blob: 4e18f0f6e5584fcfe1da59bb879dae434badc8d6 [file] [log] [blame]
Willy Tarreau6a06a402007-07-15 20:15:28 +02001 ----------------------
Willy Tarreau8317b282014-04-23 01:49:41 +02002 HAProxy
Willy Tarreau6a06a402007-07-15 20:15:28 +02003 Configuration Manual
4 ----------------------
Willy Tarreau9dc6b972019-06-16 21:49:47 +02005 version 2.1
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreaudb514072019-07-16 19:15:28 +02007 2019/07/16
Willy Tarreau6a06a402007-07-15 20:15:28 +02008
9
10This document covers the configuration language as implemented in the version
Davor Ocelice9ed2812017-12-25 17:49:28 +010011specified above. It does not provide any hints, examples, or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010012documentation, please refer to the Reference Manual or the Architecture Manual.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013The summary below is meant to help you find sections by name and navigate
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014through the document.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016Note to documentation contributors :
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017 This document is formatted with 80 columns per line, with even number of
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018 spaces for indentation and without tabs. Please follow these rules strictly
19 so that it remains easily printable everywhere. If a line needs to be
20 printed verbatim and does not fit, please end each line with a backslash
Willy Tarreau62a36c42010-08-17 15:53:10 +020021 ('\') and continue on next line, indented by two characters. It is also
Davor Ocelice9ed2812017-12-25 17:49:28 +010022 sometimes useful to prefix all output lines (logs, console outputs) with 3
23 closing angle brackets ('>>>') in order to emphasize the difference between
24 inputs and outputs when they may be ambiguous. If you add sections,
Willy Tarreau62a36c42010-08-17 15:53:10 +020025 please update the summary below for easier searching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026
27
28Summary
29-------
30
311. Quick reminder about HTTP
321.1. The HTTP transaction model
331.2. HTTP request
Davor Ocelice9ed2812017-12-25 17:49:28 +0100341.2.1. The request line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200351.2.2. The request headers
361.3. HTTP response
Davor Ocelice9ed2812017-12-25 17:49:28 +0100371.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381.3.2. The response headers
39
402. Configuring HAProxy
412.1. Configuration file format
William Lallemandf9873ba2015-05-05 17:37:14 +0200422.2. Quoting and escaping
William Lallemandb2f07452015-05-12 14:27:13 +0200432.3. Environment variables
442.4. Time format
452.5. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020046
473. Global parameters
483.1. Process management and security
493.2. Performance tuning
503.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100513.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200523.5. Peers
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200533.6. Mailers
William Lallemandc9515522019-06-12 16:32:11 +0200543.7. Programs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020055
564. Proxies
574.1. Proxy keywords matrix
584.2. Alphabetically sorted keywords reference
59
Davor Ocelice9ed2812017-12-25 17:49:28 +0100605. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200615.1. Bind options
625.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200635.3. Server DNS resolution
645.3.1. Global overview
655.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020066
Willy Tarreau74ca5042013-06-11 23:12:07 +0200677. Using ACLs and fetching samples
687.1. ACL basics
697.1.1. Matching booleans
707.1.2. Matching integers
717.1.3. Matching strings
727.1.4. Matching regular expressions (regexes)
737.1.5. Matching arbitrary data blocks
747.1.6. Matching IPv4 and IPv6 addresses
757.2. Using ACLs to form conditions
767.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200777.3.1. Converters
787.3.2. Fetching samples from internal states
797.3.3. Fetching samples at Layer 4
807.3.4. Fetching samples at Layer 5
817.3.5. Fetching samples from buffer contents (Layer 6)
827.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +0200837.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020084
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200856. Cache
866.1. Limitation
876.2. Setup
886.2.1. Cache section
896.2.2. Proxy section
90
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200918. Logging
928.1. Log levels
938.2. Log formats
948.2.1. Default log format
958.2.2. TCP log format
968.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +0100978.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +0100988.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200998.3. Advanced logging options
1008.3.1. Disabling logging of external tests
1018.3.2. Logging before waiting for the session to terminate
1028.3.3. Raising log level upon errors
1038.3.4. Disabling logging of successful connections
1048.4. Timing events
1058.5. Session state at disconnection
1068.6. Non-printable characters
1078.7. Capturing HTTP cookies
1088.8. Capturing HTTP headers
1098.9. Examples of logs
110
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001119. Supported filters
1129.1. Trace
1139.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001149.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001159.4. Cache
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200116
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117
1181. Quick reminder about HTTP
119----------------------------
120
Davor Ocelice9ed2812017-12-25 17:49:28 +0100121When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200122fully analyzed and indexed, thus it becomes possible to build matching criteria
123on almost anything found in the contents.
124
125However, it is important to understand how HTTP requests and responses are
126formed, and how HAProxy decomposes them. It will then become easier to write
127correct rules and to debug existing configurations.
128
129
1301.1. The HTTP transaction model
131-------------------------------
132
133The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100134to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100135from the client to the server, a request is sent by the client through the
136connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200137will involve a new connection :
138
139 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
140
141In this mode, called the "HTTP close" mode, there are as many connection
142establishments as there are HTTP transactions. Since the connection is closed
143by the server after the response, the client does not need to know the content
144length.
145
146Due to the transactional nature of the protocol, it was possible to improve it
147to avoid closing a connection between two subsequent transactions. In this mode
148however, it is mandatory that the server indicates the content length for each
149response so that the client does not wait indefinitely. For this, a special
150header is used: "Content-length". This mode is called the "keep-alive" mode :
151
152 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
153
154Its advantages are a reduced latency between transactions, and less processing
155power required on the server side. It is generally better than the close mode,
156but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200157a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158
Willy Tarreau95c4e142017-11-26 12:18:55 +0100159Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200160keep-alive, but the client does not wait for the first response to send the
161second request. This is useful for fetching large number of images composing a
162page :
163
164 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
165
166This can obviously have a tremendous benefit on performance because the network
167latency is eliminated between subsequent requests. Many HTTP agents do not
168correctly support pipelining since there is no way to associate a response with
169the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100170server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171
Willy Tarreau95c4e142017-11-26 12:18:55 +0100172The next improvement is the multiplexed mode, as implemented in HTTP/2. This
173time, each transaction is assigned a single stream identifier, and all streams
174are multiplexed over an existing connection. Many requests can be sent in
175parallel by the client, and responses can arrive in any order since they also
176carry the stream identifier.
177
Willy Tarreau70dffda2014-01-30 03:07:23 +0100178By default HAProxy operates in keep-alive mode with regards to persistent
179connections: for each connection it processes each request and response, and
180leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100181start of a new request. When it receives HTTP/2 connections from a client, it
182processes all the requests in parallel and leaves the connection idling,
183waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200184
Christopher Faulet315b39c2018-09-21 16:26:19 +0200185HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100186 - keep alive : all requests and responses are processed (default)
187 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100188 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100189 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200190 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100191
Davor Ocelice9ed2812017-12-25 17:49:28 +0100192For HTTP/2, the connection mode resembles more the "server close" mode : given
193the independence of all streams, there is currently no place to hook the idle
Willy Tarreau95c4e142017-11-26 12:18:55 +0100194server connection after a response, so it is closed after the response. HTTP/2
195is only supported for incoming connections, not on connections going to
196servers.
197
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198
1991.2. HTTP request
200-----------------
201
202First, let's consider this HTTP request :
203
204 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100205 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200206 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
207 2 Host: www.mydomain.com
208 3 User-agent: my small browser
209 4 Accept: image/jpeg, image/gif
210 5 Accept: image/png
211
212
2131.2.1. The Request line
214-----------------------
215
216Line 1 is the "request line". It is always composed of 3 fields :
217
218 - a METHOD : GET
219 - a URI : /serv/login.php?lang=en&profile=2
220 - a version tag : HTTP/1.1
221
222All of them are delimited by what the standard calls LWS (linear white spaces),
223which are commonly spaces, but can also be tabs or line feeds/carriage returns
224followed by spaces/tabs. The method itself cannot contain any colon (':') and
225is limited to alphabetic letters. All those various combinations make it
226desirable that HAProxy performs the splitting itself rather than leaving it to
227the user to write a complex or inaccurate regular expression.
228
229The URI itself can have several forms :
230
231 - A "relative URI" :
232
233 /serv/login.php?lang=en&profile=2
234
235 It is a complete URL without the host part. This is generally what is
236 received by servers, reverse proxies and transparent proxies.
237
238 - An "absolute URI", also called a "URL" :
239
240 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
241
242 It is composed of a "scheme" (the protocol name followed by '://'), a host
243 name or address, optionally a colon (':') followed by a port number, then
244 a relative URI beginning at the first slash ('/') after the address part.
245 This is generally what proxies receive, but a server supporting HTTP/1.1
246 must accept this form too.
247
248 - a star ('*') : this form is only accepted in association with the OPTIONS
249 method and is not relayable. It is used to inquiry a next hop's
250 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100251
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200252 - an address:port combination : 192.168.0.12:80
253 This is used with the CONNECT method, which is used to establish TCP
254 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
255 other protocols too.
256
257In a relative URI, two sub-parts are identified. The part before the question
258mark is called the "path". It is typically the relative path to static objects
259on the server. The part after the question mark is called the "query string".
260It is mostly used with GET requests sent to dynamic scripts and is very
261specific to the language, framework or application in use.
262
Willy Tarreau95c4e142017-11-26 12:18:55 +0100263HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100264assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100265However, haproxy natively processes HTTP/1.x requests and headers, so requests
266received over an HTTP/2 connection are transcoded to HTTP/1.1 before being
267processed. This explains why they still appear as "HTTP/1.1" in haproxy's logs
268as well as in server logs.
269
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200270
2711.2.2. The request headers
272--------------------------
273
274The headers start at the second line. They are composed of a name at the
275beginning of the line, immediately followed by a colon (':'). Traditionally,
276an LWS is added after the colon but that's not required. Then come the values.
277Multiple identical headers may be folded into one single line, delimiting the
278values with commas, provided that their order is respected. This is commonly
279encountered in the "Cookie:" field. A header may span over multiple lines if
280the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
281define a total of 3 values for the "Accept:" header.
282
Davor Ocelice9ed2812017-12-25 17:49:28 +0100283Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200284their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100285"Connection:" header). In HTTP/2, header names are always sent in lower case,
286as can be seen when running in debug mode.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200287
288The end of the headers is indicated by the first empty line. People often say
289that it's a double line feed, which is not exact, even if a double line feed
290is one valid form of empty line.
291
292Fortunately, HAProxy takes care of all these complex combinations when indexing
293headers, checking values and counting them, so there is no reason to worry
294about the way they could be written, but it is important not to accuse an
295application of being buggy if it does unusual, valid things.
296
297Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000298 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200299 in the middle of headers by LWS in order to join multi-line headers. This
300 is necessary for proper analysis and helps less capable HTTP parsers to work
301 correctly and not to be fooled by such complex constructs.
302
303
3041.3. HTTP response
305------------------
306
307An HTTP response looks very much like an HTTP request. Both are called HTTP
308messages. Let's consider this HTTP response :
309
310 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100311 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200312 1 HTTP/1.1 200 OK
313 2 Content-length: 350
314 3 Content-Type: text/html
315
Willy Tarreau816b9792009-09-15 21:25:21 +0200316As a special case, HTTP supports so called "Informational responses" as status
317codes 1xx. These messages are special in that they don't convey any part of the
318response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100319continue to post its request for instance. In the case of a status 100 response
320the requested information will be carried by the next non-100 response message
321following the informational one. This implies that multiple responses may be
322sent to a single request, and that this only works when keep-alive is enabled
323(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
324correctly forward and skip them, and only process the next non-100 response. As
325such, these messages are neither logged nor transformed, unless explicitly
326state otherwise. Status 101 messages indicate that the protocol is changing
327over the same connection and that haproxy must switch to tunnel mode, just as
328if a CONNECT had occurred. Then the Upgrade header would contain additional
329information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200330
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200331
Davor Ocelice9ed2812017-12-25 17:49:28 +01003321.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200333------------------------
334
335Line 1 is the "response line". It is always composed of 3 fields :
336
337 - a version tag : HTTP/1.1
338 - a status code : 200
339 - a reason : OK
340
341The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100342 - 1xx = informational message to be skipped (e.g. 100, 101)
343 - 2xx = OK, content is following (e.g. 200, 206)
344 - 3xx = OK, no content following (e.g. 302, 304)
345 - 4xx = error caused by the client (e.g. 401, 403, 404)
346 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200347
Lukas Tribus23953682017-04-28 13:24:30 +0000348Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100349"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200350found there, but it's a common practice to respect the well-established
351messages. It can be composed of one or multiple words, such as "OK", "Found",
352or "Authentication Required".
353
Davor Ocelice9ed2812017-12-25 17:49:28 +0100354HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200355
356 Code When / reason
357 200 access to stats page, and when replying to monitoring requests
358 301 when performing a redirection, depending on the configured code
359 302 when performing a redirection, depending on the configured code
360 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100361 307 when performing a redirection, depending on the configured code
362 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200363 400 for an invalid or too large request
364 401 when an authentication is required to perform the action (when
365 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200366 403 when a request is forbidden by a "http-request deny" rule
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200367 408 when the request timeout strikes before the request is complete
368 500 when haproxy encounters an unrecoverable internal error, such as a
369 memory allocation failure, which should never happen
370 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200371 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200372 503 when no server was available to handle the request, or in response to
373 monitoring requests which match the "monitor fail" condition
374 504 when the response timeout strikes before the server responds
375
376The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3774.2).
378
379
3801.3.2. The response headers
381---------------------------
382
383Response headers work exactly like request headers, and as such, HAProxy uses
384the same parsing function for both. Please refer to paragraph 1.2.2 for more
385details.
386
387
3882. Configuring HAProxy
389----------------------
390
3912.1. Configuration file format
392------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200393
394HAProxy's configuration process involves 3 major sources of parameters :
395
396 - the arguments from the command-line, which always take precedence
397 - the "global" section, which sets process-wide parameters
398 - the proxies sections which can take form of "defaults", "listen",
399 "frontend" and "backend".
400
Willy Tarreau0ba27502007-12-24 16:55:16 +0100401The configuration file syntax consists in lines beginning with a keyword
402referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200403delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100404
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200405
William Lallemandf9873ba2015-05-05 17:37:14 +02004062.2. Quoting and escaping
407-------------------------
408
409HAProxy's configuration introduces a quoting and escaping system similar to
410many programming languages. The configuration file supports 3 types: escaping
411with a backslash, weak quoting with double quotes, and strong quoting with
412single quotes.
413
414If spaces have to be entered in strings, then they must be escaped by preceding
415them by a backslash ('\') or by quoting them. Backslashes also have to be
416escaped by doubling or strong quoting them.
417
418Escaping is achieved by preceding a special character by a backslash ('\'):
419
420 \ to mark a space and differentiate it from a delimiter
421 \# to mark a hash and differentiate it from a comment
422 \\ to use a backslash
423 \' to use a single quote and differentiate it from strong quoting
424 \" to use a double quote and differentiate it from weak quoting
425
426Weak quoting is achieved by using double quotes (""). Weak quoting prevents
427the interpretation of:
428
429 space as a parameter separator
430 ' single quote as a strong quoting delimiter
431 # hash as a comment start
432
William Lallemandb2f07452015-05-12 14:27:13 +0200433Weak quoting permits the interpretation of variables, if you want to use a non
434-interpreted dollar within a double quoted string, you should escape it with a
435backslash ("\$"), it does not work outside weak quoting.
436
437Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200438quoting.
439
440Strong quoting is achieved by using single quotes (''). Inside single quotes,
441nothing is interpreted, it's the efficient way to quote regexes.
442
443Quoted and escaped strings are replaced in memory by their interpreted
444equivalent, it allows you to perform concatenation.
445
446 Example:
447 # those are equivalents:
448 log-format %{+Q}o\ %t\ %s\ %{-Q}r
449 log-format "%{+Q}o %t %s %{-Q}r"
450 log-format '%{+Q}o %t %s %{-Q}r'
451 log-format "%{+Q}o %t"' %s %{-Q}r'
452 log-format "%{+Q}o %t"' %s'\ %{-Q}r
453
454 # those are equivalents:
455 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
456 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
457 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
458 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
459
460
William Lallemandb2f07452015-05-12 14:27:13 +02004612.3. Environment variables
462--------------------------
463
464HAProxy's configuration supports environment variables. Those variables are
465interpreted only within double quotes. Variables are expanded during the
466configuration parsing. Variable names must be preceded by a dollar ("$") and
467optionally enclosed with braces ("{}") similarly to what is done in Bourne
468shell. Variable names can contain alphanumerical characters or the character
469underscore ("_") but should not start with a digit.
470
471 Example:
472
473 bind "fd@${FD_APP1}"
474
475 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
476
477 user "$HAPROXY_USER"
478
William Lallemand4d03e432019-06-14 15:35:37 +0200479Some variables are defined by HAProxy, they can be used in the configuration
480file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200481
William Lallemand4d03e432019-06-14 15:35:37 +0200482* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
483 name of the local peer. (See "-L" in the management guide.)
484
485* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
486 separated by semicolons. Can be useful in the case you specified a
487 directory.
488
489* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
490
John Roeslerfb2fce12019-07-10 15:45:51 -0500491* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200492 processes, separated by semicolons.
493
John Roeslerfb2fce12019-07-10 15:45:51 -0500494* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200495 CLI, separated by semicolons.
496
497See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200498
4992.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200500----------------
501
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100502Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100503values are generally expressed in milliseconds (unless explicitly stated
504otherwise) but may be expressed in any other unit by suffixing the unit to the
505numeric value. It is important to consider this because it will not be repeated
506for every keyword. Supported units are :
507
508 - us : microseconds. 1 microsecond = 1/1000000 second
509 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
510 - s : seconds. 1s = 1000ms
511 - m : minutes. 1m = 60s = 60000ms
512 - h : hours. 1h = 60m = 3600s = 3600000ms
513 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
514
515
Lukas Tribusaa83a312017-03-21 09:25:09 +00005162.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200517-------------
518
519 # Simple configuration for an HTTP proxy listening on port 80 on all
520 # interfaces and forwarding requests to a single backend "servers" with a
521 # single server "server1" listening on 127.0.0.1:8000
522 global
523 daemon
524 maxconn 256
525
526 defaults
527 mode http
528 timeout connect 5000ms
529 timeout client 50000ms
530 timeout server 50000ms
531
532 frontend http-in
533 bind *:80
534 default_backend servers
535
536 backend servers
537 server server1 127.0.0.1:8000 maxconn 32
538
539
540 # The same configuration defined with a single listen block. Shorter but
541 # less expressive, especially in HTTP mode.
542 global
543 daemon
544 maxconn 256
545
546 defaults
547 mode http
548 timeout connect 5000ms
549 timeout client 50000ms
550 timeout server 50000ms
551
552 listen http-in
553 bind *:80
554 server server1 127.0.0.1:8000 maxconn 32
555
556
557Assuming haproxy is in $PATH, test these configurations in a shell with:
558
Willy Tarreauccb289d2010-12-11 20:19:38 +0100559 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200560
561
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005623. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200563--------------------
564
565Parameters in the "global" section are process-wide and often OS-specific. They
566are generally set once for all and do not need being changed once correct. Some
567of them have command-line equivalents.
568
569The following keywords are supported in the "global" section :
570
571 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200572 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200573 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200574 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200575 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200576 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200577 - description
578 - deviceatlas-json-file
579 - deviceatlas-log-level
580 - deviceatlas-separator
581 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900582 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200583 - gid
584 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100585 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200586 - h1-case-adjust
587 - h1-case-adjust-file
Willy Tarreau6a06a402007-07-15 20:15:28 +0200588 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200589 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100590 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200591 - lua-load
William Lallemand27edc4b2019-05-07 17:49:33 +0200592 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200593 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200594 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200595 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200596 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100597 - presetenv
598 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200599 - uid
600 - ulimit-n
601 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200602 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100603 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200604 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200605 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200606 - ssl-default-bind-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200607 - ssl-default-bind-options
608 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200609 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200610 - ssl-default-server-options
611 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100612 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100613 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100614 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100615 - 51degrees-data-file
616 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200617 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200618 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200619 - wurfl-data-file
620 - wurfl-information-list
621 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200622 - wurfl-cache-size
Willy Tarreaud72758d2010-01-12 10:42:19 +0100623
Willy Tarreau6a06a402007-07-15 20:15:28 +0200624 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200625 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200626 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200627 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100628 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100629 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100630 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200631 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200632 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200633 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200634 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200635 - noepoll
636 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000637 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200638 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100639 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300640 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000641 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100642 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200643 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200644 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200645 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000646 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000647 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200648 - tune.buffers.limit
649 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200650 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200651 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100652 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200653 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200654 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200655 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100656 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200657 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200658 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100659 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100660 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100661 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100662 - tune.lua.session-timeout
663 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200664 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100665 - tune.maxaccept
666 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200667 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200668 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200669 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100670 - tune.rcvbuf.client
671 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100672 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200673 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100674 - tune.sndbuf.client
675 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100676 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100677 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200678 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100679 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200680 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200681 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100682 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200683 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100684 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200685 - tune.vars.reqres-max-size
686 - tune.vars.sess-max-size
687 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100688 - tune.zlib.memlevel
689 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100690
Willy Tarreau6a06a402007-07-15 20:15:28 +0200691 * Debugging
692 - debug
693 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200694
695
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006963.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200697------------------------------------
698
Emeric Brunc8e8d122012-10-02 18:42:10 +0200699ca-base <dir>
700 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200701 relative path is used with "ca-file" or "crl-file" directives. Absolute
702 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200703
Willy Tarreau6a06a402007-07-15 20:15:28 +0200704chroot <jail dir>
705 Changes current directory to <jail dir> and performs a chroot() there before
706 dropping privileges. This increases the security level in case an unknown
707 vulnerability would be exploited, since it would make it very hard for the
708 attacker to exploit the system. This only works when the process is started
709 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100710 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100711
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100712cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
713 On Linux 2.6 and above, it is possible to bind a process or a thread to a
714 specific CPU set. This means that the process or the thread will never run on
715 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
716 sets. The first argument is a process set, eventually followed by a thread
717 set. These sets have the format
718
719 all | odd | even | number[-[number]]
720
721 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100722 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100723 ignored. It is possible to specify a range with two such number delimited by
724 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100725 "all", only odd numbers using "odd" or even numbers using "even", just like
726 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100727 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 +0100728 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100729 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100730 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100731 specified. Each "cpu-map" directive will replace the previous ones when they
732 overlap. A thread will be bound on the intersection of its mapping and the
733 one of the process on which it is attached. If the intersection is null, no
734 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100735
Christopher Fauletff4121f2017-11-22 16:38:49 +0100736 Ranges can be partially defined. The higher bound can be omitted. In such
737 case, it is replaced by the corresponding maximum value, 32 or 64 depending
738 on the machine's word size.
739
Christopher Faulet26028f62017-11-22 15:01:51 +0100740 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100741 automatically bind a process or a thread to a CPU by incrementing
742 process/thread and CPU sets. To be valid, both sets must have the same
743 size. No matter the declaration order of the CPU sets, it will be bound from
744 the lowest to the highest bound. Having a process and a thread range with the
745 "auto:" prefix is not supported. Only one range is supported, the other one
746 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100747
748 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100749 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
750
751 cpu-map 1/all 0-3 # bind all threads of the first process on the
752 # first 4 CPUs
753
754 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
755 # or "cpu-map 1-32 0-31" depending on the machine's
756 # word size.
757
Christopher Faulet26028f62017-11-22 15:01:51 +0100758 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100759 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100760 cpu-map auto:1-4 0-3
761 cpu-map auto:1-4 0-1 2-3
762 cpu-map auto:1-4 3 2 1 0
763
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100764 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
765 # and so on.
766 cpu-map auto:1/1-4 0-3
767 cpu-map auto:1/1-4 0-1 2-3
768 cpu-map auto:1/1-4 3 2 1 0
769
Davor Ocelice9ed2812017-12-25 17:49:28 +0100770 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100771 cpu-map auto:all 0-63
772 cpu-map auto:even 0-31
773 cpu-map auto:odd 32-63
774
775 # invalid cpu-map because process and CPU sets have different sizes.
