
17615request headers block. These blocks are displayed just before the HTTP request
17616in the logs.
17617
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017618As a special case, it is possible to specify an HTTP header capture in a TCP
17619frontend. The purpose is to enable logging of headers which will be parsed in
17620an HTTP backend if the request is then switched to this HTTP backend.
17621
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017622 Example :
17623 # This instance chains to the outgoing proxy
17624 listen proxy-out
17625 mode http
17626 option httplog
17627 option logasap
17628 log global
17629 server cache1 192.168.1.1:3128
17630
17631 # log the name of the virtual server
17632 capture request header Host len 20
17633
17634 # log the amount of data uploaded during a POST
17635 capture request header Content-Length len 10
17636
17637 # log the beginning of the referrer
17638 capture request header Referer len 20
17639
17640 # server name (useful for outgoing proxies only)
17641 capture response header Server len 20
17642
17643 # logging the content-length is useful with "option logasap"
17644 capture response header Content-Length len 10
17645
Davor Ocelice9ed2812017-12-25 17:49:28 +010017646 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017647 capture response header Cache-Control len 8
17648
17649 # the Via header will report the next proxy's name
17650 capture response header Via len 20
17651
17652 # log the URL location during a redirection
17653 capture response header Location len 20
17654
17655 >>> Aug 9 20:26:09 localhost \
17656 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
17657 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
17658 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
17659 "GET http://fr.adserver.yahoo.com/"
17660
17661 >>> Aug 9 20:30:46 localhost \
17662 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
17663 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
17664 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017665 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017666
17667 >>> Aug 9 20:30:46 localhost \
17668 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
17669 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
17670 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
17671 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017672 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017673
17674
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176758.9. Examples of logs
17676---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017677
17678These are real-world examples of logs accompanied with an explanation. Some of
17679them have been made up by hand. The syslog part has been removed for better
17680reading. Their sole purpose is to explain how to decipher them.
17681
17682 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17683 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17684 "HEAD / HTTP/1.0"
17685
17686 => long request (6.5s) entered by hand through 'telnet'. The server replied
17687 in 147 ms, and the session ended normally ('----')
17688
17689 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17690 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17691 0/9 "HEAD / HTTP/1.0"
17692
17693 => Idem, but the request was queued in the global queue behind 9 other
17694 requests, and waited there for 1230 ms.
17695
17696 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17697 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17698 "GET /image.iso HTTP/1.0"
17699
17700 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017701 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017702 14 ms, 243 bytes of headers were sent to the client, and total time from
17703 accept to first data byte is 30 ms.
17704
17705 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17706 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17707 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17708
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020017709 => the proxy blocked a server response either because of an "http-response
17710 deny" rule, or because the response was improperly formatted and not
17711 HTTP-compliant, or because it blocked sensitive information which risked
17712 being cached. In this case, the response is replaced with a "502 bad
17713 gateway". The flags ("PH--") tell us that it was haproxy who decided to
17714 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017715
17716 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017717 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 +010017718
17719 => the client never completed its request and aborted itself ("C---") after
17720 8.5s, while the proxy was waiting for the request headers ("-R--").
17721 Nothing was sent to any server.
17722
17723 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17724 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17725
17726 => The client never completed its request, which was aborted by the
17727 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017728 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017729 send a 408 return code to the client.
17730
17731 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17732 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17733
17734 => This log was produced with "option tcplog". The client timed out after
17735 5 seconds ("c----").
17736
17737 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17738 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017739 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017740
17741 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017742 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017743 (config says 'retries 3'), and no redispatch (otherwise we would have
17744 seen "/+3"). Status code 503 was returned to the client. There were 115
17745 connections on this server, 202 connections on this proxy, and 205 on
17746 the global process. It is possible that the server refused the
17747 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017748
Willy Tarreau52b2d222011-09-07 23:48:48 +020017749
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200177509. Supported filters
17751--------------------
17752
17753Here are listed officially supported filters with the list of parameters they
17754accept. Depending on compile options, some of these filters might be
17755unavailable. The list of available filters is reported in haproxy -vv.
17756
17757See also : "filter"
17758
177599.1. Trace
17760----------
17761
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017762filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017763
17764 Arguments:
17765 <name> is an arbitrary name that will be reported in
17766 messages. If no name is provided, "TRACE" is used.
17767
17768 <random-parsing> enables the random parsing of data exchanged between
17769 the client and the server. By default, this filter
17770 parses all available data. With this parameter, it
17771 only parses a random amount of the available data.
17772
Davor Ocelice9ed2812017-12-25 17:49:28 +010017773 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017774 default, this filter forwards all previously parsed
17775 data. With this parameter, it only forwards a random
17776 amount of the parsed data.
17777
Davor Ocelice9ed2812017-12-25 17:49:28 +010017778 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017779
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017780This filter can be used as a base to develop new filters. It defines all
17781callbacks and print a message on the standard error stream (stderr) with useful
17782information for all of them. It may be useful to debug the activity of other
17783filters or, quite simply, HAProxy's activity.
17784
17785Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17786tests the behavior of a filter that parses data exchanged between a client and
17787a server by adding some latencies in the processing.
17788
17789
177909.2. HTTP compression
17791---------------------
17792
17793filter compression
17794
17795The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17796keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020017797when no other filter is used, it is enough. When used with the cache or the
17798fcgi-app enabled, it is also enough. In this case, the compression is always
17799done after the response is stored in the cache. But it is mandatory to
17800explicitly use a filter line to enable the HTTP compression when at least one
17801filter other than the cache or the fcgi-app is used for the same
17802listener/frontend/backend. This is important to know the filters evaluation
17803order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017804
Christopher Fauletb30b3102019-09-12 23:03:09 +020017805See also : "compression", section 9.4 about the cache filter and section 9.5
17806 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017807
17808
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200178099.3. Stream Processing Offload Engine (SPOE)
17810--------------------------------------------
17811
17812filter spoe [engine <name>] config <file>
17813
17814 Arguments :
17815
17816 <name> is the engine name that will be used to find the right scope in
17817 the configuration file. If not provided, all the file will be
17818 parsed.
17819
17820 <file> is the path of the engine configuration file. This file can
17821 contain configuration of several engines. In this case, each
17822 part must be placed in its own scope.
17823
17824The Stream Processing Offload Engine (SPOE) is a filter communicating with
17825external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017826streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017827exchanged with them are configured in dedicated files, for the main part. It
17828also requires dedicated backends, defined in HAProxy configuration.
17829
17830SPOE communicates with external components using an in-house binary protocol,
17831the Stream Processing Offload Protocol (SPOP).
17832
Tim Düsterhus4896c442016-11-29 02:15:19 +010017833For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017834"doc/SPOE.txt".
17835
17836Important note:
17837 The SPOE filter is highly experimental for now and was not heavily
17838 tested. It is really not production ready. So use it carefully.
17839
Christopher Faulet99a17a22018-12-11 09:18:27 +0100178409.4. Cache
17841----------
17842
17843filter cache <name>
17844
17845 Arguments :
17846
17847 <name> is name of the cache section this filter will use.
17848
17849The cache uses a filter to store cacheable responses. The HTTP rules
17850"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050017851cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020017852other filters than fcgi-app or compression are used, it is enough. In such
17853case, the compression filter is always evaluated after the cache filter. But it
17854is mandatory to explicitly use a filter line to use a cache when at least one
17855filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010017856listener/frontend/backend. This is important to know the filters evaluation
17857order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010017858
Christopher Fauletb30b3102019-09-12 23:03:09 +020017859See also : section 9.2 about the compression filter, section 9.5 about the
17860 fcgi-app filter and section 6 about cache.
17861
17862
178639.5. Fcgi-app
17864-------------
17865
17866filter fcg-app <name>
17867
17868 Arguments :
17869
17870 <name> is name of the fcgi-app section this filter will use.
17871
17872The FastCGI application uses a filter to evaluate all custom parameters on the
17873request path, and to process the headers on the response path. the <name> must
17874reference an existing fcgi-app section. The directive "use-fcgi-app" should be
17875used to define the application to use. By default the corresponding filter is
17876implicitly defined. And when no other filters than cache or compression are
17877used, it is enough. But it is mandatory to explicitly use a filter line to a
17878fcgi-app when at least one filter other than the compression or the cache is
17879used for the same backend. This is important to know the filters evaluation
17880order.