776 cpu-map auto:1-4 0 # invalid
777 cpu-map auto:1 0-3 # invalid
778
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100779 # invalid cpu-map because automatic binding is used with a process range
780 # and a thread range.
781 cpu-map auto:all/all 0 # invalid
782 cpu-map auto:all/1-4 0 # invalid
783 cpu-map auto:1-4/all 0 # invalid
784
Emeric Brunc8e8d122012-10-02 18:42:10 +0200785crt-base <dir>
786 Assigns a default directory to fetch SSL certificates from when a relative
787 path is used with "crtfile" directives. Absolute locations specified after
788 "crtfile" prevail and ignore "crt-base".
789
Willy Tarreau6a06a402007-07-15 20:15:28 +0200790daemon
791 Makes the process fork into background. This is the recommended mode of
792 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100793 disabled by the command line "-db" argument. This option is ignored in
794 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200795
David Carlier8167f302015-06-01 13:50:06 +0200796deviceatlas-json-file <path>
797 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100798 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200799
800deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100801 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200802 optional and set to 0 by default if not set.
803
804deviceatlas-separator <char>
805 Sets the character separator for the API properties results. This directive
806 is optional and set to | by default if not set.
807
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100808deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200809 Sets the client cookie's name used for the detection if the DeviceAtlas
810 Client-side component was used during the request. This directive is optional
811 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100812
Simon Horman98637e52014-06-20 12:30:16 +0900813external-check
814 Allows the use of an external agent to perform health checks.
815 This is disabled by default as a security precaution.
816 See "option external-check".
817
Willy Tarreau6a06a402007-07-15 20:15:28 +0200818gid <number>
819 Changes the process' group ID to <number>. It is recommended that the group
820 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
821 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100822 Note that if haproxy is started from a user having supplementary groups, it
823 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200824 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100825
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100826hard-stop-after <time>
827 Defines the maximum time allowed to perform a clean soft-stop.
828
829 Arguments :
830 <time> is the maximum time (by default in milliseconds) for which the
831 instance will remain alive when a soft-stop is received via the
832 SIGUSR1 signal.
833
834 This may be used to ensure that the instance will quit even if connections
835 remain opened during a soft-stop (for example with long timeouts for a proxy
836 in tcp mode). It applies both in TCP and HTTP mode.
837
838 Example:
839 global
840 hard-stop-after 30s
841
Christopher Faulet98fbe952019-07-22 16:18:24 +0200842h1-case-adjust <from> <to>
843 Defines the case adjustment to apply, when enabled, to the header name
844 <from>, to change it to <to> before sending it to HTTP/1 clients or
845 servers. <from> must be in lower case, and <from> and <to> must not differ
846 except for their case. It may be repeated if several header names need to be
847 ajusted. Duplicate entries are not allowed. If a lot of header names have to
848 be adjusted, it might be more convenient to use "h1-case-adjust-file".
849 Please note that no transformation will be applied unless "option
850 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
851 specified in a proxy.
852
853 There is no standard case for header names because, as stated in RFC7230,
854 they are case-insensitive. So applications must handle them in a case-
855 insensitive manner. But some bogus applications violate the standards and
856 erroneously rely on the cases most commonly used by browsers. This problem
857 becomes critical with HTTP/2 because all header names must be exchanged in
858 lower case, and HAProxy follows the same convention. All header names are
859 sent in lower case to clients and servers, regardless of the HTTP version.
860
861 Applications which fail to properly process requests or responses may require
862 to temporarily use such workarounds to adjust header names sent to them for
863 the time it takes the application to be fixed. Please note that an
864 application which requires such workarounds might be vulnerable to content
865 smuggling attacks and must absolutely be fixed.
866
867 Example:
868 global
869 h1-case-adjust content-length Content-Length
870
871 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
872 "option h1-case-adjust-bogus-server".
873
874h1-case-adjust-file <hdrs-file>
875 Defines a file containing a list of key/value pairs used to adjust the case
876 of some header names before sending them to HTTP/1 clients or servers. The
877 file <hdrs-file> must contain 2 header names per line. The first one must be
878 in lower case and both must not differ except for their case. Lines which
879 start with '#' are ignored, just like empty lines. Leading and trailing tabs
880 and spaces are stripped. Duplicate entries are not allowed. Please note that
881 no transformation will be applied unless "option h1-case-adjust-bogus-client"
882 or "option h1-case-adjust-bogus-server" is specified in a proxy.
883
884 If this directive is repeated, only the last one will be processed. It is an
885 alternative to the directive "h1-case-adjust" if a lot of header names need
886 to be adjusted. Please read the risks associated with using this.
887
888 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
889 "option h1-case-adjust-bogus-server".
890
Willy Tarreau6a06a402007-07-15 20:15:28 +0200891group <group name>
892 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
893 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100894
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200895log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
896 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100897 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100898 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100899 configured with "log global".
900
901 <address> can be one of:
902
Willy Tarreau2769aa02007-12-27 18:26:09 +0100903 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100904 no port is specified, 514 is used by default (the standard syslog
905 port).
906
David du Colombier24bb5f52011-03-17 10:40:23 +0100907 - An IPv6 address followed by a colon and optionally a UDP port. If
908 no port is specified, 514 is used by default (the standard syslog
909 port).
910
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100911 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100912 considerations for chroot (be sure the path is accessible inside
913 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100914 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100915
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100916 - A file descriptor number in the form "fd@<number>", which may point
917 to a pipe, terminal, or socket. In this case unbuffered logs are used
918 and one writev() call per log is performed. This is a bit expensive
919 but acceptable for most workloads. Messages sent this way will not be
920 truncated but may be dropped, in which case the DroppedLogs counter
921 will be incremented. The writev() call is atomic even on pipes for
922 messages up to PIPE_BUF size, which POSIX recommends to be at least
923 512 and which is 4096 bytes on most modern operating systems. Any
924 larger message may be interleaved with messages from other processes.
925 Exceptionally for debugging purposes the file descriptor may also be
926 directed to a file, but doing so will significantly slow haproxy down
927 as non-blocking calls will be ignored. Also there will be no way to
928 purge nor rotate this file without restarting the process. Note that
929 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100930 for use with a TCP syslog server. See also the "short" and "raw"
931 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100932
933 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
934 "fd@2", see above.
935
William Lallemandb2f07452015-05-12 14:27:13 +0200936 You may want to reference some environment variables in the address
937 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100938
Willy Tarreau18324f52014-06-27 18:10:07 +0200939 <length> is an optional maximum line length. Log lines larger than this value
940 will be truncated before being sent. The reason is that syslog
941 servers act differently on log line length. All servers support the
942 default value of 1024, but some servers simply drop larger lines
943 while others do log them. If a server supports long lines, it may
944 make sense to set this value here in order to avoid truncating long
945 lines. Similarly, if a server drops long lines, it is preferable to
946 truncate them before sending them. Accepted values are 80 to 65535
947 inclusive. The default value of 1024 is generally fine for all
948 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100949 JSON-formatted logs may require larger values. You may also need to
950 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200951
Dragan Dosen7ad31542015-09-28 17:16:47 +0200952 <format> is the log format used when generating syslog messages. It may be
953 one of the following :
954
955 rfc3164 The RFC3164 syslog message format. This is the default.
956 (https://tools.ietf.org/html/rfc3164)
957
958 rfc5424 The RFC5424 syslog message format.
959 (https://tools.ietf.org/html/rfc5424)
960
Willy Tarreaue8746a02018-11-12 08:45:00 +0100961 short A message containing only a level between angle brackets such as
962 '<3>', followed by the text. The PID, date, time, process name
963 and system name are omitted. This is designed to be used with a
964 local log server. This format is compatible with what the systemd
965 logger consumes.
966
Willy Tarreauc1b06452018-11-12 11:57:56 +0100967 raw A message containing only the text. The level, PID, date, time,
968 process name and system name are omitted. This is designed to be
969 used in containers or during development, where the severity only
970 depends on the file descriptor used (stdout/stderr).
971
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200972 <ranges> A list of comma-separated ranges to identify the logs to sample.
973 This is used to balance the load of the logs to send to the log
974 server. The limits of the ranges cannot be null. They are numbered
975 from 1. The size or period (in number of logs) of the sample must be
976 set with <sample_size> parameter.
977
978 <sample_size>
979 The size of the sample in number of logs to consider when balancing
980 their logging loads. It is used to balance the load of the logs to
981 send to the syslog server. This size must be greater or equal to the
982 maximum of the high limits of the ranges.
983 (see also <ranges> parameter).
984
Robert Tsai81ae1952007-12-05 10:47:29 +0100985 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200986
Willy Tarreaue8746a02018-11-12 08:45:00 +0100987 kern user mail daemon auth syslog lpr news
988 uucp cron auth2 ftp ntp audit alert cron2
989 local0 local1 local2 local3 local4 local5 local6 local7
990
Willy Tarreauc1b06452018-11-12 11:57:56 +0100991 Note that the facility is ignored for the "short" and "raw"
992 formats, but still required as a positional field. It is
993 recommended to use "daemon" in this case to make it clear that
994 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200995
996 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200997 all messages are sent. If a maximum level is specified, only messages with a
998 severity at least as important as this level will be sent. An optional minimum
999 level can be specified. If it is set, logs emitted with a more severe level
1000 than this one will be capped to this level. This is used to avoid sending
1001 "emerg" messages on all terminals on some default syslog configurations.
1002 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001003
Cyril Bontédc4d9032012-04-08 21:57:39 +02001004 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001005
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001006log-send-hostname [<string>]
1007 Sets the hostname field in the syslog header. If optional "string" parameter
1008 is set the header is set to the string contents, otherwise uses the hostname
1009 of the system. Generally used if one is not relaying logs through an
1010 intermediate syslog server or for simply customizing the hostname printed in
1011 the logs.
1012
Kevinm48936af2010-12-22 16:08:21 +00001013log-tag <string>
1014 Sets the tag field in the syslog header to this string. It defaults to the
1015 program name as launched from the command line, which usually is "haproxy".
1016 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001017 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001018
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001019lua-load <file>
1020 This global directive loads and executes a Lua file. This directive can be
1021 used multiple times.
1022
William Lallemand4cfede82017-11-24 22:02:34 +01001023master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001024 Master-worker mode. It is equivalent to the command line "-W" argument.
1025 This mode will launch a "master" which will monitor the "workers". Using
1026 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001027 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001028 or daemon mode. It is recommended to use this mode with multiprocess and
1029 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001030 By default, if a worker exits with a bad return code, in the case of a
1031 segfault for example, all workers will be killed, and the master will leave.
1032 It is convenient to combine this behavior with Restart=on-failure in a
1033 systemd unit file in order to relaunch the whole process. If you don't want
1034 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001035
William Lallemand4cfede82017-11-24 22:02:34 +01001036 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001037
William Lallemand27edc4b2019-05-07 17:49:33 +02001038mworker-max-reloads <number>
1039 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001040 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001041 number of reloads is greater than this number, the worker will receive a
1042 SIGTERM. This option helps to keep under control the number of workers.
1043 See also "show proc" in the Management Guide.
1044
Willy Tarreau6a06a402007-07-15 20:15:28 +02001045nbproc <number>
1046 Creates <number> processes when going daemon. This requires the "daemon"
1047 mode. By default, only one process is created, which is the recommended mode
1048 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001049 process, it may be needed to fork multiple daemons. When set to a value
1050 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +01001051 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
1052 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001053
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001054nbthread <number>
1055 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001056 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1057 "nbproc" historically used to be the only way to use multiple processors, it
1058 also involved a number of shortcomings related to the lack of synchronization
1059 between processes (health-checks, peers, stick-tables, stats, ...) which do
1060 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001061 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1062 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1063 when nbproc is not used, the default "nbthread" value is automatically set to
1064 the number of CPUs the process is bound to upon startup. This means that the
1065 thread count can easily be adjusted from the calling process using commands
1066 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1067 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001068
Willy Tarreau6a06a402007-07-15 20:15:28 +02001069pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001070 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +02001071 the "-p" command line argument. The file must be accessible to the user
1072 starting the process. See also "daemon".
1073
Willy Tarreau1d549722016-02-16 12:41:57 +01001074presetenv <name> <value>
1075 Sets environment variable <name> to value <value>. If the variable exists, it
1076 is NOT overwritten. The changes immediately take effect so that the next line
1077 in the configuration file sees the new value. See also "setenv", "resetenv",
1078 and "unsetenv".
1079
1080resetenv [<name> ...]
1081 Removes all environment variables except the ones specified in argument. It
1082 allows to use a clean controlled environment before setting new values with
1083 setenv or unsetenv. Please note that some internal functions may make use of
1084 some environment variables, such as time manipulation functions, but also
1085 OpenSSL or even external checks. This must be used with extreme care and only
1086 after complete validation. The changes immediately take effect so that the
1087 next line in the configuration file sees the new environment. See also
1088 "setenv", "presetenv", and "unsetenv".
1089
Christopher Fauletff4121f2017-11-22 16:38:49 +01001090stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001091 Limits the stats socket to a certain set of processes numbers. By default the
1092 stats socket is bound to all processes, causing a warning to be emitted when
1093 nbproc is greater than 1 because there is no way to select the target process
1094 when connecting. However, by using this setting, it becomes possible to pin
1095 the stats socket to a specific set of processes, typically the first one. The
1096 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001097 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001098 word size (32 or 64). Ranges can be partially defined. The higher bound can
1099 be omitted. In such case, it is replaced by the corresponding maximum
1100 value. A better option consists in using the "process" setting of the "stats
1101 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001102
Baptiste Assmann5626f482015-08-23 10:00:10 +02001103server-state-base <directory>
1104 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001105 file names which do not start with a '/'. See also "server-state-file",
1106 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001107
1108server-state-file <file>
1109 Specifies the path to the file containing state of servers. If the path starts
1110 with a slash ('/'), it is considered absolute, otherwise it is considered
1111 relative to the directory specified using "server-state-base" (if set) or to
1112 the current directory. Before reloading HAProxy, it is possible to save the
1113 servers' current state using the stats command "show servers state". The
1114 output of this command must be written in the file pointed by <file>. When
1115 starting up, before handling traffic, HAProxy will read, load and apply state
1116 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001117 configuration. See also "server-state-base" and "show servers state",
1118 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001119
Willy Tarreau1d549722016-02-16 12:41:57 +01001120setenv <name> <value>
1121 Sets environment variable <name> to value <value>. If the variable exists, it
1122 is overwritten. The changes immediately take effect so that the next line in
1123 the configuration file sees the new value. See also "presetenv", "resetenv",
1124 and "unsetenv".
1125
Willy Tarreau636848a2019-04-15 19:38:50 +02001126set-dumpable
1127 This option is better left disabled by default and enabled only upon a
1128 developer's request. It has no impact on performance nor stability but will
1129 try hard to re-enable core dumps that were possibly disabled by file size
1130 limitations (ulimit -f), core size limitations (ulimit -c), or "dumpability"
1131 of a process after changing its UID/GID (such as /proc/sys/fs/suid_dumpable
1132 on Linux). Core dumps might still be limited by the current directory's
1133 permissions (check what directory the file is started from), the chroot
1134 directory's permission (it may be needed to temporarily disable the chroot
1135 directive or to move it to a dedicated writable location), or any other
1136 system-specific constraint. For example, some Linux flavours are notorious
1137 for replacing the default core file with a path to an executable not even
1138 installed on the system (check /proc/sys/kernel/core_pattern). Often, simply
1139 writing "core", "core.%p" or "/var/log/core/core.%p" addresses the issue.
1140 When trying to enable this option waiting for a rare issue to re-appear, it's
1141 often a good idea to first try to obtain such a dump by issuing, for example,
1142 "kill -11" to the haproxy process and verify that it leaves a core where
1143 expected when dying.
1144
Willy Tarreau610f04b2014-02-13 11:36:41 +01001145ssl-default-bind-ciphers <ciphers>
1146 This setting is only available when support for OpenSSL was built in. It sets
1147 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001148 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001149 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001150 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1151 information and recommendations see e.g.
1152 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1153 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1154 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1155 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001156
1157ssl-default-bind-ciphersuites <ciphersuites>
1158 This setting is only available when support for OpenSSL was built in and
1159 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1160 describing the list of cipher algorithms ("cipher suite") that are negotiated
1161 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1162 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001163 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1164 cipher configuration for TLSv1.2 and earlier, please check the
1165 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001166 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001167
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001168ssl-default-bind-options [<option>]...
1169 This setting is only available when support for OpenSSL was built in. It sets
1170 default ssl-options to force on all "bind" lines. Please check the "bind"
1171 keyword to see available options.
1172
1173 Example:
1174 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001175 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001176
Willy Tarreau610f04b2014-02-13 11:36:41 +01001177ssl-default-server-ciphers <ciphers>
1178 This setting is only available when support for OpenSSL was built in. It
1179 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001180 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001181 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001182 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1183 information and recommendations see e.g.
1184 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1185 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1186 For TLSv1.3 cipher configuration, please check the
1187 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1188 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001189
1190ssl-default-server-ciphersuites <ciphersuites>
1191 This setting is only available when support for OpenSSL was built in and
1192 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1193 string describing the list of cipher algorithms that are negotiated during
1194 the TLSv1.3 handshake with the server, for all "server" lines which do not
1195 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001196 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1197 cipher configuration for TLSv1.2 and earlier, please check the
1198 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1199 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001200
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001201ssl-default-server-options [<option>]...
1202 This setting is only available when support for OpenSSL was built in. It sets
1203 default ssl-options to force on all "server" lines. Please check the "server"
1204 keyword to see available options.
1205
Remi Gacogne47783ef2015-05-29 15:53:22 +02001206ssl-dh-param-file <file>
1207 This setting is only available when support for OpenSSL was built in. It sets
1208 the default DH parameters that are used during the SSL/TLS handshake when
1209 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001210 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001211 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001212 are not specified either by using ssl-dh-param-file or by setting them
1213 directly in the certificate file, pre-generated DH parameters of the size
1214 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1215 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001216 Custom DH parameters may be generated by using the OpenSSL command
1217 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1218 parameters should not be considered secure anymore.
1219
Emeric Brun850efd52014-01-29 12:24:34 +01001220ssl-server-verify [none|required]
1221 The default behavior for SSL verify on servers side. If specified to 'none',
1222 servers certificates are not verified. The default is 'required' except if
1223 forced using cmdline option '-dV'.
1224
Willy Tarreauabb175f2012-09-24 12:43:26 +02001225stats socket [<address:port>|<path>] [param*]
1226 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1227 Connections to this socket will return various statistics outputs and even
1228 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001229 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001230 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001231
Willy Tarreauabb175f2012-09-24 12:43:26 +02001232 All parameters supported by "bind" lines are supported, for instance to
1233 restrict access to some users or their access rights. Please consult
1234 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001235
1236stats timeout <timeout, in milliseconds>
1237 The default timeout on the stats socket is set to 10 seconds. It is possible
1238 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001239 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001240
1241stats maxconn <connections>
1242 By default, the stats socket is limited to 10 concurrent connections. It is
1243 possible to change this value with "stats maxconn".
1244
Willy Tarreau6a06a402007-07-15 20:15:28 +02001245uid <number>
1246 Changes the process' user ID to <number>. It is recommended that the user ID
1247 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1248 be started with superuser privileges in order to be able to switch to another
1249 one. See also "gid" and "user".
1250
1251ulimit-n <number>
1252 Sets the maximum number of per-process file-descriptors to <number>. By
1253 default, it is automatically computed, so it is recommended not to use this
1254 option.
1255
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001256unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1257 [ group <group> ] [ gid <gid> ]
1258
1259 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1260 This is mainly used to simplify declaration of those UNIX sockets and reduce
1261 the risk of errors, since those settings are most commonly required but are
1262 also process-specific. The <prefix> setting can be used to force all socket
1263 path to be relative to that directory. This might be needed to access another
1264 component's chroot. Note that those paths are resolved before haproxy chroots
1265 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1266 all have the same meaning as their homonyms used by the "bind" statement. If
1267 both are specified, the "bind" statement has priority, meaning that the
1268 "unix-bind" settings may be seen as process-wide default settings.
1269
Willy Tarreau1d549722016-02-16 12:41:57 +01001270unsetenv [<name> ...]
1271 Removes environment variables specified in arguments. This can be useful to
1272 hide some sensitive information that are occasionally inherited from the
1273 user's environment during some operations. Variables which did not exist are
1274 silently ignored so that after the operation, it is certain that none of
1275 these variables remain. The changes immediately take effect so that the next
1276 line in the configuration file will not see these variables. See also
1277 "setenv", "presetenv", and "resetenv".
1278
Willy Tarreau6a06a402007-07-15 20:15:28 +02001279user <user name>
1280 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1281 See also "uid" and "group".
1282
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001283node <name>
1284 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1285
1286 This statement is useful in HA configurations where two or more processes or
1287 servers share the same IP address. By setting a different node-name on all
1288 nodes, it becomes easy to immediately spot what server is handling the
1289 traffic.
1290
1291description <text>
1292 Add a text that describes the instance.
1293
1294 Please note that it is required to escape certain characters (# for example)
1295 and this text is inserted into a html page so you should avoid using
1296 "<" and ">" characters.
1297
Thomas Holmesdb04f192015-05-18 13:21:39 +0100129851degrees-data-file <file path>
1299 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001300 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001301
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001302 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001303 compiled with USE_51DEGREES.
1304
Ben Shillitof25e8e52016-12-02 14:25:37 +0000130551degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001306 A list of 51Degrees property names to be load from the dataset. A full list
1307 of names is available on the 51Degrees website:
1308 https://51degrees.com/resources/property-dictionary
1309
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001310 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001311 compiled with USE_51DEGREES.
1312
Dragan Dosen93b38d92015-06-29 16:43:25 +0200131351degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001314 A char that will be appended to every property value in a response header
1315 containing 51Degrees results. If not set that will be set as ','.
1316
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001317 Please note that this option is only available when haproxy has been
1318 compiled with USE_51DEGREES.
1319
132051degrees-cache-size <number>
1321 Sets the size of the 51Degrees converter cache to <number> entries. This
1322 is an LRU cache which reminds previous device detections and their results.
1323 By default, this cache is disabled.
1324
1325 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001326 compiled with USE_51DEGREES.
1327
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001328wurfl-data-file <file path>
1329 The path of the WURFL data file to provide device detection services. The
1330 file should be accessible by HAProxy with relevant permissions.
1331
1332 Please note that this option is only available when haproxy has been compiled
1333 with USE_WURFL=1.
1334
1335wurfl-information-list [<capability>]*
1336 A space-delimited list of WURFL capabilities, virtual capabilities, property
1337 names we plan to use in injected headers. A full list of capability and
1338 virtual capability names is available on the Scientiamobile website :
1339
1340 https://www.scientiamobile.com/wurflCapability
1341
1342 Valid WURFL properties are:
1343 - wurfl_id Contains the device ID of the matched device.
1344
1345 - wurfl_root_id Contains the device root ID of the matched
1346 device.
1347
1348 - wurfl_isdevroot Tells if the matched device is a root device.
1349 Possible values are "TRUE" or "FALSE".
1350
1351 - wurfl_useragent The original useragent coming with this
1352 particular web request.
1353
1354 - wurfl_api_version Contains a string representing the currently
1355 used Libwurfl API version.
1356
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001357 - wurfl_info A string containing information on the parsed
1358 wurfl.xml and its full path.
1359
1360 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1361 WURFL has been loaded successfully.
1362
1363 - wurfl_normalized_useragent The normalized useragent.
1364
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001365 Please note that this option is only available when haproxy has been compiled
1366 with USE_WURFL=1.
1367
1368wurfl-information-list-separator <char>
1369 A char that will be used to separate values in a response header containing
1370 WURFL results. If not set that a comma (',') will be used by default.
1371
1372 Please note that this option is only available when haproxy has been compiled
1373 with USE_WURFL=1.
1374
1375wurfl-patch-file [<file path>]
1376 A list of WURFL patch file paths. Note that patches are loaded during startup
1377 thus before the chroot.
1378
1379 Please note that this option is only available when haproxy has been compiled
1380 with USE_WURFL=1.
1381
paulborilebad132c2019-04-18 11:57:04 +02001382wurfl-cache-size <size>
1383 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1384 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001385 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001386 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001387
1388 Please note that this option is only available when haproxy has been compiled
1389 with USE_WURFL=1.