17881
17882See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
17883 about the cache filter and section 10 about FastCGI application.
17884
17885
1788610. FastCGI applications
17887-------------------------
17888
17889HAProxy is able to send HTTP requests to Responder FastCGI applications. This
17890feature was added in HAProxy 2.1. To do so, servers must be configured to use
17891the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
17892FastCGI application must be configured and used by the backend managing these
17893servers (using the keyword "use-fcgi-app" into the proxy section). Several
17894FastCGI applications may be defined, but only one can be used at a time by a
17895backend.
17896
17897HAProxy implements all features of the FastCGI specification for Responder
17898application. Especially it is able to multiplex several requests on a simple
17899connection.
17900
1790110.1. Setup
17902-----------
17903
1790410.1.1. Fcgi-app section
17905--------------------------
17906
17907fcgi-app <name>
17908 Declare a FastCGI application named <name>. To be valid, at least the
17909 document root must be defined.
17910
17911acl <aclname> <criterion> [flags] [operator] <value> ...
17912 Declare or complete an access list.
17913
17914 See "acl" keyword in section 4.2 and section 7 about ACL usage for
17915 details. ACLs defined for a FastCGI application are private. They cannot be
17916 used by any other application or by any proxy. In the same way, ACLs defined
17917 in any other section are not usable by a FastCGI application. However,
17918 Pre-defined ACLs are available.
17919
17920docroot <path>
17921 Define the document root on the remote host. <path> will be used to build
17922 the default value of FastCGI parameters SCRIPT_FILENAME and
17923 PATH_TRANSLATED. It is a mandatory setting.
17924
17925index <script-name>
17926 Define the script name that will be appended after an URI that ends with a
17927 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
17928 is an optional setting.
17929
17930 Example :
17931 index index.php
17932
17933log-stderr global
17934log-stderr <address> [len <length>] [format <format>]
17935 [sample <ranges>:<smp_size>] <facility> [<level> [<minlevel>]]
17936 Enable logging of STDERR messages reported by the FastCGI application.
17937
17938 See "log" keyword in section 4.2 for details. It is an optional setting. By
17939 default STDERR messages are ignored.
17940
17941pass-header <name> [ { if | unless } <condition> ]
17942 Specify the name of a request header which will be passed to the FastCGI
17943 application. It may optionally be followed by an ACL-based condition, in
17944 which case it will only be evaluated if the condition is true.
17945
17946 Most request headers are already available to the FastCGI application,
17947 prefixed with "HTTP_". Thus, this directive is only required to pass headers
17948 that are purposefully omitted. Currently, the headers "Authorization",
17949 "Proxy-Authorization" and hop-by-hop headers are omitted.
17950
17951 Note that the headers "Content-type" and "Content-length" are never passed to
17952 the FastCGI application because they are already converted into parameters.
17953
17954path-info <regex>
17955 Define a regular expression to extract the script-name and the path-info
17956 from the URI. Thus, <regex> should have two captures: the first one to
17957 capture the script name and the second one to capture the path-info. It is an
17958 optional setting. If it is not defined, no matching is performed on the
17959 URI. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not filled.
17960
17961 Example :
17962 path-info ^(/.+\.php)(/.*)?$
17963
17964option get-values
17965no option get-values
17966 Enable or disable the retrieve of variables about connection management.
17967
17968 HAproxy is able to send the record FCGI_GET_VALUES on connection
17969 establishment to retrieve the value for following variables:
17970
17971 * FCGI_MAX_REQS The maximum number of concurrent requests this
17972 application will accept.
17973
William Lallemand93e548e2019-09-30 13:54:02 +020017974 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
17975 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020017976
17977 Some FastCGI applications does not support this feature. Some others close
17978 the connexion immediately after sending their response. So, by default, this
17979 option is disabled.
17980
17981 Note that the maximum number of concurrent requests accepted by a FastCGI
17982 application is a connection variable. It only limits the number of streams
17983 per connection. If the global load must be limited on the application, the
17984 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
17985 an application does not support connection multiplexing, the maximum number
17986 of concurrent requests is automatically set to 1.