1390
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013913.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001392-----------------------
1393
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001394busy-polling
1395 In some situations, especially when dealing with low latency on processors
1396 supporting a variable frequency or when running inside virtual machines, each
1397 time the process waits for an I/O using the poller, the processor goes back
1398 to sleep or is offered to another VM for a long time, and it causes
1399 excessively high latencies. This option provides a solution preventing the
1400 processor from sleeping by always using a null timeout on the pollers. This
1401 results in a significant latency reduction (30 to 100 microseconds observed)
1402 at the expense of a risk to overheat the processor. It may even be used with
1403 threads, in which case improperly bound threads may heavily conflict,
1404 resulting in a worse performance and high values for the CPU stolen fields
1405 in "show info" output, indicating which threads are misconfigured. It is
1406 important not to let the process run on the same processor as the network
1407 interrupts when this option is used. It is also better to avoid using it on
1408 multiple CPU threads sharing the same core. This option is disabled by
1409 default. If it has been enabled, it may still be forcibly disabled by
1410 prefixing it with the "no" keyword. It is ignored by the "select" and
1411 "poll" pollers.
1412
Willy Tarreau1746eec2014-04-25 10:46:47 +02001413max-spread-checks <delay in milliseconds>
1414 By default, haproxy tries to spread the start of health checks across the
1415 smallest health check interval of all the servers in a farm. The principle is
1416 to avoid hammering services running on the same server. But when using large
1417 check intervals (10 seconds or more), the last servers in the farm take some
1418 time before starting to be tested, which can be a problem. This parameter is
1419 used to enforce an upper bound on delay between the first and the last check,
1420 even if the servers' check intervals are larger. When servers run with
1421 shorter intervals, their intervals will be respected though.
1422
Willy Tarreau6a06a402007-07-15 20:15:28 +02001423maxconn <number>
1424 Sets the maximum per-process number of concurrent connections to <number>. It
1425 is equivalent to the command-line argument "-n". Proxies will stop accepting
1426 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001427 automatically adjusted according to this value. See also "ulimit-n". Note:
1428 the "select" poller cannot reliably use more than 1024 file descriptors on
1429 some platforms. If your platform only supports select and reports "select
1430 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001431 below 500 in general). If this value is not set, it will automatically be
1432 calculated based on the current file descriptors limit reported by the
1433 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1434 is enforced, based on the buffer size, memory allocated to compression, SSL
1435 cache size, and use or not of SSL and the associated maxsslconn (which can
1436 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001437
Willy Tarreau81c25d02011-09-07 15:17:21 +02001438maxconnrate <number>
1439 Sets the maximum per-process number of connections per second to <number>.
1440 Proxies will stop accepting connections when this limit is reached. It can be
1441 used to limit the global capacity regardless of each frontend capacity. It is
1442 important to note that this can only be used as a service protection measure,
1443 as there will not necessarily be a fair share between frontends when the
1444 limit is reached, so it's a good idea to also limit each frontend to some
1445 value close to its expected share. Also, lowering tune.maxaccept can improve
1446 fairness.
1447
William Lallemandd85f9172012-11-09 17:05:39 +01001448maxcomprate <number>
1449 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001450 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001451 level will be decreased during the session. If the maximum is reached at the
1452 beginning of a session, the session will not compress at all. If the maximum
1453 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001454 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001455 default value.
1456
William Lallemand072a2bf2012-11-20 17:01:01 +01001457maxcompcpuusage <number>
1458 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1459 for new requests or decreasing the compression level of current requests.
1460 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1461 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1462 case of multiple processes (nbproc > 1), each process manages its individual
1463 usage. A value of 100 disable the limit. The default value is 100. Setting
1464 a lower value will prevent the compression work from slowing the whole
1465 process down and from introducing high latencies.
1466
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001467maxpipes <number>
1468 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1469 are only used by kernel-based tcp splicing. Since a pipe contains two file
1470 descriptors, the "ulimit-n" value will be increased accordingly. The default
1471 value is maxconn/4, which seems to be more than enough for most heavy usages.
1472 The splice code dynamically allocates and releases pipes, and can fall back
1473 to standard copy, so setting this value too low may only impact performance.
1474
Willy Tarreau93e7c002013-10-07 18:51:07 +02001475maxsessrate <number>
1476 Sets the maximum per-process number of sessions per second to <number>.
1477 Proxies will stop accepting connections when this limit is reached. It can be
1478 used to limit the global capacity regardless of each frontend capacity. It is
1479 important to note that this can only be used as a service protection measure,
1480 as there will not necessarily be a fair share between frontends when the
1481 limit is reached, so it's a good idea to also limit each frontend to some
1482 value close to its expected share. Also, lowering tune.maxaccept can improve
1483 fairness.
1484
Willy Tarreau403edff2012-09-06 11:58:37 +02001485maxsslconn <number>
1486 Sets the maximum per-process number of concurrent SSL connections to
1487 <number>. By default there is no SSL-specific limit, which means that the
1488 global maxconn setting will apply to all connections. Setting this limit
1489 avoids having openssl use too much memory and crash when malloc returns NULL
1490 (since it unfortunately does not reliably check for such conditions). Note
1491 that the limit applies both to incoming and outgoing connections, so one
1492 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001493 If this value is not set, but a memory limit is enforced, this value will be
1494 automatically computed based on the memory limit, maxconn, the buffer size,
1495 memory allocated to compression, SSL cache size, and use of SSL in either
1496 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1497 when there is a memory limit, haproxy will automatically adjust these values
1498 so that 100% of the connections can be made over SSL with no risk, and will
1499 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001500
Willy Tarreaue43d5322013-10-07 20:01:52 +02001501maxsslrate <number>
1502 Sets the maximum per-process number of SSL sessions per second to <number>.
1503 SSL listeners will stop accepting connections when this limit is reached. It
1504 can be used to limit the global SSL CPU usage regardless of each frontend
1505 capacity. It is important to note that this can only be used as a service
1506 protection measure, as there will not necessarily be a fair share between
1507 frontends when the limit is reached, so it's a good idea to also limit each
1508 frontend to some value close to its expected share. It is also important to
1509 note that the sessions are accounted before they enter the SSL stack and not
1510 after, which also protects the stack against bad handshakes. Also, lowering
1511 tune.maxaccept can improve fairness.
1512
William Lallemand9d5f5482012-11-07 16:12:57 +01001513maxzlibmem <number>
1514 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1515 When the maximum amount is reached, future sessions will not compress as long
1516 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001517 The default value is 0. The value is available in bytes on the UNIX socket
1518 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1519 "ZlibMemUsage" in bytes.
1520
Willy Tarreau6a06a402007-07-15 20:15:28 +02001521noepoll
1522 Disables the use of the "epoll" event polling system on Linux. It is
1523 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001524 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001525
1526nokqueue
1527 Disables the use of the "kqueue" event polling system on BSD. It is
1528 equivalent to the command-line argument "-dk". The next polling system
1529 used will generally be "poll". See also "nopoll".
1530
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001531noevports
1532 Disables the use of the event ports event polling system on SunOS systems
1533 derived from Solaris 10 and later. It is equivalent to the command-line
1534 argument "-dv". The next polling system used will generally be "poll". See
1535 also "nopoll".
1536
Willy Tarreau6a06a402007-07-15 20:15:28 +02001537nopoll
1538 Disables the use of the "poll" event polling system. It is equivalent to the
1539 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001540 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001541 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1542 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001543
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001544nosplice
1545 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001546 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001547 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001548 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001549 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1550 be used. This option makes it easier to globally disable kernel splicing in
1551 case of doubt. See also "option splice-auto", "option splice-request" and
1552 "option splice-response".
1553
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001554nogetaddrinfo
1555 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1556 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1557
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001558noreuseport
1559 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1560 command line argument "-dR".
1561
Willy Tarreaud2d33482019-04-25 17:09:07 +02001562profiling.tasks { auto | on | off }
1563 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1564 the profiling automatically turns on a thread when it starts to suffer from
1565 an average latency of 1000 microseconds or higher as reported in the
1566 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05001567 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001568 loops so it does not vary quickly and tends to significantly smooth short
1569 spikes). It may also spontaneously trigger from time to time on overloaded
1570 systems, containers, or virtual machines, or when the system swaps (which
1571 must absolutely never happen on a load balancer).
1572
1573 CPU profiling per task can be very convenient to report where the time is
1574 spent and which requests have what effect on which other request. Enabling
1575 it will typically affect the overall's performance by less than 1%, thus it
1576 is recommended to leave it to the default 'auto' value so that it only
1577 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001578 supporting the clock_gettime(2) syscall with clock identifiers
1579 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1580 be zero. This option may be changed at run time using "set profiling" on the
1581 CLI.
1582
Willy Tarreaufe255b72007-10-14 23:09:26 +02001583spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001584 Sometimes it is desirable to avoid sending agent and health checks to
1585 servers at exact intervals, for instance when many logical servers are
1586 located on the same physical server. With the help of this parameter, it
1587 becomes possible to add some randomness in the check interval between 0
1588 and +/- 50%. A value between 2 and 5 seems to show good results. The
1589 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001590
Davor Ocelice9ed2812017-12-25 17:49:28 +01001591ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001592 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001593 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001594 multiple times, it will simply enable multiple crypto engines. Referencing an
1595 unsupported engine will prevent haproxy from starting. Note that many engines
1596 will lead to lower HTTPS performance than pure software with recent
1597 processors. The optional command "algo" sets the default algorithms an ENGINE
1598 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001599 of "ALL" uses the engine for all cryptographic operations. If no list of
1600 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001601 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1602 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1603 openssl configuration file uses:
1604 https://www.openssl.org/docs/man1.0.2/apps/config.html
1605
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001606ssl-mode-async
1607 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001608 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001609 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1610 doesn't support moving read/write buffers and is not compliant with
1611 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05001612 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001613 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001614
Willy Tarreau33cb0652014-12-23 22:52:37 +01001615tune.buffers.limit <number>
1616 Sets a hard limit on the number of buffers which may be allocated per process.
1617 The default value is zero which means unlimited. The minimum non-zero value
1618 will always be greater than "tune.buffers.reserve" and should ideally always
1619 be about twice as large. Forcing this value can be particularly useful to
1620 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001621 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001622 another one to be released by another session. Since buffers are dynamically
1623 allocated and released, the waiting time is very short and not perceptible
1624 provided that limits remain reasonable. In fact sometimes reducing the limit
1625 may even increase performance by increasing the CPU cache's efficiency. Tests
1626 have shown good results on average HTTP traffic with a limit to 1/10 of the
1627 expected global maxconn setting, which also significantly reduces memory
1628 usage. The memory savings come from the fact that a number of connections
1629 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1630 advised to do so by an haproxy core developer.
1631
Willy Tarreau1058ae72014-12-23 22:40:40 +01001632tune.buffers.reserve <number>
1633 Sets the number of buffers which are pre-allocated and reserved for use only
1634 during memory shortage conditions resulting in failed memory allocations. The
1635 minimum value is 2 and is also the default. There is no reason a user would
1636 want to change this value, it's mostly aimed at haproxy core developers.
1637
Willy Tarreau27a674e2009-08-17 07:23:33 +02001638tune.bufsize <number>
1639 Sets the buffer size to this size (in bytes). Lower values allow more
1640 sessions to coexist in the same amount of RAM, and higher values allow some
1641 applications with very large cookies to work. The default value is 16384 and
1642 can be changed at build time. It is strongly recommended not to change this
1643 from the default value, as very low values will break some services such as
1644 statistics, and values larger than default size will increase memory usage,
1645 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001646 parameter should be decreased by the same factor as this one is increased. In
1647 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1648 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001649 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001650 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1651 value set using this parameter will automatically be rounded up to the next
1652 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001653
Willy Tarreau43961d52010-10-04 20:39:20 +02001654tune.chksize <number>
1655 Sets the check buffer size to this size (in bytes). Higher values may help
1656 find string or regex patterns in very large pages, though doing so may imply
1657 more memory and CPU usage. The default value is 16384 and can be changed at
1658 build time. It is not recommended to change this value, but to use better
1659 checks whenever possible.
1660
William Lallemandf3747832012-11-09 12:33:10 +01001661tune.comp.maxlevel <number>
1662 Sets the maximum compression level. The compression level affects CPU
1663 usage during compression. This value affects CPU usage during compression.
1664 Each session using compression initializes the compression algorithm with
1665 this value. The default value is 1.
1666
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001667tune.fail-alloc
1668 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1669 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1670 success). This is useful to debug and make sure memory failures are handled
1671 gracefully.
1672
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001673tune.h2.header-table-size <number>
1674 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1675 cannot be larger than 65536 bytes. A larger value may help certain clients
1676 send more compact requests, depending on their capabilities. This amount of
1677 memory is consumed for each HTTP/2 connection. It is recommended not to
1678 change it.
1679
Willy Tarreaue6baec02017-07-27 11:45:11 +02001680tune.h2.initial-window-size <number>
1681 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001682 can upload before waiting for an acknowledgment from haproxy. This setting
1683 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001684 The default value is 65535, which roughly allows up to 5 Mbps of upload
1685 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1686 over a 1-ms local network. It can make sense to increase this value to allow
1687 faster uploads, or to reduce it to increase fairness when dealing with many
1688 clients. It doesn't affect resource usage.
1689
Willy Tarreau5242ef82017-07-27 11:47:28 +02001690tune.h2.max-concurrent-streams <number>
1691 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1692 number of outstanding requests on a single connection). The default value is
1693 100. A larger one may slightly improve page load time for complex sites when
1694 visited over high latency networks, but increases the amount of resources a
1695 single client may allocate. A value of zero disables the limit so a single
1696 client may create as many streams as allocatable by haproxy. It is highly
1697 recommended not to change this value.
1698
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001699tune.h2.max-frame-size <number>
1700 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1701 receive to its peers. The default value is the largest between 16384 and the
1702 buffer size (tune.bufsize). In any case, haproxy will not announce support
1703 for frame sizes larger than buffers. The main purpose of this setting is to
1704 allow to limit the maximum frame size setting when using large buffers. Too
1705 large frame sizes might have performance impact or cause some peers to
1706 misbehave. It is highly recommended not to change this value.
1707
Willy Tarreau193b8c62012-11-22 00:17:38 +01001708tune.http.cookielen <number>
1709 Sets the maximum length of captured cookies. This is the maximum value that
1710 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1711 will automatically be truncated to this one. It is important not to set too
1712 high a value because all cookie captures still allocate this size whatever
1713 their configured value (they share a same pool). This value is per request
1714 per response, so the memory allocated is twice this value per connection.
1715 When not specified, the limit is set to 63 characters. It is recommended not
1716 to change this value.
1717
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001718tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001719 Sets the maximum length of request URI in logs. This prevents truncating long
1720 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001721 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001722 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001723 configuration directives too.
1724 The default value is 1024.
1725
Willy Tarreauac1932d2011-10-24 19:14:41 +02001726tune.http.maxhdr <number>
1727 Sets the maximum number of headers in a request. When a request comes with a
1728 number of headers greater than this value (including the first line), it is
1729 rejected with a "400 Bad Request" status code. Similarly, too large responses
1730 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1731 for all usages, considering that the widely deployed Apache server uses the
1732 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001733 a buggy application to work by the time it gets fixed. The accepted range is
1734 1..32767. Keep in mind that each new header consumes 32bits of memory for
1735 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001736
Willy Tarreau7e312732014-02-12 16:35:14 +01001737tune.idletimer <timeout>
1738 Sets the duration after which haproxy will consider that an empty buffer is
1739 probably associated with an idle stream. This is used to optimally adjust
1740 some packet sizes while forwarding large and small data alternatively. The
1741 decision to use splice() or to send large buffers in SSL is modulated by this
1742 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1743 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001744 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05001745 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01001746 tune.ssl.maxrecord below.
1747
Willy Tarreau7ac908b2019-02-27 12:02:18 +01001748tune.listener.multi-queue { on | off }
1749 Enables ('on') or disables ('off') the listener's multi-queue accept which
1750 spreads the incoming traffic to all threads a "bind" line is allowed to run
1751 on instead of taking them for itself. This provides a smoother traffic
1752 distribution and scales much better, especially in environments where threads
1753 may be unevenly loaded due to external activity (network interrupts colliding
1754 with one thread for example). This option is enabled by default, but it may
1755 be forcefully disabled for troubleshooting or for situations where it is
1756 estimated that the operating system already provides a good enough
1757 distribution and connections are extremely short-lived.
1758
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001759tune.lua.forced-yield <number>
1760 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001761 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001762 HAProxy scheduler to process other tasks like accepting connections or
1763 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001764 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001765 lowered. If the Lua code is quite long and its result is absolutely required
1766 to process the data, the <number> can be increased.
1767
Willy Tarreau32f61e22015-03-18 17:54:59 +01001768tune.lua.maxmem
1769 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1770 default it is zero which means unlimited. It is important to set a limit to
1771 ensure that a bug in a script will not result in the system running out of
1772 memory.
1773
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001774tune.lua.session-timeout <timeout>
1775 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001776 preventing infinite loops or spending too much time in Lua. This timeout
1777 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001778 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001779
1780tune.lua.task-timeout <timeout>
1781 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1782 dedicated to the tasks. By default, this timeout isn't set because a task may
1783 remain alive during of the lifetime of HAProxy. For example, a task used to
1784 check servers.
1785
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001786tune.lua.service-timeout <timeout>
1787 This is the execution timeout for the Lua services. This is useful for
1788 preventing infinite loops or spending too much time in Lua. This timeout
1789 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001790 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001791
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001792tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001793 Sets the maximum number of consecutive connections a process may accept in a
1794 row before switching to other work. In single process mode, higher numbers
1795 give better performance at high connection rates. However in multi-process
1796 modes, keeping a bit of fairness between processes generally is better to
1797 increase performance. This value applies individually to each listener, so
1798 that the number of processes a listener is bound to is taken into account.
1799 This value defaults to 64. In multi-process mode, it is divided by twice
1800 the number of processes the listener is bound to. Setting this value to -1
1801 completely disables the limitation. It should normally not be needed to tweak
1802 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001803
1804tune.maxpollevents <number>
1805 Sets the maximum amount of events that can be processed at once in a call to
1806 the polling system. The default value is adapted to the operating system. It
1807 has been noticed that reducing it below 200 tends to slightly decrease
1808 latency at the expense of network bandwidth, and increasing it above 200
1809 tends to trade latency for slightly increased bandwidth.
1810
Willy Tarreau27a674e2009-08-17 07:23:33 +02001811tune.maxrewrite <number>
1812 Sets the reserved buffer space to this size in bytes. The reserved space is
1813 used for header rewriting or appending. The first reads on sockets will never
1814 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1815 bufsize, though that does not make much sense since there are rarely large
1816 numbers of headers to add. Setting it too high prevents processing of large
1817 requests or responses. Setting it too low prevents addition of new headers
1818 to already large requests or to POST requests. It is generally wise to set it
1819 to about 1024. It is automatically readjusted to half of bufsize if it is
1820 larger than that. This means you don't have to worry about it when changing
1821 bufsize.
1822
Willy Tarreauf3045d22015-04-29 16:24:50 +02001823tune.pattern.cache-size <number>
1824 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1825 cache which reminds previous lookups and their results. It is used by ACLs
1826 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1827 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1828 strings. It applies to pattern expressions which means that it will be able
1829 to memorize the result of a lookup among all the patterns specified on a
1830 configuration line (including all those loaded from files). It automatically
1831 invalidates entries which are updated using HTTP actions or on the CLI. The
1832 default cache size is set to 10000 entries, which limits its footprint to
1833 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1834 risk of collision in this cache, which is in the order of the size of the
1835 cache divided by 2^64. Typically, at 10000 requests per second with the
1836 default cache size of 10000 entries, there's 1% chance that a brute force
1837 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1838 This is considered much lower than the risk of a memory corruption caused by
1839 aging components. If this is not acceptable, the cache can be disabled by
1840 setting this parameter to 0.
1841
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001842tune.pipesize <number>
1843 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1844 are the default size for the system. But sometimes when using TCP splicing,
1845 it can improve performance to increase pipe sizes, especially if it is
1846 suspected that pipes are not filled and that many calls to splice() are
1847 performed. This has an impact on the kernel's memory footprint, so this must
1848 not be changed if impacts are not understood.
1849
Olivier Houchard88698d92019-04-16 19:07:22 +02001850tune.pool-low-fd-ratio <number>
1851 This setting sets the max number of file descriptors (in percentage) used by
1852 haproxy globally against the maximum number of file descriptors haproxy can
1853 use before we stop putting connection into the idle pool for reuse. The
1854 default is 20.
1855
1856tune.pool-high-fd-ratio <number>
1857 This setting sets the max number of file descriptors (in percentage) used by
1858 haproxy globally against the maximum number of file descriptors haproxy can
1859 use before we start killing idle connections when we can't reuse a connection
1860 and we have to create a new one. The default is 25 (one quarter of the file
1861 descriptor will mean that roughly half of the maximum front connections can
1862 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05001863 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02001864
Willy Tarreaue803de22010-01-21 17:43:04 +01001865tune.rcvbuf.client <number>
1866tune.rcvbuf.server <number>
1867 Forces the kernel socket receive buffer size on the client or the server side
1868 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1869 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05001870 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001871 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001872 order to save kernel memory by preventing it from buffering too large amounts
1873 of received data. Lower values will significantly increase CPU usage though.
1874
Willy Tarreaub22fc302015-12-14 12:04:35 +01001875tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001876 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001877 socket buffers. One of them is that a read returns more than <recv_enough>
1878 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1879 may be changed by this setting to better deal with workloads involving lots
1880 of short messages such as telnet or SSH sessions.
1881
Olivier Houchard1599b802018-05-24 18:59:04 +02001882tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05001883 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02001884 tasks. The default value is 200. Increasing it may incur latency when
1885 dealing with I/Os, making it too small can incur extra overhead.
1886
Willy Tarreaue803de22010-01-21 17:43:04 +01001887tune.sndbuf.client <number>
1888tune.sndbuf.server <number>
1889 Forces the kernel socket send buffer size on the client or the server side to
1890 the specified value in bytes. This value applies to all TCP/HTTP frontends
1891 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05001892 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001893 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001894 order to save kernel memory by preventing it from buffering too large amounts
1895 of received data. Lower values will significantly increase CPU usage though.
1896 Another use case is to prevent write timeouts with extremely slow clients due
1897 to the kernel waiting for a large part of the buffer to be read before
1898 notifying haproxy again.
1899
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001900tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001901 Sets the size of the global SSL session cache, in a number of blocks. A block
1902 is large enough to contain an encoded session without peer certificate.
1903 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001904 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001905 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001906 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001907 and reassigned. Higher values reduce the occurrence of such a purge, hence
1908 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1909 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001910 and are shared between all processes if "nbproc" is greater than 1. Setting
1911 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001912
Emeric Brun8dc60392014-05-09 13:52:00 +02001913tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001914 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001915 should normally not be used since it will force many renegotiations due to
1916 clients hitting a random process. But it may be required on some operating
1917 systems where none of the SSL cache synchronization method may be used. In
1918 this case, adding a first layer of hash-based load balancing before the SSL
1919 layer might limit the impact of the lack of session sharing.
1920
Emeric Brun4f65bff2012-11-16 15:11:00 +01001921tune.ssl.lifetime <timeout>
1922 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001923 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001924 does not guarantee that sessions will last that long, because if the cache is
1925 full, the longest idle sessions will be purged despite their configured
1926 lifetime. The real usefulness of this setting is to prevent sessions from
1927 being used for too long.
1928
Willy Tarreaubfd59462013-02-21 07:46:09 +01001929tune.ssl.maxrecord <number>
1930 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1931 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1932 data only once it has received a full record. With large records, it means
1933 that clients might have to download up to 16kB of data before starting to
1934 process them. Limiting the value can improve page load times on browsers
1935 located over high latency or low bandwidth networks. It is suggested to find
1936 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1937 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1938 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1939 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001940 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001941 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001942
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001943tune.ssl.default-dh-param <number>
1944 Sets the maximum size of the Diffie-Hellman parameters used for generating
1945 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1946 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1947 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1948 this maximum value. Default value if 1024. Only 1024 or higher values are
1949 allowed. Higher values will increase the CPU load, and values greater than
1950 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001951 used if static Diffie-Hellman parameters are supplied either directly
1952 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001953
Christopher Faulet31af49d2015-06-09 17:29:50 +02001954tune.ssl.ssl-ctx-cache-size <number>
1955 Sets the size of the cache used to store generated certificates to <number>
1956 entries. This is a LRU cache. Because generating a SSL certificate
1957 dynamically is expensive, they are cached. The default cache size is set to
1958 1000 entries.
1959
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001960tune.ssl.capture-cipherlist-size <number>
1961 Sets the maximum size of the buffer used for capturing client-hello cipher
1962 list. If the value is 0 (default value) the capture is disabled, otherwise
1963 a buffer is allocated for each SSL/TLS connection.
1964
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001965tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001966tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001967tune.vars.reqres-max-size <size>
1968tune.vars.sess-max-size <size>
1969tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001970 These five tunes help to manage the maximum amount of memory used by the
1971 variables system. "global" limits the overall amount of memory available for
1972 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1973 memory for the session scope, "txn" for the transaction scope, and "reqres"
1974 limits the memory for each request or response processing.
1975 Memory accounting is hierarchical, meaning more coarse grained limits include
1976 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1977 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001978
Daniel Schneller0b547052016-03-21 20:46:57 +01001979 For example, when "tune.vars.sess-max-size" is limited to 100,
1980 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1981 100 either. If we create a variable "txn.var" that contains 100 bytes,
1982 all available space is consumed.
1983 Notice that exceeding the limits at runtime will not result in an error
1984 message, but values might be cut off or corrupted. So make sure to accurately
1985 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001986
William Lallemanda509e4c2012-11-07 16:54:34 +01001987tune.zlib.memlevel <number>
1988 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001989 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001990 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01001991 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01001992 between 1 and 9. The default value is 8.
1993
1994tune.zlib.windowsize <number>
1995 Sets the window size (the size of the history buffer) as a parameter of the
1996 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01001997 in better compression at the expense of memory usage. Can be a value between
1998 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001999
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020003.3. Debugging
2001--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002002
2003debug
2004 Enables debug mode which dumps to stdout all exchanges, and disables forking
2005 into background. It is the equivalent of the command-line argument "-d". It
2006 should never be used in a production configuration since it may prevent full
2007 system startup.
2008
2009quiet
2010 Do not display any message during startup. It is equivalent to the command-
2011 line argument "-q".
2012
Emeric Brunf099e792010-09-27 12:05:28 +02002013
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010020143.4. Userlists
2015--------------
2016It is possible to control access to frontend/backend/listen sections or to
2017http stats by allowing only authenticated and authorized users. To do this,
2018it is required to create at least one userlist and to define users.
2019
2020userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002021 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002022 used to store authentication & authorization data for independent customers.
2023
2024group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002025 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002026 attach users to this group by using a comma separated list of names
2027 proceeded by "users" keyword.
2028
Cyril Bontéf0c60612010-02-06 14:44:47 +01002029user <username> [password|insecure-password <password>]
2030 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002031 Adds user <username> to the current userlist. Both secure (encrypted) and
2032 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002033 evaluated using the crypt(3) function, so depending on the system's
2034 capabilities, different algorithms are supported. For example, modern Glibc
2035 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2036 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002037
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002038 Attention: Be aware that using encrypted passwords might cause significantly
2039 increased CPU usage, depending on the number of requests, and the algorithm
2040 used. For any of the hashed variants, the password for each request must
2041 be processed through the chosen algorithm, before it can be compared to the
2042 value specified in the config file. Most current algorithms are deliberately
2043 designed to be expensive to compute to achieve resistance against brute
2044 force attacks. They do not simply salt/hash the clear text password once,
2045 but thousands of times. This can quickly become a major factor in haproxy's
2046 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002047
2048 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002049 userlist L1
2050 group G1 users tiger,scott
2051 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002052
Cyril Bontéf0c60612010-02-06 14:44:47 +01002053 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2054 user scott insecure-password elgato
2055 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002056
Cyril Bontéf0c60612010-02-06 14:44:47 +01002057 userlist L2
2058 group G1
2059 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002060
Cyril Bontéf0c60612010-02-06 14:44:47 +01002061 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2062 user scott insecure-password elgato groups G1,G2
2063 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002064
2065 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002066
Emeric Brunf099e792010-09-27 12:05:28 +02002067
20683.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002069----------
Emeric Brun94900952015-06-11 18:25:54 +02002070It is possible to propagate entries of any data-types in stick-tables between
2071several haproxy instances over TCP connections in a multi-master fashion. Each
2072instance pushes its local updates and insertions to remote peers. The pushed
2073values overwrite remote ones without aggregation. Interrupted exchanges are
2074automatically detected and recovered from the last known point.
2075In addition, during a soft restart, the old process connects to the new one
2076using such a TCP connection to push all its entries before the new process
2077tries to connect to other peers. That ensures very fast replication during a
2078reload, it typically takes a fraction of a second even for large tables.
2079Note that Server IDs are used to identify servers remotely, so it is important
2080that configurations look similar or at least that the same IDs are forced on
2081each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002082
2083peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002084 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002085 which is referenced by one or more stick-tables.
2086
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002087bind [<address>]:<port_range> [, ...] [param*]
2088 Defines the binding parameters of the local peer of this "peers" section.
2089 Such lines are not supported with "peer" line in the same "peers" section.
2090
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002091disabled
2092 Disables a peers section. It disables both listening and any synchronization
2093 related to this section. This is provided to disable synchronization of stick
2094 tables without having to comment out all "peers" references.
2095
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002096default-bind [param*]
2097 Defines the binding parameters for the local peer, excepted its address.
2098
2099default-server [param*]
2100 Change default options for a server in a "peers" section.
2101
2102 Arguments:
2103 <param*> is a list of parameters for this server. The "default-server"
2104 keyword accepts an important number of options and has a complete
2105 section dedicated to it. Please refer to section 5 for more
2106 details.
2107
2108
2109 See also: "server" and section 5 about server options
2110
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002111enable
2112 This re-enables a disabled peers section which was previously disabled.
2113
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002114peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002115 Defines a peer inside a peers section.
2116 If <peername> is set to the local peer name (by default hostname, or forced
2117 using "-L" command line option), haproxy will listen for incoming remote peer
2118 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
2119 to join the remote peer, and <peername> is used at the protocol level to
2120 identify and validate the remote peer on the server side.
2121
2122 During a soft restart, local peer <ip>:<port> is used by the old instance to
2123 connect the new one and initiate a complete replication (teaching process).
2124
2125 It is strongly recommended to have the exact same peers declaration on all
2126 peers and to only rely on the "-L" command line argument to change the local
2127 peer name. This makes it easier to maintain coherent configuration files
2128 across all peers.
2129
William Lallemandb2f07452015-05-12 14:27:13 +02002130 You may want to reference some environment variables in the address
2131 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002132
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002133 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2134 "server" keyword explanation below).
2135
2136server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002137 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002138 with a support for all "server" parameters found in 5.2 paragraph.
2139 If the underlying peer is local, <ip>:<port> parameters must not be present.
2140 These parameters must be provided on a "bind" line (see "bind" keyword
2141 of this "peers" section).
2142 Some of these parameters are irrelevant for "peers" sections.
2143
2144
Cyril Bontédc4d9032012-04-08 21:57:39 +02002145 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002146 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002147 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002148 peer haproxy1 192.168.0.1:1024
2149 peer haproxy2 192.168.0.2:1024
2150 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002151
2152 backend mybackend
2153 mode tcp
2154 balance roundrobin
2155 stick-table type ip size 20k peers mypeers
2156 stick on src
2157
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002158 server srv1 192.168.0.30:80
2159 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002160
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002161 Example:
2162 peers mypeers
2163 bind 127.0.0.11:10001 ssl crt mycerts/pem
2164 default-server ssl verify none
2165 server hostA 127.0.0.10:10000
2166 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002167
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002168
2169table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2170 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2171
2172 Configure a stickiness table for the current section. This line is parsed
2173 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002174 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002175 mandatory first parameter to designate the stick-table. Contrary to others
2176 sections, there may be several "table" lines in "peers" sections (see also
2177 "stick-table" keyword).
2178
2179 Also be aware of the fact that "peers" sections have their own stick-table
2180 namespaces to avoid collisions between stick-table names identical in
2181 different "peers" section. This is internally handled prepending the "peers"
2182 sections names to the name of the stick-tables followed by a '/' character.
2183 If somewhere else in the configuration file you have to refer to such
2184 stick-tables declared in "peers" sections you must use the prefixed version
2185 of the stick-table name as follows:
2186
2187 peers mypeers
2188 peer A ...
2189 peer B ...
2190 table t1 ...
2191
2192 frontend fe1
2193 tcp-request content track-sc0 src table mypeers/t1
2194
2195 This is also this prefixed version of the stick-table names which must be
2196 used to refer to stick-tables through the CLI.
2197
2198 About "peers" protocol, as only "peers" belonging to the same section may
2199 communicate with each others, there is no need to do such a distinction.
2200 Several "peers" sections may declare stick-tables with the same name.
2201 This is shorter version of the stick-table name which is sent over the network.
2202 There is only a '/' character as prefix to avoid stick-table name collisions between
2203 stick-tables declared as backends and stick-table declared in "peers" sections
2204 as follows in this weird but supported configuration:
2205
2206 peers mypeers
2207 peer A ...
2208 peer B ...
2209 table t1 type string size 10m store gpc0
2210
2211 backend t1
2212 stick-table type string size 10m store gpc0 peers mypeers
2213
2214 Here "t1" table declared in "mypeeers" section has "mypeers/t1" as global name.
2215 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2216 level the former table is named "/t1", the latter is again named "t1".
2217
Simon Horman51a1cf62015-02-03 13:00:44 +090022183.6. Mailers
2219------------
2220It is possible to send email alerts when the state of servers changes.
2221If configured email alerts are sent to each mailer that is configured
2222in a mailers section. Email is sent to mailers using SMTP.
2223
Pieter Baauw386a1272015-08-16 15:26:24 +02002224mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002225 Creates a new mailer list with the name <mailersect>. It is an
2226 independent section which is referenced by one or more proxies.
2227
2228mailer <mailername> <ip>:<port>
2229 Defines a mailer inside a mailers section.
2230
2231 Example:
2232 mailers mymailers
2233 mailer smtp1 192.168.0.1:587
2234 mailer smtp2 192.168.0.2:587
2235
2236 backend mybackend
2237 mode tcp
2238 balance roundrobin
2239
2240 email-alert mailers mymailers
2241 email-alert from test1@horms.org
2242 email-alert to test2@horms.org
2243
2244 server srv1 192.168.0.30:80
2245 server srv2 192.168.0.31:80
2246
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002247timeout mail <time>
2248 Defines the time available for a mail/connection to be made and send to
2249 the mail-server. If not defined the default value is 10 seconds. To allow
2250 for at least two SYN-ACK packets to be send during initial TCP handshake it
2251 is advised to keep this value above 4 seconds.
2252
2253 Example:
2254 mailers mymailers
2255 timeout mail 20s
2256 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002257
William Lallemandc9515522019-06-12 16:32:11 +020022583.7. Programs
2259-------------
2260In master-worker mode, it is possible to launch external binaries with the
2261master, these processes are called programs. These programs are launched and
2262managed the same way as the workers.
2263
2264During a reload of HAProxy, those processes are dealing with the same
2265sequence as a worker:
2266
2267 - the master is re-executed
2268 - the master sends a SIGUSR1 signal to the program
2269 - if "option start-on-reload" is not disabled, the master launches a new
2270 instance of the program
2271
2272During a stop, or restart, a SIGTERM is sent to the programs.
2273
2274program <name>
2275 This is a new program section, this section will create an instance <name>
2276 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2277 the management guide).
2278
2279command <command> [arguments*]
2280 Define the command to start with optional arguments. The command is looked
2281 up in the current PATH if it does not include an absolute path. This is a
2282 mandatory option of the program section. Arguments containing spaces must
2283 be enclosed in quotes or double quotes or be prefixed by a backslash.
2284
Andrew Heberle97236962019-07-12 11:50:26 +08002285user <user name>
2286 Changes the executed command user ID to the <user name> from /etc/passwd.
2287 See also "group".
2288
2289group <group name>
2290 Changes the executed command group ID to the <group name> from /etc/group.
2291 See also "user".
2292
William Lallemandc9515522019-06-12 16:32:11 +02002293option start-on-reload
2294no option start-on-reload
2295 Start (or not) a new instance of the program upon a reload of the master.
2296 The default is to start a new instance. This option may only be used in a
2297 program section.
2298
2299
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023004. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002301----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002302
Willy Tarreau6a06a402007-07-15 20:15:28 +02002303Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002304 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002305 - frontend <name>
2306 - backend <name>
2307 - listen <name>
2308
2309A "defaults" section sets default parameters for all other sections following
2310its declaration. Those default parameters are reset by the next "defaults"
2311section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002312section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002313
2314A "frontend" section describes a set of listening sockets accepting client
2315connections.
2316
2317A "backend" section describes a set of servers to which the proxy will connect
2318to forward incoming connections.
2319
2320A "listen" section defines a complete proxy with its frontend and backend
2321parts combined in one section. It is generally useful for TCP-only traffic.
2322
Willy Tarreau0ba27502007-12-24 16:55:16 +01002323All proxy names must be formed from upper and lower case letters, digits,
2324'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2325case-sensitive, which means that "www" and "WWW" are two different proxies.
2326
2327Historically, all proxy names could overlap, it just caused troubles in the
2328logs. Since the introduction of content switching, it is mandatory that two
2329proxies with overlapping capabilities (frontend/backend) have different names.
2330However, it is still permitted that a frontend and a backend share the same
2331name, as this configuration seems to be commonly encountered.
2332
2333Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2334and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002335bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002336protocol, and can interact with it by allowing, blocking, switching, adding,
2337modifying, or removing arbitrary contents in requests or responses, based on
2338arbitrary criteria.
2339
Willy Tarreau70dffda2014-01-30 03:07:23 +01002340In HTTP mode, the processing applied to requests and responses flowing over
2341a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02002342the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002343
2344 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2345 requests and responses are processed, and connections remain open but idle
2346 between responses and new requests.
2347
Willy Tarreau70dffda2014-01-30 03:07:23 +01002348 - SCL: server close ("option http-server-close") : the server-facing
2349 connection is closed after the end of the response is received, but the
2350 client-facing connection remains open.
2351
Christopher Faulet315b39c2018-09-21 16:26:19 +02002352 - CLO: close ("option httpclose"): the connection is closed after the end of
2353 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002354
2355The effective mode that will be applied to a connection passing through a
2356frontend and a backend can be determined by both proxy modes according to the
2357following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002358weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002359
Christopher Faulet315b39c2018-09-21 16:26:19 +02002360 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002361
Christopher Faulet315b39c2018-09-21 16:26:19 +02002362 | KAL | SCL | CLO
2363 ----+-----+-----+----
2364 KAL | KAL | SCL | CLO
2365 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02002366 mode SCL | SCL | SCL | CLO
2367 ----+-----+-----+----
2368 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002369
Willy Tarreau0ba27502007-12-24 16:55:16 +01002370
Willy Tarreau70dffda2014-01-30 03:07:23 +01002371
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023724.1. Proxy keywords matrix
2373--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002374
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002375The following list of keywords is supported. Most of them may only be used in a
2376limited set of section types. Some of them are marked as "deprecated" because
2377they are inherited from an old syntax which may be confusing or functionally
2378limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002379marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002380option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002381and must be disabled for a specific instance. Such options may also be prefixed
2382with "default" in order to restore default settings regardless of what has been
2383specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002384
Willy Tarreau6a06a402007-07-15 20:15:28 +02002385
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002386 keyword defaults frontend listen backend
2387------------------------------------+----------+----------+---------+---------
2388acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002389backlog X X X -
2390balance X - X X
2391bind - X X -
2392bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002393capture cookie - X X -
2394capture request header - X X -
2395capture response header - X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002396compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002397cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002398declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002399default-server X - X X
2400default_backend X X X -
2401description - X X X
2402disabled X X X X
2403dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002404email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002405email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002406email-alert mailers X X X X
2407email-alert myhostname X X X X
2408email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002409enabled X X X X
2410errorfile X X X X
2411errorloc X X X X
2412errorloc302 X X X X
2413-- keyword -------------------------- defaults - frontend - listen -- backend -
2414errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002415force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002416filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002417fullconn X - X X
2418grace X X X X
2419hash-type X - X X
2420http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002421http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002422http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002423http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002424http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002425http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002426http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002427id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002428ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002429load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002430log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002431log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002432log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002433log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002434max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002435maxconn X X X -
2436mode X X X X
2437monitor fail - X X -
2438monitor-net X X X -
2439monitor-uri X X X -
2440option abortonclose (*) X - X X
2441option accept-invalid-http-request (*) X X X -
2442option accept-invalid-http-response (*) X - X X
2443option allbackups (*) X - X X
2444option checkcache (*) X - X X
2445option clitcpka (*) X X X -
2446option contstats (*) X X X -
2447option dontlog-normal (*) X X X -
2448option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002449-- keyword -------------------------- defaults - frontend - listen -- backend -
2450option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02002451option h1-case-adjust-bogus-client (*) X X X -
2452option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002453option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002454option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002455option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002456option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002457option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002458option http-server-close (*) X X X X
2459option http-use-proxy-header (*) X X X -
2460option httpchk X - X X
2461option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01002462option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002463option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002464option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002465option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002466option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002467option log-health-checks (*) X - X X
2468option log-separate-errors (*) X X X -
2469option logasap (*) X X X -
2470option mysql-check X - X X
2471option nolinger (*) X X X X
2472option originalto X X X X
2473option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002474option pgsql-check X - X X
2475option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002476option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002477option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002478option smtpchk X - X X
2479option socket-stats (*) X X X -
2480option splice-auto (*) X X X X
2481option splice-request (*) X X X X
2482option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002483option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002484option srvtcpka (*) X - X X
2485option ssl-hello-chk X - X X
2486-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002487option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002488option tcp-smart-accept (*) X X X -
2489option tcp-smart-connect (*) X - X X
2490option tcpka X X X X
2491option tcplog X X X X
2492option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002493external-check command X - X X
2494external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002495persist rdp-cookie X - X X
2496rate-limit sessions X X X -
2497redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002498-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002499retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02002500retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002501server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002502server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002503server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002504source X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002505stats admin - X X X
2506stats auth X X X X
2507stats enable X X X X
2508stats hide-version X X X X
2509stats http-request - X X X
2510stats realm X X X X
2511stats refresh X X X X
2512stats scope X X X X
2513stats show-desc X X X X
2514stats show-legends X X X X
2515stats show-node X X X X
2516stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002517-- keyword -------------------------- defaults - frontend - listen -- backend -
2518stick match - - X X
2519stick on - - X X
2520stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002521stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002522stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002523tcp-check connect - - X X
2524tcp-check expect - - X X
2525tcp-check send - - X X
2526tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002527tcp-request connection - X X -
2528tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002529tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002530tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002531tcp-response content - - X X
2532tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002533timeout check X - X X
2534timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002535timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002536timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002537timeout http-keep-alive X X X X
2538timeout http-request X X X X
2539timeout queue X - X X
2540timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002541timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002542timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002543timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002544transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002545unique-id-format X X X -
2546unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002547use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002548use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002549------------------------------------+----------+----------+---------+---------
2550 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002551
Willy Tarreau0ba27502007-12-24 16:55:16 +01002552
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025534.2. Alphabetically sorted keywords reference
2554---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002555
2556This section provides a description of each keyword and its usage.
2557
2558
2559acl <aclname> <criterion> [flags] [operator] <value> ...
2560 Declare or complete an access list.
2561 May be used in sections : defaults | frontend | listen | backend
2562 no | yes | yes | yes
2563 Example:
2564 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2565 acl invalid_src src_port 0:1023
2566 acl local_dst hdr(host) -i localhost
2567
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002568 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002569
2570
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002571backlog <conns>
2572 Give hints to the system about the approximate listen backlog desired size
2573 May be used in sections : defaults | frontend | listen | backend
2574 yes | yes | yes | no
2575 Arguments :
2576 <conns> is the number of pending connections. Depending on the operating
2577 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002578 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002579
2580 In order to protect against SYN flood attacks, one solution is to increase
2581 the system's SYN backlog size. Depending on the system, sometimes it is just
2582 tunable via a system parameter, sometimes it is not adjustable at all, and
2583 sometimes the system relies on hints given by the application at the time of
2584 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2585 to the listen() syscall. On systems which can make use of this value, it can
2586 sometimes be useful to be able to specify a different value, hence this
2587 backlog parameter.
2588
2589 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2590 used as a hint and the system accepts up to the smallest greater power of
2591 two, and never more than some limits (usually 32768).
2592
2593 See also : "maxconn" and the target operating system's tuning guide.
2594
2595
Willy Tarreau0ba27502007-12-24 16:55:16 +01002596balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002597balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002598 Define the load balancing algorithm to be used in a backend.
2599 May be used in sections : defaults | frontend | listen | backend
2600 yes | no | yes | yes
2601 Arguments :
2602 <algorithm> is the algorithm used to select a server when doing load
2603 balancing. This only applies when no persistence information
2604 is available, or when a connection is redispatched to another
2605 server. <algorithm> may be one of the following :
2606
2607 roundrobin Each server is used in turns, according to their weights.
2608 This is the smoothest and fairest algorithm when the server's
2609 processing time remains equally distributed. This algorithm
2610 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002611 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002612 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002613 large farms, when a server becomes up after having been down
2614 for a very short time, it may sometimes take a few hundreds
2615 requests for it to be re-integrated into the farm and start
2616 receiving traffic. This is normal, though very rare. It is
2617 indicated here in case you would have the chance to observe
2618 it, so that you don't worry.
2619
2620 static-rr Each server is used in turns, according to their weights.
2621 This algorithm is as similar to roundrobin except that it is
2622 static, which means that changing a server's weight on the
2623 fly will have no effect. On the other hand, it has no design
2624 limitation on the number of servers, and when a server goes
2625 up, it is always immediately reintroduced into the farm, once
2626 the full map is recomputed. It also uses slightly less CPU to
2627 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002628
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002629 leastconn The server with the lowest number of connections receives the
2630 connection. Round-robin is performed within groups of servers
2631 of the same load to ensure that all servers will be used. Use
2632 of this algorithm is recommended where very long sessions are
2633 expected, such as LDAP, SQL, TSE, etc... but is not very well
2634 suited for protocols using short sessions such as HTTP. This
2635 algorithm is dynamic, which means that server weights may be
2636 adjusted on the fly for slow starts for instance.
2637
Willy Tarreauf09c6602012-02-13 17:12:08 +01002638 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002639 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002640 identifier to the highest (see server parameter "id"), which
2641 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002642 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002643 not make sense to use this algorithm without setting maxconn.
2644 The purpose of this algorithm is to always use the smallest
2645 number of servers so that extra servers can be powered off
2646 during non-intensive hours. This algorithm ignores the server
2647 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002648 or IMAP than HTTP, though it can be useful there too. In
2649 order to use this algorithm efficiently, it is recommended
2650 that a cloud controller regularly checks server usage to turn
2651 them off when unused, and regularly checks backend queue to
2652 turn new servers on when the queue inflates. Alternatively,
2653 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002654
Willy Tarreau0ba27502007-12-24 16:55:16 +01002655 source The source IP address is hashed and divided by the total
2656 weight of the running servers to designate which server will
2657 receive the request. This ensures that the same client IP
2658 address will always reach the same server as long as no
2659 server goes down or up. If the hash result changes due to the
2660 number of running servers changing, many clients will be
2661 directed to a different server. This algorithm is generally
2662 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002663 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002664 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002665 static by default, which means that changing a server's
2666 weight on the fly will have no effect, but this can be
2667 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002668
Oskar Stolc8dc41842012-05-19 10:19:54 +01002669 uri This algorithm hashes either the left part of the URI (before
2670 the question mark) or the whole URI (if the "whole" parameter
2671 is present) and divides the hash value by the total weight of
2672 the running servers. The result designates which server will
2673 receive the request. This ensures that the same URI will
2674 always be directed to the same server as long as no server
2675 goes up or down. This is used with proxy caches and
2676 anti-virus proxies in order to maximize the cache hit rate.