17987
17988option keep-conn
17989no option keep-conn
17990 Instruct the FastCGI application to keep the connection open or not after
17991 sending a response.
17992
17993 If disabled, the FastCGI application closes the connection after responding
17994 to this request. By default, this option is enabled.
17995
17996option max-reqs <reqs>
17997 Define the maximum number of concurrent requests this application will
17998 accept.
17999
18000 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
18001 during connection establishment. Furthermore, if the application does not
18002 support connection multiplexing, this option will be ignored. By default set
18003 to 1.
18004
18005option mpxs-conns
18006no option mpxs-conns
18007 Enable or disable the support of connection multiplexing.
18008
18009 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
18010 during connection establishment. It is disabled by default.
18011
18012set-param <name> <fmt> [ { if | unless } <condition> ]
18013 Set a FastCGI parameter that should be passed to this application. Its
18014 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
18015 "Custom Log format"). It may optionally be followed by an ACL-based
18016 condition, in which case it will only be evaluated if the condition is true.
18017
18018 With this directive, it is possible to overwrite the value of default FastCGI
18019 parameters. If the value is evaluated to an empty string, the rule is
18020 ignored. These directives are evaluated in their declaration order.
18021
18022 Example :
18023 # PHP only, required if PHP was built with --enable-force-cgi-redirect
18024 set-param REDIRECT_STATUS 200
18025
18026 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
18027
18028
1802910.1.2. Proxy section
18030---------------------
18031
18032use-fcgi-app <name>
18033 Define the FastCGI application to use for the backend.
18034
18035 Arguments :
18036 <name> is the name of the FastCGI application to use.
18037
18038 This keyword is only available for HTTP proxies with the backend capability
18039 and with at least one FastCGI server. However, FastCGI servers can be mixed
18040 with HTTP servers. But except there is a good reason to do so, it is not
18041 recommended (see section 10.3 about the limitations for details). Only one
18042 application may be defined at a time per backend.
18043
18044 Note that, once a FastCGI application is referenced for a backend, depending
18045 on the configuration some processing may be done even if the request is not
18046 sent to a FastCGI server. Rules to set parameters or pass headers to an
18047 application are evaluated.
18048
18049
1805010.1.3. Example
18051---------------
18052
18053 frontend front-http
18054 mode http
18055 bind *:80
18056 bind *:
18057
18058 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
18059 default_backend back-static
18060
18061 backend back-static
18062 mode http
18063 server www A.B.C.D:80
18064
18065 backend back-dynamic
18066 mode http
18067 use-fcgi-app php-fpm
18068 server php-fpm A.B.C.D:9000 proto fcgi
18069
18070 fcgi-app php-fpm
18071 log-stderr global
18072 option keep-conn
18073
18074 docroot /var/www/my-app
18075 index index.php
18076 path-info ^(/.+\.php)(/.*)?$
18077
18078
1807910.2. Default parameters
18080------------------------
18081
18082A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
18083the CGI/1.1 specification (RFC3875), several variables must be passed to the
18084scipt. So HAProxy set them and some others commonly used by FastCGI
18085applications. All these variables may be overwritten, with caution though.