2677 Note that this algorithm may only be used in an HTTP backend.
2678 This algorithm is static by default, which means that
2679 changing a server's weight on the fly will have no effect,
2680 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002681
Oskar Stolc8dc41842012-05-19 10:19:54 +01002682 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002683 "depth", both followed by a positive integer number. These
2684 options may be helpful when it is needed to balance servers
2685 based on the beginning of the URI only. The "len" parameter
2686 indicates that the algorithm should only consider that many
2687 characters at the beginning of the URI to compute the hash.
2688 Note that having "len" set to 1 rarely makes sense since most
2689 URIs start with a leading "/".
2690
2691 The "depth" parameter indicates the maximum directory depth
2692 to be used to compute the hash. One level is counted for each
2693 slash in the request. If both parameters are specified, the
2694 evaluation stops when either is reached.
2695
Willy Tarreau0ba27502007-12-24 16:55:16 +01002696 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002697 the query string of each HTTP GET request.
2698
2699 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002700 request entity will be searched for the parameter argument,
2701 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002702 ('?') in the URL. The message body will only start to be
2703 analyzed once either the advertised amount of data has been
2704 received or the request buffer is full. In the unlikely event
2705 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002706 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002707 be randomly balanced if at all. This keyword used to support
2708 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002709
2710 If the parameter is found followed by an equal sign ('=') and
2711 a value, then the value is hashed and divided by the total
2712 weight of the running servers. The result designates which
2713 server will receive the request.
2714
2715 This is used to track user identifiers in requests and ensure
2716 that a same user ID will always be sent to the same server as
2717 long as no server goes up or down. If no value is found or if
2718 the parameter is not found, then a round robin algorithm is
2719 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002720 backend. This algorithm is static by default, which means
2721 that changing a server's weight on the fly will have no
2722 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002723
Cyril Bontédc4d9032012-04-08 21:57:39 +02002724 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2725 request. Just as with the equivalent ACL 'hdr()' function,
2726 the header name in parenthesis is not case sensitive. If the
2727 header is absent or if it does not contain any value, the
2728 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002729
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002730 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002731 reducing the hash algorithm to the main domain part with some
2732 specific headers such as 'Host'. For instance, in the Host
2733 value "haproxy.1wt.eu", only "1wt" will be considered.
2734
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002735 This algorithm is static by default, which means that
2736 changing a server's weight on the fly will have no effect,
2737 but this can be changed using "hash-type".
2738
Willy Tarreau21c741a2019-01-14 18:14:27 +01002739 random
2740 random(<draws>)
2741 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02002742 hashing function. This means that the servers' weights are
2743 respected, dynamic weight changes immediately take effect, as
2744 well as new server additions. Random load balancing can be
2745 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01002746 or removed as it may avoid the hammering effect that could
2747 result from roundrobin or leastconn in this situation. The
2748 hash-balance-factor directive can be used to further improve
2749 fairness of the load balancing, especially in situations
2750 where servers show highly variable response times. When an
2751 argument <draws> is present, it must be an integer value one
2752 or greater, indicating the number of draws before selecting
2753 the least loaded of these servers. It was indeed demonstrated
2754 that picking the least loaded of two servers is enough to
2755 significantly improve the fairness of the algorithm, by
2756 always avoiding to pick the most loaded server within a farm
2757 and getting rid of any bias that could be induced by the
2758 unfair distribution of the consistent list. Higher values N
2759 will take away N-1 of the highest loaded servers at the
2760 expense of performance. With very high values, the algorithm
2761 will converge towards the leastconn's result but much slower.
2762 The default value is 2, which generally shows very good
2763 distribution and performance. This algorithm is also known as
2764 the Power of Two Random Choices and is described here :
2765 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02002766
Emeric Brun736aa232009-06-30 17:56:00 +02002767 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002768 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002769 The RDP cookie <name> (or "mstshash" if omitted) will be
2770 looked up and hashed for each incoming TCP request. Just as
2771 with the equivalent ACL 'req_rdp_cookie()' function, the name
2772 is not case-sensitive. This mechanism is useful as a degraded
2773 persistence mode, as it makes it possible to always send the
2774 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002775 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002776 used instead.
2777
2778 Note that for this to work, the frontend must ensure that an
2779 RDP cookie is already present in the request buffer. For this
2780 you must use 'tcp-request content accept' rule combined with
2781 a 'req_rdp_cookie_cnt' ACL.
2782
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002783 This algorithm is static by default, which means that
2784 changing a server's weight on the fly will have no effect,
2785 but this can be changed using "hash-type".
2786
Cyril Bontédc4d9032012-04-08 21:57:39 +02002787 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002788
Willy Tarreau0ba27502007-12-24 16:55:16 +01002789 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002790 algorithms. Right now, only "url_param" and "uri" support an
2791 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002792
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002793 The load balancing algorithm of a backend is set to roundrobin when no other
2794 algorithm, mode nor option have been set. The algorithm may only be set once
2795 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002796
Lukas Tribus80512b12018-10-27 20:07:40 +02002797 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05002798 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02002799 to be routed to different backend servers, breaking the invalid assumptions
2800 NTLM relies on.
2801
Willy Tarreau0ba27502007-12-24 16:55:16 +01002802 Examples :
2803 balance roundrobin
2804 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002805 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002806 balance hdr(User-Agent)
2807 balance hdr(host)
2808 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002809
2810 Note: the following caveats and limitations on using the "check_post"
2811 extension with "url_param" must be considered :
2812
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002813 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002814 to determine if the parameters will be found in the body or entity which
2815 may contain binary data. Therefore another method may be required to
2816 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02002817 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002818
2819 - using a <max_wait> value larger than the request buffer size does not
2820 make sense and is useless. The buffer size is set at build time, and
2821 defaults to 16 kB.
2822
2823 - Content-Encoding is not supported, the parameter search will probably
2824 fail; and load balancing will fall back to Round Robin.
2825
2826 - Expect: 100-continue is not supported, load balancing will fall back to
2827 Round Robin.
2828
Lukas Tribus23953682017-04-28 13:24:30 +00002829 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002830 If the entire parameter value is not present in the first chunk, the
2831 selection of server is undefined (actually, defined by how little
2832 actually appeared in the first chunk).
2833
2834 - This feature does not support generation of a 100, 411 or 501 response.
2835
2836 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002837 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002838 white space or control characters are found, indicating the end of what
2839 might be a URL parameter list. This is probably not a concern with SGML
2840 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002841
Willy Tarreau294d0f02015-08-10 19:40:12 +02002842 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002843
2844
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002845bind [<address>]:<port_range> [, ...] [param*]
2846bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002847 Define one or several listening addresses and/or ports in a frontend.
2848 May be used in sections : defaults | frontend | listen | backend
2849 no | yes | yes | no
2850 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002851 <address> is optional and can be a host name, an IPv4 address, an IPv6
2852 address, or '*'. It designates the address the frontend will
2853 listen on. If unset, all IPv4 addresses of the system will be
2854 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002855 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002856 Optionally, an address family prefix may be used before the
2857 address to force the family regardless of the address format,
2858 which can be useful to specify a path to a unix socket with
2859 no slash ('/'). Currently supported prefixes are :
2860 - 'ipv4@' -> address is always IPv4
2861 - 'ipv6@' -> address is always IPv6
2862 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002863 - 'abns@' -> address is in abstract namespace (Linux only).
2864 Note: since abstract sockets are not "rebindable", they
2865 do not cope well with multi-process mode during
2866 soft-restart, so it is better to avoid them if
2867 nbproc is greater than 1. The effect is that if the
2868 new process fails to start, only one of the old ones
2869 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002870 - 'fd@<n>' -> use file descriptor <n> inherited from the
2871 parent. The fd must be bound and may or may not already
2872 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002873 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2874 connected unix socket or of a socketpair. The bind waits
2875 to receive a FD over the unix socket and uses it as if it
2876 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002877 You may want to reference some environment variables in the
2878 address parameter, see section 2.3 about environment
2879 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002880
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002881 <port_range> is either a unique TCP port, or a port range for which the
2882 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002883 above. The port is mandatory for TCP listeners. Note that in
2884 the case of an IPv6 address, the port is always the number
2885 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002886 - a numerical port (ex: '80')
2887 - a dash-delimited ports range explicitly stating the lower
2888 and upper bounds (ex: '2000-2100') which are included in
2889 the range.
2890
2891 Particular care must be taken against port ranges, because
2892 every <address:port> couple consumes one socket (= a file
2893 descriptor), so it's easy to consume lots of descriptors
2894 with a simple range, and to run out of sockets. Also, each
2895 <address:port> couple must be used only once among all
2896 instances running on a same system. Please note that binding
2897 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002898 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002899 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002900
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002901 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002902 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002903 receive UNIX connections on the socket located at this place.
2904 The path must begin with a slash and by default is absolute.
2905 It can be relative to the prefix defined by "unix-bind" in
2906 the global section. Note that the total length of the prefix
2907 followed by the socket path cannot exceed some system limits
2908 for UNIX sockets, which commonly are set to 107 characters.
2909
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002910 <param*> is a list of parameters common to all sockets declared on the
2911 same line. These numerous parameters depend on OS and build
2912 options and have a complete section dedicated to them. Please
2913 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002914
Willy Tarreau0ba27502007-12-24 16:55:16 +01002915 It is possible to specify a list of address:port combinations delimited by
2916 commas. The frontend will then listen on all of these addresses. There is no
2917 fixed limit to the number of addresses and ports which can be listened on in
2918 a frontend, as well as there is no limit to the number of "bind" statements
2919 in a frontend.
2920
2921 Example :
2922 listen http_proxy
2923 bind :80,:443
2924 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002925 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002926
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002927 listen http_https_proxy
2928 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002929 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002930
Willy Tarreau24709282013-03-10 21:32:12 +01002931 listen http_https_proxy_explicit
2932 bind ipv6@:80
2933 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2934 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2935
Willy Tarreaudad36a32013-03-11 01:20:04 +01002936 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002937 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002938
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002939 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2940 sun_path length is used for the address length. Some other programs
2941 such as socat use the string length only by default. Pass the option
2942 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2943 make it compatible with HAProxy's.
2944
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002945 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002946 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002947
2948
Christopher Fauletff4121f2017-11-22 16:38:49 +01002949bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002950 Limit visibility of an instance to a certain set of processes numbers.
2951 May be used in sections : defaults | frontend | listen | backend
2952 yes | yes | yes | yes
2953 Arguments :
2954 all All process will see this instance. This is the default. It
2955 may be used to override a default value.
2956
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002957 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002958 option may be combined with other numbers.
2959
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002960 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002961 option may be combined with other numbers. Do not use it
2962 with less than 2 processes otherwise some instances might be
2963 missing from all processes.
2964
Christopher Fauletff4121f2017-11-22 16:38:49 +01002965 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002966 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002967 the machine's word size. Ranges can be partially defined. The
2968 higher bound can be omitted. In such case, it is replaced by
2969 the corresponding maximum value. If a proxy is bound to
2970 process numbers greater than the configured global.nbproc, it
2971 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02002972 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002973
2974 This keyword limits binding of certain instances to certain processes. This
2975 is useful in order not to have too many processes listening to the same
2976 ports. For instance, on a dual-core machine, it might make sense to set
2977 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2978 and 'even' instances.
2979
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002980 At the moment, it is not possible to reference more than 32 or 64 processes
2981 using this keyword, but this should be more than enough for most setups.
2982 Please note that 'all' really means all processes regardless of the machine's
2983 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002984
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002985 Each "bind" line may further be limited to a subset of the proxy's processes,
2986 please consult the "process" bind keyword in section 5.1.
2987
Willy Tarreaub369a042014-09-16 13:21:03 +02002988 When a frontend has no explicit "bind-process" line, it tries to bind to all
2989 the processes referenced by its "bind" lines. That means that frontends can
2990 easily adapt to their listeners' processes.
2991
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002992 If some backends are referenced by frontends bound to other processes, the
2993 backend automatically inherits the frontend's processes.
2994
2995 Example :
2996 listen app_ip1
2997 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002998 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002999
3000 listen app_ip2
3001 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003002 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003003
3004 listen management
3005 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003006 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003007
Willy Tarreau110ecc12012-11-15 17:50:01 +01003008 listen management
3009 bind 10.0.0.4:80
3010 bind-process 1-4
3011
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003012 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003013
3014
Willy Tarreau0ba27502007-12-24 16:55:16 +01003015capture cookie <name> len <length>
3016 Capture and log a cookie in the request and in the response.
3017 May be used in sections : defaults | frontend | listen | backend
3018 no | yes | yes | no
3019 Arguments :
3020 <name> is the beginning of the name of the cookie to capture. In order
3021 to match the exact name, simply suffix the name with an equal
3022 sign ('='). The full name will appear in the logs, which is
3023 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003024 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003025
3026 <length> is the maximum number of characters to report in the logs, which
3027 include the cookie name, the equal sign and the value, all in the
3028 standard "name=value" form. The string will be truncated on the
3029 right if it exceeds <length>.
3030
3031 Only the first cookie is captured. Both the "cookie" request headers and the
3032 "set-cookie" response headers are monitored. This is particularly useful to
3033 check for application bugs causing session crossing or stealing between
3034 users, because generally the user's cookies can only change on a login page.
3035
3036 When the cookie was not presented by the client, the associated log column
3037 will report "-". When a request does not cause a cookie to be assigned by the
3038 server, a "-" is reported in the response column.
3039
3040 The capture is performed in the frontend only because it is necessary that
3041 the log format does not change for a given frontend depending on the
3042 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003043 "capture cookie" statement in a frontend. The maximum capture length is set
3044 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3045 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003046
3047 Example:
3048 capture cookie ASPSESSION len 32
3049
3050 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003051 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003052
3053
3054capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003055 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003056 May be used in sections : defaults | frontend | listen | backend
3057 no | yes | yes | no
3058 Arguments :
3059 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003060 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003061 appear in the requests, with the first letter of each word in
3062 upper case. The header name will not appear in the logs, only the
3063 value is reported, but the position in the logs is respected.
3064
3065 <length> is the maximum number of characters to extract from the value and
3066 report in the logs. The string will be truncated on the right if
3067 it exceeds <length>.
3068
Willy Tarreau4460d032012-11-21 23:37:37 +01003069 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003070 value will be added to the logs between braces ('{}'). If multiple headers
3071 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003072 in the same order they were declared in the configuration. Non-existent
3073 headers will be logged just as an empty string. Common uses for request
3074 header captures include the "Host" field in virtual hosting environments, the
3075 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003076 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003077 environments to find where the request came from.
3078
3079 Note that when capturing headers such as "User-agent", some spaces may be
3080 logged, making the log analysis more difficult. Thus be careful about what
3081 you log if you know your log parser is not smart enough to rely on the
3082 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003083
Willy Tarreau0900abb2012-11-22 00:21:46 +01003084 There is no limit to the number of captured request headers nor to their
3085 length, though it is wise to keep them low to limit memory usage per session.
3086 In order to keep log format consistent for a same frontend, header captures
3087 can only be declared in a frontend. It is not possible to specify a capture
3088 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003089
3090 Example:
3091 capture request header Host len 15
3092 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003093 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003094
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003095 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003096 about logging.
3097
3098
3099capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003100 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003101 May be used in sections : defaults | frontend | listen | backend
3102 no | yes | yes | no
3103 Arguments :
3104 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003105 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003106 appear in the response, with the first letter of each word in
3107 upper case. The header name will not appear in the logs, only the
3108 value is reported, but the position in the logs is respected.
3109
3110 <length> is the maximum number of characters to extract from the value and
3111 report in the logs. The string will be truncated on the right if
3112 it exceeds <length>.
3113
Willy Tarreau4460d032012-11-21 23:37:37 +01003114 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003115 result will be added to the logs between braces ('{}') after the captured
3116 request headers. If multiple headers are captured, they will be delimited by
3117 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003118 the configuration. Non-existent headers will be logged just as an empty
3119 string. Common uses for response header captures include the "Content-length"
3120 header which indicates how many bytes are expected to be returned, the
3121 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003122
Willy Tarreau0900abb2012-11-22 00:21:46 +01003123 There is no limit to the number of captured response headers nor to their
3124 length, though it is wise to keep them low to limit memory usage per session.
3125 In order to keep log format consistent for a same frontend, header captures
3126 can only be declared in a frontend. It is not possible to specify a capture
3127 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003128
3129 Example:
3130 capture response header Content-length len 9
3131 capture response header Location len 15
3132
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003133 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003134 about logging.
3135
3136
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003137compression algo <algorithm> ...
3138compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003139compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003140 Enable HTTP compression.
3141 May be used in sections : defaults | frontend | listen | backend
3142 yes | yes | yes | yes
3143 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003144 algo is followed by the list of supported compression algorithms.
3145 type is followed by the list of MIME types that will be compressed.
3146 offload makes haproxy work as a compression offloader only (see notes).
3147
3148 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003149 identity this is mostly for debugging, and it was useful for developing
3150 the compression feature. Identity does not apply any change on
3151 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003152
Willy Tarreauc91840a2015-03-28 17:00:39 +01003153 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003154 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003155
3156 deflate same as "gzip", but with deflate algorithm and zlib format.
3157 Note that this algorithm has ambiguous support on many
3158 browsers and no support at all from recent ones. It is
3159 strongly recommended not to use it for anything else than
3160 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003161 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003162
Willy Tarreauc91840a2015-03-28 17:00:39 +01003163 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3164 alternative when the browser wants "deflate". All major
3165 browsers understand it and despite violating the standards,
3166 it is known to work better than "deflate", at least on MSIE
3167 and some versions of Safari. Do not use it in conjunction
3168 with "deflate", use either one or the other since both react
3169 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003170 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003171
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003172 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003173 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003174 If backend servers support HTTP compression, these directives
3175 will be no-op: haproxy will see the compressed response and will not
3176 compress again. If backend servers do not support HTTP compression and
3177 there is Accept-Encoding header in request, haproxy will compress the
3178 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003179
3180 The "offload" setting makes haproxy remove the Accept-Encoding header to
3181 prevent backend servers from compressing responses. It is strongly
3182 recommended not to do this because this means that all the compression work
3183 will be done on the single point where haproxy is located. However in some
3184 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003185 with broken HTTP compression implementation which can't be turned off.
3186 In that case haproxy can be used to prevent that gateway from emitting
3187 invalid payloads. In this case, simply removing the header in the
3188 configuration does not work because it applies before the header is parsed,
3189 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003190 then be used for such scenarios. Note: for now, the "offload" setting is
3191 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003192
William Lallemand05097442012-11-20 12:14:28 +01003193 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003194 * the request does not advertise a supported compression algorithm in the
3195 "Accept-Encoding" header
3196 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003197 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003198 * response contain neither a "Content-Length" header nor a
3199 "Transfer-Encoding" whose last value is "chunked"
3200 * response contains a "Content-Type" header whose first value starts with
3201 "multipart"
3202 * the response contains the "no-transform" value in the "Cache-control"
3203 header
3204 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3205 and later
3206 * The response contains a "Content-Encoding" header, indicating that the
3207 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003208 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003209
Tim Duesterhusb229f012019-01-29 16:38:56 +01003210 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003211
William Lallemand82fe75c2012-10-23 10:25:10 +02003212 Examples :
3213 compression algo gzip
3214 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003215
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003216
Willy Tarreau55165fe2009-05-10 12:02:55 +02003217cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003218 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3219 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003220 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003221 Enable cookie-based persistence in a backend.
3222 May be used in sections : defaults | frontend | listen | backend
3223 yes | no | yes | yes
3224 Arguments :
3225 <name> is the name of the cookie which will be monitored, modified or
3226 inserted in order to bring persistence. This cookie is sent to
3227 the client via a "Set-Cookie" header in the response, and is
3228 brought back by the client in a "Cookie" header in all requests.
3229 Special care should be taken to choose a name which does not
3230 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003231 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003232 HTTP/HTTPS), care should be taken to use different cookie names
3233 between all backends if persistence between them is not desired.
3234
3235 rewrite This keyword indicates that the cookie will be provided by the
3236 server and that haproxy will have to modify its value to set the
3237 server's identifier in it. This mode is handy when the management
3238 of complex combinations of "Set-cookie" and "Cache-control"
3239 headers is left to the application. The application can then
3240 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003241 cookie. Since all responses should be monitored, this mode
3242 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003243 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003244 start with this mode for new deployments. This keyword is
3245 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003246
3247 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003248 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003249
Willy Tarreaua79094d2010-08-31 22:54:15 +02003250 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003251 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003252 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003253 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003254 existing configurations running in the "rewrite" mode. The cookie
3255 will only be a session cookie and will not be stored on the
3256 client's disk. By default, unless the "indirect" option is added,
3257 the server will see the cookies emitted by the client. Due to
3258 caching effects, it is generally wise to add the "nocache" or
3259 "postonly" keywords (see below). The "insert" keyword is not
3260 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003261
3262 prefix This keyword indicates that instead of relying on a dedicated
3263 cookie for the persistence, an existing one will be completed.
3264 This may be needed in some specific environments where the client
3265 does not support more than one single cookie and the application
3266 already needs it. In this case, whenever the server sets a cookie
3267 named <name>, it will be prefixed with the server's identifier
3268 and a delimiter. The prefix will be removed from all client
3269 requests so that the server still finds the cookie it emitted.
3270 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003271 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003272 not compatible with "rewrite" and "insert". Note: it is highly
3273 recommended not to use "indirect" with "prefix", otherwise server
3274 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003275
Willy Tarreaua79094d2010-08-31 22:54:15 +02003276 indirect When this option is specified, no cookie will be emitted to a
3277 client which already has a valid one for the server which has
3278 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003279 it will be removed, unless the "preserve" option is also set. In
3280 "insert" mode, this will additionally remove cookies from the
3281 requests transmitted to the server, making the persistence
3282 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003283 Note: it is highly recommended not to use "indirect" with
3284 "prefix", otherwise server cookie updates would not be sent to
3285 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003286
3287 nocache This option is recommended in conjunction with the insert mode
3288 when there is a cache between the client and HAProxy, as it
3289 ensures that a cacheable response will be tagged non-cacheable if
3290 a cookie needs to be inserted. This is important because if all
3291 persistence cookies are added on a cacheable home page for
3292 instance, then all customers will then fetch the page from an
3293 outer cache and will all share the same persistence cookie,
3294 leading to one server receiving much more traffic than others.
3295 See also the "insert" and "postonly" options.
3296
3297 postonly This option ensures that cookie insertion will only be performed
3298 on responses to POST requests. It is an alternative to the
3299 "nocache" option, because POST responses are not cacheable, so
3300 this ensures that the persistence cookie will never get cached.
3301 Since most sites do not need any sort of persistence before the
3302 first POST which generally is a login request, this is a very
3303 efficient method to optimize caching without risking to find a
3304 persistence cookie in the cache.
3305 See also the "insert" and "nocache" options.
3306
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003307 preserve This option may only be used with "insert" and/or "indirect". It
3308 allows the server to emit the persistence cookie itself. In this
3309 case, if a cookie is found in the response, haproxy will leave it
3310 untouched. This is useful in order to end persistence after a
3311 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003312 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003313 the past. By combining this mechanism with the "disable-on-404"
3314 check option, it is possible to perform a completely graceful
3315 shutdown because users will definitely leave the server after
3316 they logout.
3317
Willy Tarreau4992dd22012-05-31 21:02:17 +02003318 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3319 when a cookie is inserted. This attribute is used so that a
3320 user agent doesn't share the cookie with non-HTTP components.
3321 Please check RFC6265 for more information on this attribute.
3322
3323 secure This option tells haproxy to add a "Secure" cookie attribute when
3324 a cookie is inserted. This attribute is used so that a user agent
3325 never emits this cookie over non-secure channels, which means
3326 that a cookie learned with this flag will be presented only over
3327 SSL/TLS connections. Please check RFC6265 for more information on
3328 this attribute.
3329
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003330 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003331 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003332 name. If the domain begins with a dot, the browser is allowed to
3333 use it for any host ending with that name. It is also possible to
3334 specify several domain names by invoking this option multiple
3335 times. Some browsers might have small limits on the number of
3336 domains, so be careful when doing that. For the record, sending
3337 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003338
Willy Tarreau996a92c2010-10-13 19:30:47 +02003339 maxidle This option allows inserted cookies to be ignored after some idle
3340 time. It only works with insert-mode cookies. When a cookie is
3341 sent to the client, the date this cookie was emitted is sent too.
3342 Upon further presentations of this cookie, if the date is older
3343 than the delay indicated by the parameter (in seconds), it will
3344 be ignored. Otherwise, it will be refreshed if needed when the
3345 response is sent to the client. This is particularly useful to
3346 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003347 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003348 this option is set and a cookie has no date, it is always
3349 accepted, but gets refreshed in the response. This maintains the
3350 ability for admins to access their sites. Cookies that have a
3351 date in the future further than 24 hours are ignored. Doing so
3352 lets admins fix timezone issues without risking kicking users off
3353 the site.
3354
3355 maxlife This option allows inserted cookies to be ignored after some life
3356 time, whether they're in use or not. It only works with insert
3357 mode cookies. When a cookie is first sent to the client, the date
3358 this cookie was emitted is sent too. Upon further presentations
3359 of this cookie, if the date is older than the delay indicated by
3360 the parameter (in seconds), it will be ignored. If the cookie in
3361 the request has no date, it is accepted and a date will be set.
3362 Cookies that have a date in the future further than 24 hours are
3363 ignored. Doing so lets admins fix timezone issues without risking
3364 kicking users off the site. Contrary to maxidle, this value is
3365 not refreshed, only the first visit date counts. Both maxidle and
3366 maxlife may be used at the time. This is particularly useful to
3367 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003368 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003369 is stronger than the maxidle method in that it forces a
3370 redispatch after some absolute delay.
3371
Olivier Houchard4e694042017-03-14 20:01:29 +01003372 dynamic Activate dynamic cookies. When used, a session cookie is
3373 dynamically created for each server, based on the IP and port
3374 of the server, and a secret key, specified in the
3375 "dynamic-cookie-key" backend directive.
3376 The cookie will be regenerated each time the IP address change,
3377 and is only generated for IPv4/IPv6.
3378
Willy Tarreau0ba27502007-12-24 16:55:16 +01003379 There can be only one persistence cookie per HTTP backend, and it can be
3380 declared in a defaults section. The value of the cookie will be the value
3381 indicated after the "cookie" keyword in a "server" statement. If no cookie
3382 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003383
Willy Tarreau0ba27502007-12-24 16:55:16 +01003384 Examples :
3385 cookie JSESSIONID prefix
3386 cookie SRV insert indirect nocache
3387 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003388 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003389
Willy Tarreau294d0f02015-08-10 19:40:12 +02003390 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003391
Willy Tarreau983e01e2010-01-11 18:42:06 +01003392
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003393declare capture [ request | response ] len <length>
3394 Declares a capture slot.
3395 May be used in sections : defaults | frontend | listen | backend
3396 no | yes | yes | no
3397 Arguments:
3398 <length> is the length allowed for the capture.
3399
3400 This declaration is only available in the frontend or listen section, but the
3401 reserved slot can be used in the backends. The "request" keyword allocates a
3402 capture slot for use in the request, and "response" allocates a capture slot
3403 for use in the response.
3404
3405 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003406 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003407 "http-request capture" and "http-response capture".
3408
3409
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003410default-server [param*]
3411 Change default options for a server in a backend
3412 May be used in sections : defaults | frontend | listen | backend
3413 yes | no | yes | yes
3414 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003415 <param*> is a list of parameters for this server. The "default-server"
3416 keyword accepts an important number of options and has a complete
3417 section dedicated to it. Please refer to section 5 for more
3418 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003419
Willy Tarreau983e01e2010-01-11 18:42:06 +01003420 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003421 default-server inter 1000 weight 13
3422
3423 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003424
Willy Tarreau983e01e2010-01-11 18:42:06 +01003425
Willy Tarreau0ba27502007-12-24 16:55:16 +01003426default_backend <backend>
3427 Specify the backend to use when no "use_backend" rule has been matched.
3428 May be used in sections : defaults | frontend | listen | backend
3429 yes | yes | yes | no
3430 Arguments :
3431 <backend> is the name of the backend to use.
3432
3433 When doing content-switching between frontend and backends using the
3434 "use_backend" keyword, it is often useful to indicate which backend will be
3435 used when no rule has matched. It generally is the dynamic backend which
3436 will catch all undetermined requests.
3437
Willy Tarreau0ba27502007-12-24 16:55:16 +01003438 Example :
3439
3440 use_backend dynamic if url_dyn
3441 use_backend static if url_css url_img extension_img
3442 default_backend dynamic
3443
Willy Tarreau98d04852015-05-26 12:18:29 +02003444 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003445
Willy Tarreau0ba27502007-12-24 16:55:16 +01003446
Baptiste Assmann27f51342013-10-09 06:51:49 +02003447description <string>
3448 Describe a listen, frontend or backend.
3449 May be used in sections : defaults | frontend | listen | backend
3450 no | yes | yes | yes
3451 Arguments : string
3452
3453 Allows to add a sentence to describe the related object in the HAProxy HTML
3454 stats page. The description will be printed on the right of the object name
3455 it describes.
3456 No need to backslash spaces in the <string> arguments.
3457
3458
Willy Tarreau0ba27502007-12-24 16:55:16 +01003459disabled
3460 Disable a proxy, frontend or backend.
3461 May be used in sections : defaults | frontend | listen | backend
3462 yes | yes | yes | yes
3463 Arguments : none
3464
3465 The "disabled" keyword is used to disable an instance, mainly in order to
3466 liberate a listening port or to temporarily disable a service. The instance
3467 will still be created and its configuration will be checked, but it will be
3468 created in the "stopped" state and will appear as such in the statistics. It
3469 will not receive any traffic nor will it send any health-checks or logs. It
3470 is possible to disable many instances at once by adding the "disabled"
3471 keyword in a "defaults" section.
3472
3473 See also : "enabled"
3474
3475
Willy Tarreau5ce94572010-06-07 14:35:41 +02003476dispatch <address>:<port>
3477 Set a default server address
3478 May be used in sections : defaults | frontend | listen | backend
3479 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003480 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003481
3482 <address> is the IPv4 address of the default server. Alternatively, a
3483 resolvable hostname is supported, but this name will be resolved
3484 during start-up.
3485
3486 <ports> is a mandatory port specification. All connections will be sent
3487 to this port, and it is not permitted to use port offsets as is
3488 possible with normal servers.
3489
Willy Tarreau787aed52011-04-15 06:45:37 +02003490 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003491 server can take the connection. In the past it was used to forward non
3492 persistent connections to an auxiliary load balancer. Due to its simple
3493 syntax, it has also been used for simple TCP relays. It is recommended not to
3494 use it for more clarity, and to use the "server" directive instead.
3495
3496 See also : "server"
3497
Olivier Houchard4e694042017-03-14 20:01:29 +01003498
3499dynamic-cookie-key <string>
3500 Set the dynamic cookie secret key for a backend.
3501 May be used in sections : defaults | frontend | listen | backend
3502 yes | no | yes | yes
3503 Arguments : The secret key to be used.
3504
3505 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003506 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003507 specified on the "server" line), using a hash of the IP address of the
3508 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003509 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003510 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003511
Willy Tarreau0ba27502007-12-24 16:55:16 +01003512enabled
3513 Enable a proxy, frontend or backend.
3514 May be used in sections : defaults | frontend | listen | backend
3515 yes | yes | yes | yes
3516 Arguments : none
3517
3518 The "enabled" keyword is used to explicitly enable an instance, when the
3519 defaults has been set to "disabled". This is very rarely used.
3520
3521 See also : "disabled"
3522
3523
3524errorfile <code> <file>
3525 Return a file contents instead of errors generated by HAProxy
3526 May be used in sections : defaults | frontend | listen | backend
3527 yes | yes | yes | yes
3528 Arguments :
3529 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003530 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3531 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003532
3533 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003534 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003535 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003536 error pages, and to use absolute paths, since files are read
3537 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003538
3539 It is important to understand that this keyword is not meant to rewrite
3540 errors returned by the server, but errors detected and returned by HAProxy.
3541 This is why the list of supported errors is limited to a small set.
3542
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003543 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3544
Willy Tarreau0ba27502007-12-24 16:55:16 +01003545 The files are returned verbatim on the TCP socket. This allows any trick such
3546 as redirections to another URL or site, as well as tricks to clean cookies,
3547 force enable or disable caching, etc... The package provides default error
3548 files returning the same contents as default errors.
3549
Willy Tarreau59140a22009-02-22 12:02:30 +01003550 The files should not exceed the configured buffer size (BUFSIZE), which
3551 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003552 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003553 loops between the client and HAProxy when all servers are down, causing an
3554 error to be returned instead of an image. For better HTTP compliance, it is
3555 recommended that all header lines end with CR-LF and not LF alone.
3556
Willy Tarreau0ba27502007-12-24 16:55:16 +01003557 The files are read at the same time as the configuration and kept in memory.
3558 For this reason, the errors continue to be returned even when the process is
3559 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003560 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003561 403 status code and interrogating a blocked URL.
3562
3563 See also : "errorloc", "errorloc302", "errorloc303"
3564
Willy Tarreau59140a22009-02-22 12:02:30 +01003565 Example :
3566 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003567 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003568 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3569 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3570
Willy Tarreau2769aa02007-12-27 18:26:09 +01003571
3572errorloc <code> <url>
3573errorloc302 <code> <url>
3574 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3575 May be used in sections : defaults | frontend | listen | backend
3576 yes | yes | yes | yes
3577 Arguments :
3578 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003579 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3580 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003581
3582 <url> it is the exact contents of the "Location" header. It may contain
3583 either a relative URI to an error page hosted on the same site,
3584 or an absolute URI designating an error page on another site.
3585 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003586 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003587
3588 It is important to understand that this keyword is not meant to rewrite
3589 errors returned by the server, but errors detected and returned by HAProxy.
3590 This is why the list of supported errors is limited to a small set.
3591
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003592 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3593
Willy Tarreau2769aa02007-12-27 18:26:09 +01003594 Note that both keyword return the HTTP 302 status code, which tells the
3595 client to fetch the designated URL using the same HTTP method. This can be
3596 quite problematic in case of non-GET methods such as POST, because the URL
3597 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003598 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003599 status code, indicating to the client that the URL must be fetched with a GET
3600 request.
3601
3602 See also : "errorfile", "errorloc303"
3603
3604
3605errorloc303 <code> <url>
3606 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3607 May be used in sections : defaults | frontend | listen | backend
3608 yes | yes | yes | yes
3609 Arguments :
3610 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003611 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3612 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003613
3614 <url> it is the exact contents of the "Location" header. It may contain
3615 either a relative URI to an error page hosted on the same site,
3616 or an absolute URI designating an error page on another site.
3617 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003618 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003619
3620 It is important to understand that this keyword is not meant to rewrite
3621 errors returned by the server, but errors detected and returned by HAProxy.
3622 This is why the list of supported errors is limited to a small set.
3623
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003624 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3625
Willy Tarreau2769aa02007-12-27 18:26:09 +01003626 Note that both keyword return the HTTP 303 status code, which tells the
3627 client to fetch the designated URL using the same HTTP GET method. This
3628 solves the usual problems associated with "errorloc" and the 302 code. It is
3629 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003630 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003631
3632 See also : "errorfile", "errorloc", "errorloc302"
3633
3634
Simon Horman51a1cf62015-02-03 13:00:44 +09003635email-alert from <emailaddr>
3636 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003637 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003638 May be used in sections: defaults | frontend | listen | backend
3639 yes | yes | yes | yes
3640
3641 Arguments :
3642
3643 <emailaddr> is the from email address to use when sending email alerts
3644
3645 Also requires "email-alert mailers" and "email-alert to" to be set
3646 and if so sending email alerts is enabled for the proxy.
3647
Simon Horman64e34162015-02-06 11:11:57 +09003648 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003649 "email-alert myhostname", "email-alert to", section 3.6 about
3650 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003651
3652
3653email-alert level <level>
3654 Declare the maximum log level of messages for which email alerts will be
3655 sent. This acts as a filter on the sending of email alerts.
3656 May be used in sections: defaults | frontend | listen | backend
3657 yes | yes | yes | yes
3658
3659 Arguments :
3660
3661 <level> One of the 8 syslog levels:
3662 emerg alert crit err warning notice info debug
3663 The above syslog levels are ordered from lowest to highest.
3664
3665 By default level is alert
3666
3667 Also requires "email-alert from", "email-alert mailers" and
3668 "email-alert to" to be set and if so sending email alerts is enabled
3669 for the proxy.
3670
Simon Horman1421e212015-04-30 13:10:35 +09003671 Alerts are sent when :
3672
3673 * An un-paused server is marked as down and <level> is alert or lower
3674 * A paused server is marked as down and <level> is notice or lower
3675 * A server is marked as up or enters the drain state and <level>
3676 is notice or lower
3677 * "option log-health-checks" is enabled, <level> is info or lower,
3678 and a health check status update occurs
3679
Simon Horman64e34162015-02-06 11:11:57 +09003680 See also : "email-alert from", "email-alert mailers",
3681 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003682 section 3.6 about mailers.
3683
3684
3685email-alert mailers <mailersect>
3686 Declare the mailers to be used when sending email alerts
3687 May be used in sections: defaults | frontend | listen | backend
3688 yes | yes | yes | yes
3689
3690 Arguments :
3691
3692 <mailersect> is the name of the mailers section to send email alerts.
3693
3694 Also requires "email-alert from" and "email-alert to" to be set
3695 and if so sending email alerts is enabled for the proxy.
3696
Simon Horman64e34162015-02-06 11:11:57 +09003697 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3698 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003699
3700
3701email-alert myhostname <hostname>
3702 Declare the to hostname address to be used when communicating with
3703 mailers.
3704 May be used in sections: defaults | frontend | listen | backend
3705 yes | yes | yes | yes
3706
3707 Arguments :
3708
Baptiste Assmann738bad92015-12-21 15:27:53 +01003709 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003710
3711 By default the systems hostname is used.
3712
3713 Also requires "email-alert from", "email-alert mailers" and
3714 "email-alert to" to be set and if so sending email alerts is enabled
3715 for the proxy.
3716
Simon Horman64e34162015-02-06 11:11:57 +09003717 See also : "email-alert from", "email-alert level", "email-alert mailers",
3718 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003719
3720
3721email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003722 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003723 header of email alerts. This is the address that email alerts are sent to.
3724 May be used in sections: defaults | frontend | listen | backend
3725 yes | yes | yes | yes
3726
3727 Arguments :
3728
3729 <emailaddr> is the to email address to use when sending email alerts
3730
3731 Also requires "email-alert mailers" and "email-alert to" to be set
3732 and if so sending email alerts is enabled for the proxy.
3733
Simon Horman64e34162015-02-06 11:11:57 +09003734 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003735 "email-alert myhostname", section 3.6 about mailers.
3736
3737
Willy Tarreau4de91492010-01-22 19:10:05 +01003738force-persist { if | unless } <condition>
3739 Declare a condition to force persistence on down servers
3740 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003741 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003742
3743 By default, requests are not dispatched to down servers. It is possible to
3744 force this using "option persist", but it is unconditional and redispatches
3745 to a valid server if "option redispatch" is set. That leaves with very little
3746 possibilities to force some requests to reach a server which is artificially
3747 marked down for maintenance operations.
3748
3749 The "force-persist" statement allows one to declare various ACL-based
3750 conditions which, when met, will cause a request to ignore the down status of
3751 a server and still try to connect to it. That makes it possible to start a
3752 server, still replying an error to the health checks, and run a specially
3753 configured browser to test the service. Among the handy methods, one could
3754 use a specific source IP address, or a specific cookie. The cookie also has
3755 the advantage that it can easily be added/removed on the browser from a test
3756 page. Once the service is validated, it is then possible to open the service
3757 to the world by returning a valid response to health checks.
3758
3759 The forced persistence is enabled when an "if" condition is met, or unless an
3760 "unless" condition is met. The final redispatch is always disabled when this
3761 is used.
3762
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003763 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003764 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003765
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003766
3767filter <name> [param*]
3768 Add the filter <name> in the filter list attached to the proxy.
3769 May be used in sections : defaults | frontend | listen | backend
3770 no | yes | yes | yes
3771 Arguments :
3772 <name> is the name of the filter. Officially supported filters are
3773 referenced in section 9.
3774
Tim Düsterhus4896c442016-11-29 02:15:19 +01003775 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003776 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003777 filter. Please refer to the documentation of the corresponding
3778 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003779
3780 Multiple occurrences of the filter line can be used for the same proxy. The
3781 same filter can be referenced many times if needed.
3782
3783 Example:
3784 listen
3785 bind *:80
3786
3787 filter trace name BEFORE-HTTP-COMP
3788 filter compression
3789 filter trace name AFTER-HTTP-COMP
3790
3791 compression algo gzip
3792 compression offload
3793
3794 server srv1 192.168.0.1:80
3795
3796 See also : section 9.
3797
Willy Tarreau4de91492010-01-22 19:10:05 +01003798
Willy Tarreau2769aa02007-12-27 18:26:09 +01003799fullconn <conns>
3800 Specify at what backend load the servers will reach their maxconn
3801 May be used in sections : defaults | frontend | listen | backend
3802 yes | no | yes | yes
3803 Arguments :
3804 <conns> is the number of connections on the backend which will make the
3805 servers use the maximal number of connections.
3806
Willy Tarreau198a7442008-01-17 12:05:32 +01003807 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003808 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003809 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003810 load. The server will then always accept at least <minconn> connections,
3811 never more than <maxconn>, and the limit will be on the ramp between both
3812 values when the backend has less than <conns> concurrent connections. This
3813 makes it possible to limit the load on the servers during normal loads, but
3814 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003815 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003816
Willy Tarreaufbb78422011-06-05 15:38:35 +02003817 Since it's hard to get this value right, haproxy automatically sets it to
3818 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003819 backend (based on "use_backend" and "default_backend" rules). That way it's
3820 safe to leave it unset. However, "use_backend" involving dynamic names are
3821 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003822
Willy Tarreau2769aa02007-12-27 18:26:09 +01003823 Example :
3824 # The servers will accept between 100 and 1000 concurrent connections each
3825 # and the maximum of 1000 will be reached when the backend reaches 10000
3826 # connections.
3827 backend dynamic
3828 fullconn 10000
3829 server srv1 dyn1:80 minconn 100 maxconn 1000
3830 server srv2 dyn2:80 minconn 100 maxconn 1000
3831
3832 See also : "maxconn", "server"
3833
3834
3835grace <time>
3836 Maintain a proxy operational for some time after a soft stop
3837 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003838 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003839 Arguments :
3840 <time> is the time (by default in milliseconds) for which the instance
3841 will remain operational with the frontend sockets still listening
3842 when a soft-stop is received via the SIGUSR1 signal.
3843
3844 This may be used to ensure that the services disappear in a certain order.
3845 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003846 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003847 needed by the equipment to detect the failure.
3848
3849 Note that currently, there is very little benefit in using this parameter,
3850 and it may in fact complicate the soft-reconfiguration process more than
3851 simplify it.
3852
Willy Tarreau0ba27502007-12-24 16:55:16 +01003853
Andrew Rodland17be45e2016-10-25 17:04:12 -04003854hash-balance-factor <factor>
3855 Specify the balancing factor for bounded-load consistent hashing
3856 May be used in sections : defaults | frontend | listen | backend
3857 yes | no | no | yes
3858 Arguments :
3859 <factor> is the control for the maximum number of concurrent requests to
3860 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01003861 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04003862
3863 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3864 enables an algorithm that prevents any one server from getting too many
3865 requests at once, even if some hash buckets receive many more requests than
3866 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3867 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3868 then no server will be allowed to have a load more than 1.5 times the average.
3869 If server weights are used, they will be respected.
3870
3871 If the first-choice server is disqualified, the algorithm will choose another
3872 server based on the request hash, until a server with additional capacity is
3873 found. A higher <factor> allows more imbalance between the servers, while a
3874 lower <factor> means that more servers will be checked on average, affecting
3875 performance. Reasonable values are from 125 to 200.
3876
Willy Tarreau760e81d2018-05-03 07:20:40 +02003877 This setting is also used by "balance random" which internally relies on the
3878 consistent hashing mechanism.
3879
Andrew Rodland17be45e2016-10-25 17:04:12 -04003880 See also : "balance" and "hash-type".
3881
3882
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003883hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003884 Specify a method to use for mapping hashes to servers
3885 May be used in sections : defaults | frontend | listen | backend
3886 yes | no | yes | yes
3887 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003888 <method> is the method used to select a server from the hash computed by
3889 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003890
Bhaskar98634f02013-10-29 23:30:51 -04003891 map-based the hash table is a static array containing all alive servers.
3892 The hashes will be very smooth, will consider weights, but
3893 will be static in that weight changes while a server is up
3894 will be ignored. This means that there will be no slow start.
3895 Also, since a server is selected by its position in the array,
3896 most mappings are changed when the server count changes. This
3897 means that when a server goes up or down, or when a server is
3898 added to a farm, most connections will be redistributed to
3899 different servers. This can be inconvenient with caches for
3900 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003901
Bhaskar98634f02013-10-29 23:30:51 -04003902 consistent the hash table is a tree filled with many occurrences of each
3903 server. The hash key is looked up in the tree and the closest
3904 server is chosen. This hash is dynamic, it supports changing
3905 weights while the servers are up, so it is compatible with the
3906 slow start feature. It has the advantage that when a server
3907 goes up or down, only its associations are moved. When a
3908 server is added to the farm, only a few part of the mappings
3909 are redistributed, making it an ideal method for caches.
3910 However, due to its principle, the distribution will never be
3911 very smooth and it may sometimes be necessary to adjust a
3912 server's weight or its ID to get a more balanced distribution.
3913 In order to get the same distribution on multiple load
3914 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003915 same IDs. Note: consistent hash uses sdbm and avalanche if no
3916 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003917
3918 <function> is the hash function to be used :
3919
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003920 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003921 reimplementation of ndbm) database library. It was found to do
3922 well in scrambling bits, causing better distribution of the keys
3923 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003924 function with good distribution, unless the total server weight
3925 is a multiple of 64, in which case applying the avalanche
3926 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003927
3928 djb2 this function was first proposed by Dan Bernstein many years ago
3929 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003930 function provides a better distribution than sdbm. It generally
3931 works well with text-based inputs though it can perform extremely
3932 poorly with numeric-only input or when the total server weight is
3933 a multiple of 33, unless the avalanche modifier is also used.
3934
Willy Tarreaua0f42712013-11-14 14:30:35 +01003935 wt6 this function was designed for haproxy while testing other
3936 functions in the past. It is not as smooth as the other ones, but
3937 is much less sensible to the input data set or to the number of
3938 servers. It can make sense as an alternative to sdbm+avalanche or
3939 djb2+avalanche for consistent hashing or when hashing on numeric
3940 data such as a source IP address or a visitor identifier in a URL
3941 parameter.
3942
Willy Tarreau324f07f2015-01-20 19:44:50 +01003943 crc32 this is the most common CRC32 implementation as used in Ethernet,
3944 gzip, PNG, etc. It is slower than the other ones but may provide
3945 a better distribution or less predictable results especially when
3946 used on strings.
3947
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003948 <modifier> indicates an optional method applied after hashing the key :
3949
3950 avalanche This directive indicates that the result from the hash
3951 function above should not be used in its raw form but that
3952 a 4-byte full avalanche hash must be applied first. The
3953 purpose of this step is to mix the resulting bits from the
3954 previous hash in order to avoid any undesired effect when
3955 the input contains some limited values or when the number of
3956 servers is a multiple of one of the hash's components (64
3957 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3958 result less predictable, but it's also not as smooth as when
3959 using the original function. Some testing might be needed
3960 with some workloads. This hash is one of the many proposed
3961 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003962
Bhaskar98634f02013-10-29 23:30:51 -04003963 The default hash type is "map-based" and is recommended for most usages. The
3964 default function is "sdbm", the selection of a function should be based on
3965 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003966
Andrew Rodland17be45e2016-10-25 17:04:12 -04003967 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003968
3969
Willy Tarreau0ba27502007-12-24 16:55:16 +01003970http-check disable-on-404
3971 Enable a maintenance mode upon HTTP/404 response to health-checks
3972 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003973 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003974 Arguments : none
3975
3976 When this option is set, a server which returns an HTTP code 404 will be
3977 excluded from further load-balancing, but will still receive persistent
3978 connections. This provides a very convenient method for Web administrators
3979 to perform a graceful shutdown of their servers. It is also important to note
3980 that a server which is detected as failed while it was in this mode will not
3981 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3982 will immediately be reinserted into the farm. The status on the stats page
3983 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01003984 option only works in conjunction with the "httpchk" option. If this option
3985 is used with "http-check expect", then it has precedence over it so that 404
3986 responses will still be considered as soft-stop.