18086
18087 +-------------------+-----------------------------------------------------+
18088 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
18089 | | to authenticate the user. Concretely, only the |
18090 | | BASIC authentication mechanism is supported. |
18091 | | |
18092 +-------------------+-----------------------------------------------------+
18093 | CONTENT_LENGTH | Contains the size of the message-body attached to |
18094 | | the request. It means only requests with a known |
18095 | | size are considered as valid and sent to the |
18096 | | application. |
18097 | | |
18098 +-------------------+-----------------------------------------------------+
18099 | CONTENT_TYPE | Contains the type of the message-body attached to |
18100 | | the request. It may not be set. |
18101 | | |
18102 +-------------------+-----------------------------------------------------+
18103 | DOCUMENT_ROOT | Contains the document root on the remote host under |
18104 | | which the script should be executed, as defined in |
18105 | | the application's configuration. |
18106 | | |
18107 +-------------------+-----------------------------------------------------+
18108 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
18109 | | to communicate with the FastCGI application. |
18110 | | Concretely, it is set to "CGI/1.1". |
18111 | | |
18112 +-------------------+-----------------------------------------------------+
18113 | PATH_INFO | Contains the portion of the URI path hierarchy |
18114 | | following the part that identifies the script |
18115 | | itself. To be set, the directive "path-info" must |
18116 | | be defined. |
18117 | | |
18118 +-------------------+-----------------------------------------------------+
18119 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
18120 | | It is the concatenation of DOCUMENT_ROOT and |
18121 | | PATH_INFO. If PATH_INFO is not set, this parameters |
18122 | | is not set too. |
18123 | | |
18124 +-------------------+-----------------------------------------------------+
18125 | QUERY_STRING | Contains the request's query string. It may not be |
18126 | | set. |
18127 | | |
18128 +-------------------+-----------------------------------------------------+
18129 | REMOTE_ADDR | Contains the network address of the client sending |
18130 | | the request. |
18131 | | |
18132 +-------------------+-----------------------------------------------------+
18133 | REMOTE_USER | Contains the user identification string supplied by |
18134 | | client as part of user authentication. |
18135 | | |
18136 +-------------------+-----------------------------------------------------+
18137 | REQUEST_METHOD | Contains the method which should be used by the |
18138 | | script to process the request. |
18139 | | |
18140 +-------------------+-----------------------------------------------------+
18141 | REQUEST_URI | Contains the request's URI. |
18142 | | |
18143 +-------------------+-----------------------------------------------------+
18144 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
18145 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
18146 | | |
18147 +-------------------+-----------------------------------------------------+
18148 | SCRIPT_NAME | Contains the name of the script. If the directive |
18149 | | "path-info" is defined, it is the first part of the |
18150 | | URI path hierarchy, ending with the script name. |
18151 | | Otherwise, it is the entire URI path. |
18152 | | |
18153 +-------------------+-----------------------------------------------------+
18154 | SERVER_NAME | Contains the name of the server host to which the |
18155 | | client request is directed. It is the value of the |
18156 | | header "Host", if defined. Otherwise, the |
18157 | | destination address of the connection on the client |
18158 | | side. |
18159 | | |
18160 +-------------------+-----------------------------------------------------+
18161 | SERVER_PORT | Contains the destination TCP port of the connection |
18162 | | on the client side, which is the port the client |
18163 | | connected to. |
18164 | | |
18165 +-------------------+-----------------------------------------------------+
18166 | SERVER_PROTOCOL | Contains the request's protocol. |
18167 | | |
18168 +-------------------+-----------------------------------------------------+
18169 | HTTPS | Set to a non-empty value ("on") if the script was |
18170 | | queried through the HTTPS protocol. |
18171 | | |
18172 +-------------------+-----------------------------------------------------+
18173
18174
1817510.3. Limitations
18176------------------
18177
18178The current implementation have some limitations. The first one is about the
18179way some request headers are hidden to the FastCGI applications. This happens
18180during the headers analysis, on the backend side, before the connection
18181establishment. At this stage, HAProxy know the backend is using a FastCGI
18182application but it don't know if the request will be routed to a FastCGI server
18183or not. But to hide request headers, it simply removes them from the HTX
18184message. So, if the request is finally routed to an HTTP server, it never see
18185these headers. For this reason, it is not recommended to mix FastCGI servers
18186and HTTP servers under the same backend.
18187
18188Similarly, the rules "set-param" and "pass-header" are evaluated during the
18189request headers analysis. So the evaluation is always performed, even if the
18190requests is finally forwarded to an HTTP server.
18191
18192About the rules "set-param", when a rule is applied, a pseudo header is added
18193into the HTX message. So, the same way than for HTTP header rewrites, it may
18194fail if the buffer is full. The rules "set-param" will compete with
18195"http-request" ones.
18196
18197Finally, all FastCGI params and HTTP headers are sent into a unique record
18198FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
18199processing error is returned. It means the record FCGI_PARAM, once encoded,
18200must not exceeds the size of a buffer. However, there is no reserve to respect
18201here.
William Lallemand86d0df02017-11-24 21:36:45 +010018202
Willy Tarreau0ba27502007-12-24 16:55:16 +010018203/*
18204 * Local variables:
18205 * fill-column: 79
18206 * End:
18207 */