3987
3988 See also : "option httpchk", "http-check expect"
3989
3990
3991http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003992 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01003993 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02003994 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01003995 Arguments :
3996 <match> is a keyword indicating how to look for a specific pattern in the
3997 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003998 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01003999 exclamation mark ("!") to negate the match. Spaces are allowed
4000 between the exclamation mark and the keyword. See below for more
4001 details on the supported keywords.
4002
4003 <pattern> is the pattern to look for. It may be a string or a regular
4004 expression. If the pattern contains spaces, they must be escaped
4005 with the usual backslash ('\').
4006
4007 By default, "option httpchk" considers that response statuses 2xx and 3xx
4008 are valid, and that others are invalid. When "http-check expect" is used,
4009 it defines what is considered valid or invalid. Only one "http-check"
4010 statement is supported in a backend. If a server fails to respond or times
4011 out, the check obviously fails. The available matches are :
4012
4013 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004014 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004015 response's status code is exactly this string. If the
4016 "status" keyword is prefixed with "!", then the response
4017 will be considered invalid if the status code matches.
4018
4019 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004020 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004021 response's status code matches the expression. If the
4022 "rstatus" keyword is prefixed with "!", then the response
4023 will be considered invalid if the status code matches.
4024 This is mostly used to check for multiple codes.
4025
4026 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004027 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004028 response's body contains this exact string. If the
4029 "string" keyword is prefixed with "!", then the response
4030 will be considered invalid if the body contains this
4031 string. This can be used to look for a mandatory word at
4032 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004033 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01004034 trace).
4035
4036 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004037 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004038 response's body matches this expression. If the "rstring"
4039 keyword is prefixed with "!", then the response will be
4040 considered invalid if the body matches the expression.
4041 This can be used to look for a mandatory word at the end
4042 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01004043 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01004044
4045 It is important to note that the responses will be limited to a certain size
4046 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
4047 Thus, too large responses may not contain the mandatory pattern when using
4048 "string" or "rstring". If a large response is absolutely required, it is
4049 possible to change the default max size by setting the global variable.
4050 However, it is worth keeping in mind that parsing very large responses can
4051 waste some CPU cycles, especially when regular expressions are used, and that
4052 it is always better to focus the checks on smaller resources.
4053
Cyril Bonté32602d22015-01-30 00:07:07 +01004054 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
4055 will automatically append a "Connection: close" header, meaning that this
4056 header should not be present in the request provided by "option httpchk".
4057
Willy Tarreaubd741542010-03-16 18:46:54 +01004058 Last, if "http-check expect" is combined with "http-check disable-on-404",
4059 then this last one has precedence when the server responds with 404.
4060
4061 Examples :
4062 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004063 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01004064
4065 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004066 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01004067
4068 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004069 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01004070
4071 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004072 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01004073
Willy Tarreaubd741542010-03-16 18:46:54 +01004074 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004075
4076
Willy Tarreauef781042010-01-27 11:53:01 +01004077http-check send-state
4078 Enable emission of a state header with HTTP health checks
4079 May be used in sections : defaults | frontend | listen | backend
4080 yes | no | yes | yes
4081 Arguments : none
4082
4083 When this option is set, haproxy will systematically send a special header
4084 "X-Haproxy-Server-State" with a list of parameters indicating to each server
4085 how they are seen by haproxy. This can be used for instance when a server is
4086 manipulated without access to haproxy and the operator needs to know whether
4087 haproxy still sees it up or not, or if the server is the last one in a farm.
4088
4089 The header is composed of fields delimited by semi-colons, the first of which
4090 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
4091 checks on the total number before transition, just as appears in the stats
4092 interface. Next headers are in the form "<variable>=<value>", indicating in
4093 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08004094 - a variable "address", containing the address of the backend server.
4095 This corresponds to the <address> field in the server declaration. For
4096 unix domain sockets, it will read "unix".
4097
4098 - a variable "port", containing the port of the backend server. This
4099 corresponds to the <port> field in the server declaration. For unix
4100 domain sockets, it will read "unix".
4101
Willy Tarreauef781042010-01-27 11:53:01 +01004102 - a variable "name", containing the name of the backend followed by a slash
4103 ("/") then the name of the server. This can be used when a server is
4104 checked in multiple backends.
4105
4106 - a variable "node" containing the name of the haproxy node, as set in the
4107 global "node" variable, otherwise the system's hostname if unspecified.
4108
4109 - a variable "weight" indicating the weight of the server, a slash ("/")
4110 and the total weight of the farm (just counting usable servers). This
4111 helps to know if other servers are available to handle the load when this
4112 one fails.
4113
4114 - a variable "scur" indicating the current number of concurrent connections
4115 on the server, followed by a slash ("/") then the total number of
4116 connections on all servers of the same backend.
4117
4118 - a variable "qcur" indicating the current number of requests in the
4119 server's queue.
4120
4121 Example of a header received by the application server :
4122 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4123 scur=13/22; qcur=0
4124
4125 See also : "option httpchk", "http-check disable-on-404"
4126
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004127
4128http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004129 Access control for Layer 7 requests
4130
4131 May be used in sections: defaults | frontend | listen | backend
4132 no | yes | yes | yes
4133
Willy Tarreau20b0de52012-12-24 15:45:22 +01004134 The http-request statement defines a set of rules which apply to layer 7
4135 processing. The rules are evaluated in their declaration order when they are
4136 met in a frontend, listen or backend section. Any rule may optionally be
4137 followed by an ACL-based condition, in which case it will only be evaluated
4138 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004139
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004140 The first keyword is the rule's action. The supported actions are described
4141 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004142
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004143 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004144
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004145 Example:
4146 acl nagios src 192.168.129.3
4147 acl local_net src 192.168.0.0/16
4148 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004149
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004150 http-request allow if nagios
4151 http-request allow if local_net auth_ok
4152 http-request auth realm Gimme if local_net auth_ok
4153 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004154
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004155 Example:
4156 acl key req.hdr(X-Add-Acl-Key) -m found
4157 acl add path /addacl
4158 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004159
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004160 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004161
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004162 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4163 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004164
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004165 Example:
4166 acl value req.hdr(X-Value) -m found
4167 acl setmap path /setmap
4168 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004169
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004170 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004171
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004172 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4173 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004174
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004175 See also : "stats http-request", section 3.4 about userlists and section 7
4176 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004177
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004178http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004179
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004180 This is used to add a new entry into an ACL. The ACL must be loaded from a
4181 file (even a dummy empty file). The file name of the ACL to be updated is
4182 passed between parentheses. It takes one argument: <key fmt>, which follows
4183 log-format rules, to collect content of the new entry. It performs a lookup
4184 in the ACL before insertion, to avoid duplicated (or more) values. This
4185 lookup is done by a linear search and can be expensive with large lists!
4186 It is the equivalent of the "add acl" command from the stats socket, but can
4187 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004188
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004189http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004190
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004191 This appends an HTTP header field whose name is specified in <name> and
4192 whose value is defined by <fmt> which follows the log-format rules (see
4193 Custom Log Format in section 8.2.4). This is particularly useful to pass
4194 connection-specific information to the server (e.g. the client's SSL
4195 certificate), or to combine several headers into one. This rule is not
4196 final, so it is possible to add other similar rules. Note that header
4197 addition is performed immediately, so one rule might reuse the resulting
4198 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004199
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004200http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004201
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004202 This stops the evaluation of the rules and lets the request pass the check.
4203 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004204
Sasha Pachev218f0642014-06-16 12:05:59 -06004205
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004206http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004207
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004208 This stops the evaluation of the rules and immediately responds with an
4209 HTTP 401 or 407 error code to invite the user to present a valid user name
4210 and password. No further "http-request" rules are evaluated. An optional
4211 "realm" parameter is supported, it sets the authentication realm that is
4212 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004213
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004214 Example:
4215 acl auth_ok http_auth_group(L1) G1
4216 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004217
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004218http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004219
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004220 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004221
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004222http-request capture <sample> [ len <length> | id <id> ]
4223 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004224
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004225 This captures sample expression <sample> from the request buffer, and
4226 converts it to a string of at most <len> characters. The resulting string is
4227 stored into the next request "capture" slot, so it will possibly appear next
4228 to some captured HTTP headers. It will then automatically appear in the logs,
4229 and it will be possible to extract it using sample fetch rules to feed it
4230 into headers or anything. The length should be limited given that this size
4231 will be allocated for each capture during the whole session life.
4232 Please check section 7.3 (Fetching samples) and "capture request header" for
4233 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004234
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004235 If the keyword "id" is used instead of "len", the action tries to store the
4236 captured string in a previously declared capture slot. This is useful to run
4237 captures in backends. The slot id can be declared by a previous directive
4238 "http-request capture" or with the "declare capture" keyword. If the slot
4239 <id> doesn't exist, then HAProxy fails parsing the configuration to prevent
4240 unexpected behavior at run time.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004241
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004242http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004243
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004244 This is used to delete an entry from an ACL. The ACL must be loaded from a
4245 file (even a dummy empty file). The file name of the ACL to be updated is
4246 passed between parentheses. It takes one argument: <key fmt>, which follows
4247 log-format rules, to collect content of the entry to delete.
4248 It is the equivalent of the "del acl" command from the stats socket, but can
4249 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004250
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004251http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004252
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004253 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004254
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004255http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004256
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004257 This is used to delete an entry from a MAP. The MAP must be loaded from a
4258 file (even a dummy empty file). The file name of the MAP to be updated is
4259 passed between parentheses. It takes one argument: <key fmt>, which follows
4260 log-format rules, to collect content of the entry to delete.
4261 It takes one argument: "file name" It is the equivalent of the "del map"
4262 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004263
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004264http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004265
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004266 This stops the evaluation of the rules and immediately rejects the request
4267 and emits an HTTP 403 error, or optionally the status code specified as an
4268 argument to "deny_status". The list of permitted status codes is limited to
4269 those that can be overridden by the "errorfile" directive.
4270 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004271
Olivier Houchard602bf7d2019-05-10 13:59:15 +02004272http-request disable-l7-retry [ { if | unless } <condition> ]
4273 This disables any attempt to retry the request if it fails for any other
4274 reason than a connection failure. This can be useful for example to make
4275 sure POST requests aren't retried on failure.
4276
Baptiste Assmann333939c2019-01-21 08:34:50 +01004277http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
4278
4279 This action performs a DNS resolution of the output of <expr> and stores
4280 the result in the variable <var>. It uses the DNS resolvers section
4281 pointed by <resolvers>.
4282 It is possible to choose a resolution preference using the optional
4283 arguments 'ipv4' or 'ipv6'.
4284 When performing the DNS resolution, the client side connection is on
4285 pause waiting till the end of the resolution.
4286 If an IP address can be found, it is stored into <var>. If any kind of
4287 error occurs, then <var> is not set.
4288 One can use this action to discover a server IP address at run time and
4289 based on information found in the request (IE a Host header).
4290 If this action is used to find the server's IP address (using the
4291 "set-dst" action), then the server IP address in the backend must be set
4292 to 0.0.0.0.
4293
4294 Example:
4295 resolvers mydns
4296 nameserver local 127.0.0.53:53
4297 nameserver google 8.8.8.8:53
4298 timeout retry 1s
4299 hold valid 10s
4300 hold nx 3s
4301 hold other 3s
4302 hold obsolete 0s
4303 accepted_payload_size 8192
4304
4305 frontend fe
4306 bind 10.42.0.1:80
4307 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
4308 http-request capture var(txn.myip) len 40
4309
4310 # return 503 when the variable is not set,
4311 # which mean DNS resolution error
4312 use_backend b_503 unless { var(txn.myip) -m found }
4313
4314 default_backend be
4315
4316 backend b_503
4317 # dummy backend used to return 503.
4318 # one can use the errorfile directive to send a nice
4319 # 503 error page to end users
4320
4321 backend be
4322 # rule to prevent HAProxy from reconnecting to services
4323 # on the local network (forged DNS name used to scan the network)
4324 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
4325 http-request set-dst var(txn.myip)
4326 server clear 0.0.0.0:0
4327
4328 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
4329 be used to scan the network or worst won't loop over itself...
4330
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004331http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4332
4333 This is used to build an HTTP 103 Early Hints response prior to any other one.
4334 This appends an HTTP header field to this response whose name is specified in
4335 <name> and whose value is defined by <fmt> which follows the log-format rules
4336 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004337 to the client some Link headers to preload resources required to render the
4338 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004339
4340 See RFC 8297 for more information.
4341
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004342http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004343
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004344 This performs an HTTP redirection based on a redirect rule. This is exactly
4345 the same as the "redirect" statement except that it inserts a redirect rule
4346 which can be processed in the middle of other "http-request" rules and that
4347 these rules use the "log-format" strings. See the "redirect" keyword for the
4348 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004349
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004350http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004351
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004352 This stops the evaluation of the rules and immediately closes the connection
4353 without sending any response. It acts similarly to the
4354 "tcp-request content reject" rules. It can be useful to force an immediate
4355 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004356
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004357http-request replace-header <name> <match-regex> <replace-fmt>
4358 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004359
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004360 This matches the regular expression in all occurrences of header field
4361 <name> according to <match-regex>, and replaces them with the
4362 <replace-fmt> argument. Format characters are allowed in replace-fmt and
4363 work like in <fmt> arguments in "http-request add-header". The match is
4364 only case-sensitive. It is important to understand that this action only
4365 considers whole header lines, regardless of the number of values they may
4366 contain. This usage is suited to headers naturally containing commas in
4367 their value, such as If-Modified-Since and so on.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004368
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004369 Example:
4370 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
William Lallemand86d0df02017-11-24 21:36:45 +01004371
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004372 # applied to:
4373 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004374
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004375 # outputs:
4376 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004377
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004378 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004379
Willy Tarreau33810222019-06-12 17:44:02 +02004380http-request replace-uri <match-regex> <replace-fmt>
4381 [ { if | unless } <condition> ]
4382
4383 This matches the regular expression in the URI part of the request
4384 according to <match-regex>, and replaces it with the <replace-fmt>
4385 argument. Standard back-references using the backslash ('\') followed by a
4386 number are supported. The <fmt> field is interpreted as a log-format string
4387 so it may contain special expressions just like the <fmt> argument passed
4388 to "http-request set-uri". The match is exclusively case-sensitive. Any
4389 optional scheme, authority or query string are considered in the matching
4390 part of the URI. It is worth noting that regular expressions may be more
4391 expensive to evaluate than certain ACLs, so rare replacements may benefit
4392 from a condition to avoid performing the evaluation at all if it does not
4393 match.
4394
4395 Example:
4396 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4397 http-request replace-uri (.*) /foo\1
4398
4399 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
4400 http-request replace-uri ([^?]*)(\?(.*))? \1/foo\2
4401
4402 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
4403 http-request replace-uri /foo/(.*) /\1
4404 # or more efficient if only some requests match :
4405 http-request replace-uri /foo/(.*) /\1 if { url_beg /foo/ }
4406
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004407http-request replace-value <name> <match-regex> <replace-fmt>
4408 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004409
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004410 This works like "replace-header" except that it matches the regex against
4411 every comma-delimited value of the header field <name> instead of the
4412 entire header. This is suited for all headers which are allowed to carry
4413 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004414
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004415 Example:
4416 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004417
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004418 # applied to:
4419 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004420
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004421 # outputs:
4422 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004423
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004424http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4425http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004426
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004427 This actions increments the GPC0 or GPC1 counter according with the sticky
4428 counter designated by <sc-id>. If an error occurs, this action silently fails
4429 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004430
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004431http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004432
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004433 This action sets the GPT0 tag according to the sticky counter designated by
4434 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4435 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004436
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004437http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004438
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004439 This is used to set the destination IP address to the value of specified
4440 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4441 but provides the correct IP in a HTTP header; or you want to mask the IP for
4442 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4443 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004444
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004445 Arguments:
4446 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4447 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004448
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004449 Example:
4450 http-request set-dst hdr(x-dst)
4451 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004452
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004453 When possible, set-dst preserves the original destination port as long as the
4454 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004455
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004456http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004457
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004458 This is used to set the destination port address to the value of specified
4459 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4460 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004461
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004462 Arguments:
4463 <expr> Is a standard HAProxy expression formed by a sample-fetch
4464 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004465
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004466 Example:
4467 http-request set-dst-port hdr(x-port)
4468 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004469
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004470 When possible, set-dst-port preserves the original destination address as
4471 long as the address family supports a port, otherwise it forces the
4472 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004473
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004474http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004475
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004476 This does the same as "http-request add-header" except that the header name
4477 is first removed if it existed. This is useful when passing security
4478 information to the server, where the header must not be manipulated by
4479 external users. Note that the new value is computed before the removal so it
4480 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004481
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004482 Example:
4483 http-request set-header X-Haproxy-Current-Date %T
4484 http-request set-header X-SSL %[ssl_fc]
4485 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4486 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4487 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4488 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4489 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4490 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4491 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004492
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004493http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004494
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004495 This is used to change the log level of the current request when a certain
4496 condition is met. Valid levels are the 8 syslog levels (see the "log"
4497 keyword) plus the special level "silent" which disables logging for this
4498 request. This rule is not final so the last matching rule wins. This rule
4499 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004500
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004501http-request set-map(<file-name>) <key fmt> <value fmt>
4502 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004503
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004504 This is used to add a new entry into a MAP. The MAP must be loaded from a
4505 file (even a dummy empty file). The file name of the MAP to be updated is
4506 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4507 log-format rules, used to collect MAP key, and <value fmt>, which follows
4508 log-format rules, used to collect content for the new entry.
4509 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4510 more) values. This lookup is done by a linear search and can be expensive
4511 with large lists! It is the equivalent of the "set map" command from the
4512 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004513
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004514http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004515
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004516 This is used to set the Netfilter MARK on all packets sent to the client to
4517 the value passed in <mark> on platforms which support it. This value is an
4518 unsigned 32 bit value which can be matched by netfilter and by the routing
4519 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4520 "0x"). This can be useful to force certain packets to take a different route
4521 (for example a cheaper network path for bulk downloads). This works on Linux
4522 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004523
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004524http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004525
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004526 This rewrites the request method with the result of the evaluation of format
4527 string <fmt>. There should be very few valid reasons for having to do so as
4528 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004529
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004530http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004531
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004532 This sets the "nice" factor of the current request being processed. It only
4533 has effect against the other requests being processed at the same time.
4534 The default value is 0, unless altered by the "nice" setting on the "bind"
4535 line. The accepted range is -1024..1024. The higher the value, the nicest
4536 the request will be. Lower values will make the request more important than
4537 other ones. This can be useful to improve the speed of some requests, or
4538 lower the priority of non-important requests. Using this setting without
4539 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004540
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004541http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004542
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004543 This rewrites the request path with the result of the evaluation of format
4544 string <fmt>. The query string, if any, is left intact. If a scheme and
4545 authority is found before the path, they are left intact as well. If the
4546 request doesn't have a path ("*"), this one is replaced with the format.
4547 This can be used to prepend a directory component in front of a path for
4548 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004549
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004550 Example :
4551 # prepend the host name before the path
4552 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004553
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004554http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004555
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004556 This is used to set the queue priority class of the current request.
4557 The value must be a sample expression which converts to an integer in the
4558 range -2047..2047. Results outside this range will be truncated.
4559 The priority class determines the order in which queued requests are
4560 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004561
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004562http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004563
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004564 This is used to set the queue priority timestamp offset of the current
4565 request. The value must be a sample expression which converts to an integer
4566 in the range -524287..524287. Results outside this range will be truncated.
4567 When a request is queued, it is ordered first by the priority class, then by
4568 the current timestamp adjusted by the given offset in milliseconds. Lower
4569 values have higher priority.
4570 Note that the resulting timestamp is is only tracked with enough precision
4571 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4572 adjusted timestamp exceeds this value, it will be misidentified as highest
4573 priority. Thus it is important to set "timeout queue" to a value, where when
4574 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004575
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004576http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004577
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004578 This rewrites the request's query string which appears after the first
4579 question mark ("?") with the result of the evaluation of format string <fmt>.
4580 The part prior to the question mark is left intact. If the request doesn't
4581 contain a question mark and the new value is not empty, then one is added at
4582 the end of the URI, followed by the new value. If a question mark was
4583 present, it will never be removed even if the value is empty. This can be
4584 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004585
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004586 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004587
4588 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004589 # replace "%3D" with "=" in the query string
4590 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004591
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004592http-request set-src <expr> [ { if | unless } <condition> ]
4593 This is used to set the source IP address to the value of specified
4594 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4595 provides the correct IP in a HTTP header; or you want to mask source IP for
4596 privacy.
4597
4598 Arguments :
4599 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4600 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004601
Cyril Bonté78caf842010-03-10 22:41:43 +01004602 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004603 http-request set-src hdr(x-forwarded-for)
4604 http-request set-src src,ipmask(24)
4605
4606 When possible, set-src preserves the original source port as long as the
4607 address family allows it, otherwise the source port is set to 0.
4608
4609http-request set-src-port <expr> [ { if | unless } <condition> ]
4610
4611 This is used to set the source port address to the value of specified
4612 expression.
4613
4614 Arguments:
4615 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4616 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004617
Willy Tarreau20b0de52012-12-24 15:45:22 +01004618 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004619 http-request set-src-port hdr(x-port)
4620 http-request set-src-port int(4000)
4621
4622 When possible, set-src-port preserves the original source address as long as
4623 the address family supports a port, otherwise it forces the source address to
4624 IPv4 "0.0.0.0" before rewriting the port.
4625
4626http-request set-tos <tos> [ { if | unless } <condition> ]
4627
4628 This is used to set the TOS or DSCP field value of packets sent to the client
4629 to the value passed in <tos> on platforms which support this. This value
4630 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4631 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4632 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4633 be used to adjust some routing behavior on border routers based on some
4634 information from the request.
4635
4636 See RFC 2474, 2597, 3260 and 4594 for more information.
4637
4638http-request set-uri <fmt> [ { if | unless } <condition> ]
4639
4640 This rewrites the request URI with the result of the evaluation of format
4641 string <fmt>. The scheme, authority, path and query string are all replaced
4642 at once. This can be used to rewrite hosts in front of proxies, or to
4643 perform complex modifications to the URI such as moving parts between the
4644 path and the query string.
4645 See also "http-request set-path" and "http-request set-query".
4646
4647http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4648
4649 This is used to set the contents of a variable. The variable is declared
4650 inline.
4651
4652 Arguments:
4653 <var-name> The name of the variable starts with an indication about its
4654 scope. The scopes allowed are:
4655 "proc" : the variable is shared with the whole process
4656 "sess" : the variable is shared with the whole session
4657 "txn" : the variable is shared with the transaction
4658 (request and response)
4659 "req" : the variable is shared only during request
4660 processing
4661 "res" : the variable is shared only during response
4662 processing
4663 This prefix is followed by a name. The separator is a '.'.
4664 The name may only contain characters 'a-z', 'A-Z', '0-9'
4665 and '_'.
4666
4667 <expr> Is a standard HAProxy expression formed by a sample-fetch
4668 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004669
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004670 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004671 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004672
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004673http-request send-spoe-group <engine-name> <group-name>
4674 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004675
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004676 This action is used to trigger sending of a group of SPOE messages. To do so,
4677 the SPOE engine used to send messages must be defined, as well as the SPOE
4678 group to send. Of course, the SPOE engine must refer to an existing SPOE
4679 filter. If not engine name is provided on the SPOE filter line, the SPOE
4680 agent name must be used.
4681
4682 Arguments:
4683 <engine-name> The SPOE engine name.
4684
4685 <group-name> The SPOE group name as specified in the engine
4686 configuration.
4687
4688http-request silent-drop [ { if | unless } <condition> ]
4689
4690 This stops the evaluation of the rules and makes the client-facing connection
4691 suddenly disappear using a system-dependent way that tries to prevent the
4692 client from being notified. The effect it then that the client still sees an
4693 established connection while there's none on HAProxy. The purpose is to
4694 achieve a comparable effect to "tarpit" except that it doesn't use any local
4695 resource at all on the machine running HAProxy. It can resist much higher
4696 loads than "tarpit", and slow down stronger attackers. It is important to
4697 understand the impact of using this mechanism. All stateful equipment placed
4698 between the client and HAProxy (firewalls, proxies, load balancers) will also
4699 keep the established connection for a long time and may suffer from this
4700 action.
4701 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4702 option is used to block the emission of a TCP reset. On other systems, the
4703 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4704 router, though it's still delivered to local networks. Do not use it unless
4705 you fully understand how it works.
4706
4707http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4708
4709 This stops the evaluation of the rules and immediately blocks the request
4710 without responding for a delay specified by "timeout tarpit" or
4711 "timeout connect" if the former is not set. After that delay, if the client
4712 is still connected, an HTTP error 500 (or optionally the status code
4713 specified as an argument to "deny_status") is returned so that the client
4714 does not suspect it has been tarpitted. Logs will report the flags "PT".
4715 The goal of the tarpit rule is to slow down robots during an attack when
4716 they're limited on the number of concurrent requests. It can be very
4717 efficient against very dumb robots, and will significantly reduce the load
4718 on firewalls compared to a "deny" rule. But when facing "correctly"
4719 developed robots, it can make things worse by forcing haproxy and the front
4720 firewall to support insane number of concurrent connections.
4721 See also the "silent-drop" action.
4722
4723http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4724http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4725http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4726
4727 This enables tracking of sticky counters from current request. These rules do
4728 not stop evaluation and do not change default action. The number of counters
4729 that may be simultaneously tracked by the same connection is set in
4730 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4731 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4732 "track-sc0" rule executed enables tracking of the counters of the specified
4733 table as the first set. The first "track-sc1" rule executed enables tracking
4734 of the counters of the specified table as the second set. The first
4735 "track-sc2" rule executed enables tracking of the counters of the specified
4736 table as the third set. It is a recommended practice to use the first set of
4737 counters for the per-frontend counters and the second set for the per-backend
4738 ones. But this is just a guideline, all may be used everywhere.
4739
4740 Arguments :
4741 <key> is mandatory, and is a sample expression rule as described in
4742 section 7.3. It describes what elements of the incoming request or
4743 connection will be analyzed, extracted, combined, and used to
4744 select which table entry to update the counters.
4745
4746 <table> is an optional table to be used instead of the default one, which
4747 is the stick-table declared in the current proxy. All the counters
4748 for the matches and updates for the key will then be performed in
4749 that table until the session ends.
4750
4751 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4752 it is not found, an entry is allocated for it. Then a pointer to that entry
4753 is kept during all the session's life, and this entry's counters are updated
4754 as often as possible, every time the session's counters are updated, and also
4755 systematically when the session ends. Counters are only updated for events
4756 that happen after the tracking has been started. As an exception, connection
4757 counters and request counters are systematically updated so that they reflect
4758 useful information.
4759
4760 If the entry tracks concurrent connection counters, one connection is counted
4761 for as long as the entry is tracked, and the entry will not expire during
4762 that time. Tracking counters also provides a performance advantage over just
4763 checking the keys, because only one table lookup is performed for all ACL
4764 checks that make use of it.
4765
4766http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4767
4768 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004769
4770 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004771 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004772
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004773http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004774
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004775 This will delay the processing of the request until the SSL handshake
4776 happened. This is mostly useful to delay processing early data until we're
4777 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004778
Willy Tarreauef781042010-01-27 11:53:01 +01004779
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004780http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004781 Access control for Layer 7 responses
4782
4783 May be used in sections: defaults | frontend | listen | backend
4784 no | yes | yes | yes
4785
4786 The http-response statement defines a set of rules which apply to layer 7
4787 processing. The rules are evaluated in their declaration order when they are
4788 met in a frontend, listen or backend section. Any rule may optionally be
4789 followed by an ACL-based condition, in which case it will only be evaluated
4790 if the condition is true. Since these rules apply on responses, the backend
4791 rules are applied first, followed by the frontend's rules.
4792
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004793 The first keyword is the rule's action. The supported actions are described
4794 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004795
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004796 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004797
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004798 Example:
4799 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004800
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004801 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06004802
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004803 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4804 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06004805
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004806 Example:
4807 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06004808
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004809 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004810
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004811 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4812 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06004813
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004814 See also : "http-request", section 3.4 about userlists and section 7 about
4815 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004816
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004817http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004818
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004819 This is used to add a new entry into an ACL. The ACL must be loaded from a
4820 file (even a dummy empty file). The file name of the ACL to be updated is
4821 passed between parentheses. It takes one argument: <key fmt>, which follows
4822 log-format rules, to collect content of the new entry. It performs a lookup
4823 in the ACL before insertion, to avoid duplicated (or more) values.
4824 This lookup is done by a linear search and can be expensive with large lists!
4825 It is the equivalent of the "add acl" command from the stats socket, but can
4826 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06004827
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004828http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004829
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004830 This appends an HTTP header field whose name is specified in <name> and whose
4831 value is defined by <fmt> which follows the log-format rules (see Custom Log
4832 Format in section 8.2.4). This may be used to send a cookie to a client for
4833 example, or to pass some internal information.
4834 This rule is not final, so it is possible to add other similar rules.
4835 Note that header addition is performed immediately, so one rule might reuse
4836 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004837
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004838http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004839
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004840 This stops the evaluation of the rules and lets the response pass the check.
4841 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06004842
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004843http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004844
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004845 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06004846
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004847http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004848
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004849 This captures sample expression <sample> from the response buffer, and
4850 converts it to a string. The resulting string is stored into the next request
4851 "capture" slot, so it will possibly appear next to some captured HTTP
4852 headers. It will then automatically appear in the logs, and it will be
4853 possible to extract it using sample fetch rules to feed it into headers or
4854 anything. Please check section 7.3 (Fetching samples) and
4855 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004856
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004857 The keyword "id" is the id of the capture slot which is used for storing the
4858 string. The capture slot must be defined in an associated frontend.
4859 This is useful to run captures in backends. The slot id can be declared by a
4860 previous directive "http-response capture" or with the "declare capture"
4861 keyword.
4862 If the slot <id> doesn't exist, then HAProxy fails parsing the configuration
4863 to prevent unexpected behavior at run time.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004864
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004865http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004866
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004867 This is used to delete an entry from an ACL. The ACL must be loaded from a
4868 file (even a dummy empty file). The file name of the ACL to be updated is
4869 passed between parentheses. It takes one argument: <key fmt>, which follows
4870 log-format rules, to collect content of the entry to delete.
4871 It is the equivalent of the "del acl" command from the stats socket, but can
4872 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004873
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004874http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004875
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004876 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004877
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004878http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02004879
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004880 This is used to delete an entry from a MAP. The MAP must be loaded from a
4881 file (even a dummy empty file). The file name of the MAP to be updated is
4882 passed between parentheses. It takes one argument: <key fmt>, which follows
4883 log-format rules, to collect content of the entry to delete.
4884 It takes one argument: "file name" It is the equivalent of the "del map"
4885 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004886
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004887http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004888
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004889 This stops the evaluation of the rules and immediately rejects the response
4890 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004891
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004892http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004893
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004894 This performs an HTTP redirection based on a redirect rule.
4895 This supports a format string similarly to "http-request redirect" rules,
4896 with the exception that only the "location" type of redirect is possible on
4897 the response. See the "redirect" keyword for the rule's syntax. When a
4898 redirect rule is applied during a response, connections to the server are
4899 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004900
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004901http-response replace-header <name> <regex-match> <replace-fmt>
4902 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004903
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004904 This matches the regular expression in all occurrences of header field <name>
4905 according to <match-regex>, and replaces them with the <replace-fmt> argument.
4906 Format characters are allowed in replace-fmt and work like in <fmt> arguments
4907 in "add-header". The match is only case-sensitive. It is important to
4908 understand that this action only considers whole header lines, regardless of
4909 the number of values they may contain. This usage is suited to headers
4910 naturally containing commas in their value, such as Set-Cookie, Expires and
4911 so on.
William Lallemand86d0df02017-11-24 21:36:45 +01004912
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004913 Example:
4914 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02004915
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004916 # applied to:
4917 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004918
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004919 # outputs:
4920 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 +02004921
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004922 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004923
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004924http-response replace-value <name> <regex-match> <replace-fmt>
4925 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004926
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004927 This works like "replace-header" except that it matches the regex against
4928 every comma-delimited value of the header field <name> instead of the entire
4929 header. This is suited for all headers which are allowed to carry more than
4930 one value. An example could be the Accept header.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004931
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004932 Example:
4933 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004934
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004935 # applied to:
4936 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01004937
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004938 # outputs:
4939 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004940
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004941http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4942http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004943
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004944 This action increments the GPC0 or GPC1 counter according with the sticky
4945 counter designated by <sc-id>. If an error occurs, this action silently fails
4946 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004947
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004948http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004949
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004950 This action sets the GPT0 tag according to the sticky counter designated by
4951 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4952 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004953
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004954http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02004955
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004956 This action is used to trigger sending of a group of SPOE messages. To do so,
4957 the SPOE engine used to send messages must be defined, as well as the SPOE
4958 group to send. Of course, the SPOE engine must refer to an existing SPOE
4959 filter. If not engine name is provided on the SPOE filter line, the SPOE
4960 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004961
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004962 Arguments:
4963 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004964
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004965 <group-name> The SPOE group name as specified in the engine
4966 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004967
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004968http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004969
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004970 This does the same as "add-header" except that the header name is first
4971 removed if it existed. This is useful when passing security information to
4972 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004973
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004974http-response set-log-level <level> [ { if | unless } <condition> ]
4975
4976 This is used to change the log level of the current request when a certain
4977 condition is met. Valid levels are the 8 syslog levels (see the "log"
4978 keyword) plus the special level "silent" which disables logging for this
4979 request. This rule is not final so the last matching rule wins. This rule can
4980 be useful to disable health checks coming from another equipment.
4981
4982http-response set-map(<file-name>) <key fmt> <value fmt>
4983
4984 This is used to add a new entry into a MAP. The MAP must be loaded from a
4985 file (even a dummy empty file). The file name of the MAP to be updated is
4986 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4987 log-format rules, used to collect MAP key, and <value fmt>, which follows
4988 log-format rules, used to collect content for the new entry. It performs a
4989 lookup in the MAP before insertion, to avoid duplicated (or more) values.
4990 This lookup is done by a linear search and can be expensive with large lists!
4991 It is the equivalent of the "set map" command from the stats socket, but can
4992 be triggered by an HTTP response.
4993
4994http-response set-mark <mark> [ { if | unless } <condition> ]
4995
4996 This is used to set the Netfilter MARK on all packets sent to the client to
4997 the value passed in <mark> on platforms which support it. This value is an
4998 unsigned 32 bit value which can be matched by netfilter and by the routing
4999 table. It can be expressed both in decimal or hexadecimal format (prefixed
5000 by "0x"). This can be useful to force certain packets to take a different
5001 route (for example a cheaper network path for bulk downloads). This works on
5002 Linux kernels 2.6.32 and above and requires admin privileges.
5003
5004http-response set-nice <nice> [ { if | unless } <condition> ]
5005
5006 This sets the "nice" factor of the current request being processed.
5007 It only has effect against the other requests being processed at the same
5008 time. The default value is 0, unless altered by the "nice" setting on the
5009 "bind" line. The accepted range is -1024..1024. The higher the value, the
5010 nicest the request will be. Lower values will make the request more important
5011 than other ones. This can be useful to improve the speed of some requests, or
5012 lower the priority of non-important requests. Using this setting without
5013 prior experimentation can cause some major slowdown.
5014
5015http-response set-status <status> [reason <str>]
5016 [ { if | unless } <condition> ]
5017
5018 This replaces the response status code with <status> which must be an integer
5019 between 100 and 999. Optionally, a custom reason text can be provided defined
5020 by <str>, or the default reason for the specified code will be used as a
5021 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005022
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005023 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005024 # return "431 Request Header Fields Too Large"
5025 http-response set-status 431
5026 # return "503 Slow Down", custom reason
5027 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005028
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005029http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005030
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005031 This is used to set the TOS or DSCP field value of packets sent to the client
5032 to the value passed in <tos> on platforms which support this.
5033 This value represents the whole 8 bits of the IP TOS field, and can be
5034 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
5035 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
5036 always 0. This can be used to adjust some routing behavior on border routers
5037 based on some information from the request.
5038
5039 See RFC 2474, 2597, 3260 and 4594 for more information.
5040
5041http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5042
5043 This is used to set the contents of a variable. The variable is declared
5044 inline.
5045
5046 Arguments:
5047 <var-name> The name of the variable starts with an indication about its
5048 scope. The scopes allowed are:
5049 "proc" : the variable is shared with the whole process
5050 "sess" : the variable is shared with the whole session
5051 "txn" : the variable is shared with the transaction
5052 (request and response)
5053 "req" : the variable is shared only during request
5054 processing
5055 "res" : the variable is shared only during response
5056 processing
5057 This prefix is followed by a name. The separator is a '.'.
5058 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5059 and '_'.
5060
5061 <expr> Is a standard HAProxy expression formed by a sample-fetch
5062 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005063
5064 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005065 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005066
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005067http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005068
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005069 This stops the evaluation of the rules and makes the client-facing connection
5070 suddenly disappear using a system-dependent way that tries to prevent the
5071 client from being notified. The effect it then that the client still sees an
5072 established connection while there's none on HAProxy. The purpose is to
5073 achieve a comparable effect to "tarpit" except that it doesn't use any local
5074 resource at all on the machine running HAProxy. It can resist much higher
5075 loads than "tarpit", and slow down stronger attackers. It is important to
5076 understand the impact of using this mechanism. All stateful equipment placed
5077 between the client and HAProxy (firewalls, proxies, load balancers) will also
5078 keep the established connection for a long time and may suffer from this
5079 action.
5080 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
5081 option is used to block the emission of a TCP reset. On other systems, the
5082 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
5083 router, though it's still delivered to local networks. Do not use it unless
5084 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005085
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005086http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
5087http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
5088http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005089
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005090 This enables tracking of sticky counters from current response. Please refer
5091 to "http-request track-sc" for a complete description. The only difference
5092 from "http-request track-sc" is the <key> sample expression can only make use
5093 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
5094 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
5095 supported, haproxy will fail and warn while parsing the config.
5096
5097http-response unset-var(<var-name>) [ { if | unless } <condition> ]
5098
5099 This is used to unset a variable. See "http-response set-var" for details
5100 about <var-name>.
5101
5102 Example:
5103 http-response unset-var(sess.last_redir)
5104
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02005105
Willy Tarreau30631952015-08-06 15:05:24 +02005106http-reuse { never | safe | aggressive | always }
5107 Declare how idle HTTP connections may be shared between requests
5108
5109 May be used in sections: defaults | frontend | listen | backend
5110 yes | no | yes | yes
5111
5112 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01005113 which is considered safe for reuse is moved back to the server's idle
5114 connections pool so that any other request can make use of it. This is the
5115 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02005116
5117 The argument indicates the desired connection reuse strategy :
5118
Olivier Houchard86006a52018-12-14 19:37:49 +01005119 - "never" : idle connections are never shared between sessions. This mode
5120 may be enforced to cancel a different strategy inherited from
5121 a defaults section or for troubleshooting. For example, if an
5122 old bogus application considers that multiple requests over
5123 the same connection come from the same client and it is not
5124 possible to fix the application, it may be desirable to
5125 disable connection sharing in a single backend. An example of
5126 such an application could be an old haproxy using cookie
5127 insertion in tunnel mode and not checking any request past the
5128 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02005129
Olivier Houchard86006a52018-12-14 19:37:49 +01005130 - "safe" : this is the default and the recommended strategy. The first
5131 request of a session is always sent over its own connection,
5132 and only subsequent requests may be dispatched over other
5133 existing connections. This ensures that in case the server
5134 closes the connection when the request is being sent, the
5135 browser can decide to silently retry it. Since it is exactly
5136 equivalent to regular keep-alive, there should be no side
5137 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02005138
5139 - "aggressive" : this mode may be useful in webservices environments where
5140 all servers are not necessarily known and where it would be
5141 appreciable to deliver most first requests over existing
5142 connections. In this case, first requests are only delivered
5143 over existing connections that have been reused at least once,
5144 proving that the server correctly supports connection reuse.
5145 It should only be used when it's sure that the client can
5146 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02005147 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02005148 downsides of rare connection failures.
5149
5150 - "always" : this mode is only recommended when the path to the server is
5151 known for never breaking existing connections quickly after
5152 releasing them. It allows the first request of a session to be
5153 sent to an existing connection. This can provide a significant
5154 performance increase over the "safe" strategy when the backend
5155 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005156 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02005157 sharing. It is recommended that the "http-keep-alive" timeout
5158 remains low in this mode so that no dead connections remain
5159 usable. In most cases, this will lead to the same performance
5160 gains as "aggressive" but with more risks. It should only be
5161 used when it improves the situation over "aggressive".
5162
5163 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005164 connection properties and compatibility. Specifically :
5165 - connections made with "usesrc" followed by a client-dependent value
5166 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005167
5168 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01005169 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005170
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02005171 - connections with certain bogus authentication schemes (relying on the
5172 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005173
5174 No connection pool is involved, once a session dies, the last idle connection
5175 it was attached to is deleted at the same time. This ensures that connections
5176 may not last after all sessions are closed.
5177
5178 Note: connection reuse improves the accuracy of the "server maxconn" setting,
5179 because almost no new connection will be established while idle connections
5180 remain available. This is particularly true with the "always" strategy.
5181
5182 See also : "option http-keep-alive", "server maxconn"
5183
5184
Mark Lamourinec2247f02012-01-04 13:02:01 -05005185http-send-name-header [<header>]
5186 Add the server name to a request. Use the header string given by <header>
5187
5188 May be used in sections: defaults | frontend | listen | backend
5189 yes | no | yes | yes
5190
5191 Arguments :
5192
5193 <header> The header string to use to send the server name
5194
5195 The "http-send-name-header" statement causes the name of the target
Davor Ocelice9ed2812017-12-25 17:49:28 +01005196 server to be added to the headers of an HTTP request. The name
Mark Lamourinec2247f02012-01-04 13:02:01 -05005197 is added with the header string proved.
5198
5199 See also : "server"
5200
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005201id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005202 Set a persistent ID to a proxy.
5203 May be used in sections : defaults | frontend | listen | backend
5204 no | yes | yes | yes
5205 Arguments : none
5206
5207 Set a persistent ID for the proxy. This ID must be unique and positive.
5208 An unused ID will automatically be assigned if unset. The first assigned
5209 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005210
5211
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005212ignore-persist { if | unless } <condition>
5213 Declare a condition to ignore persistence
5214 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005215 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005216
5217 By default, when cookie persistence is enabled, every requests containing
5218 the cookie are unconditionally persistent (assuming the target server is up
5219 and running).
5220
5221 The "ignore-persist" statement allows one to declare various ACL-based
5222 conditions which, when met, will cause a request to ignore persistence.
5223 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005224 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005225 persistence for a specific User-Agent (for example, some web crawler bots).
5226
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005227 The persistence is ignored when an "if" condition is met, or unless an
5228 "unless" condition is met.
5229
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005230 Example:
5231 acl url_static path_beg /static /images /img /css
5232 acl url_static path_end .gif .png .jpg .css .js
5233 ignore-persist if url_static
5234
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005235 See also : "force-persist", "cookie", and section 7 about ACL usage.
5236
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005237load-server-state-from-file { global | local | none }
5238 Allow seamless reload of HAProxy
5239 May be used in sections: defaults | frontend | listen | backend
5240 yes | no | yes | yes
5241
5242 This directive points HAProxy to a file where server state from previous
5243 running process has been saved. That way, when starting up, before handling
5244 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005245 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005246 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5247 loading state or load states from a file containing all backends and servers.
5248 The state file can be generated by running the command "show servers state"
5249 over the stats socket and redirect output.
5250
Davor Ocelice9ed2812017-12-25 17:49:28 +01005251 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005252 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005253 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005254
5255 Arguments:
5256 global load the content of the file pointed by the global directive
5257 named "server-state-file".
5258
5259 local load the content of the file pointed by the directive
5260 "server-state-file-name" if set. If not set, then the backend
5261 name is used as a file name.
5262
5263 none don't load any stat for this backend
5264
5265 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005266 - server's IP address is preserved across reloads by default, but the
5267 order can be changed thanks to the server's "init-addr" setting. This
5268 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005269 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005270 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005271
5272 - server's weight is applied from previous running process unless it has
5273 has changed between previous and new configuration files.
5274
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005275 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005276
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005277 global
5278 stats socket /tmp/socket
5279 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005280
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005281 defaults
5282 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005283
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005284 backend bk
5285 server s1 127.0.0.1:22 check weight 11
5286 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005287
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005288
5289 Then one can run :
5290
5291 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5292
5293 Content of the file /tmp/server_state would be like this:
5294
5295 1
5296 # <field names skipped for the doc example>
5297 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5298 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5299
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005300 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005301
5302 global
5303 stats socket /tmp/socket
5304 server-state-base /etc/haproxy/states
5305
5306 defaults
5307 load-server-state-from-file local
5308
5309 backend bk
5310 server s1 127.0.0.1:22 check weight 11
5311 server s2 127.0.0.1:22 check weight 12
5312
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005313
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005314 Then one can run :
5315
5316 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5317
5318 Content of the file /etc/haproxy/states/bk would be like this:
5319
5320 1
5321 # <field names skipped for the doc example>
5322 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5323 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5324
5325 See also: "server-state-file", "server-state-file-name", and
5326 "show servers state"
5327
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005328
Willy Tarreau2769aa02007-12-27 18:26:09 +01005329log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005330log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
5331 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005332no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005333 Enable per-instance logging of events and traffic.
5334 May be used in sections : defaults | frontend | listen | backend
5335 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005336
5337 Prefix :
5338 no should be used when the logger list must be flushed. For example,
5339 if you don't want to inherit from the default logger list. This
5340 prefix does not allow arguments.
5341
Willy Tarreau2769aa02007-12-27 18:26:09 +01005342 Arguments :
5343 global should be used when the instance's logging parameters are the
5344 same as the global ones. This is the most common usage. "global"
5345 replaces <address>, <facility> and <level> with those of the log
5346 entries found in the "global" section. Only one "log global"
5347 statement may be used per instance, and this form takes no other
5348 parameter.
5349
5350 <address> indicates where to send the logs. It takes the same format as
5351 for the "global" section's logs, and can be one of :
5352
5353 - An IPv4 address optionally followed by a colon (':') and a UDP
5354 port. If no port is specified, 514 is used by default (the
5355 standard syslog port).
5356
David du Colombier24bb5f52011-03-17 10:40:23 +01005357 - An IPv6 address followed by a colon (':') and optionally a UDP
5358 port. If no port is specified, 514 is used by default (the
5359 standard syslog port).
5360
Willy Tarreau2769aa02007-12-27 18:26:09 +01005361 - A filesystem path to a UNIX domain socket, keeping in mind
5362 considerations for chroot (be sure the path is accessible
5363 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005364 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005365
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005366 - A file descriptor number in the form "fd@<number>", which may
5367 point to a pipe, terminal, or socket. In this case unbuffered
5368 logs are used and one writev() call per log is performed. This
5369 is a bit expensive but acceptable for most workloads. Messages
5370 sent this way will not be truncated but may be dropped, in
5371 which case the DroppedLogs counter will be incremented. The
5372 writev() call is atomic even on pipes for messages up to
5373 PIPE_BUF size, which POSIX recommends to be at least 512 and
5374 which is 4096 bytes on most modern operating systems. Any
5375 larger message may be interleaved with messages from other
5376 processes. Exceptionally for debugging purposes the file
5377 descriptor may also be directed to a file, but doing so will
5378 significantly slow haproxy down as non-blocking calls will be
5379 ignored. Also there will be no way to purge nor rotate this
5380 file without restarting the process. Note that the configured
5381 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005382 with a TCP syslog server. See also the "short" and "raw"
5383 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005384
5385 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5386 and "fd@2", see above.
5387
5388 You may want to reference some environment variables in the
5389 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005390
Willy Tarreau18324f52014-06-27 18:10:07 +02005391 <length> is an optional maximum line length. Log lines larger than this
5392 value will be truncated before being sent. The reason is that
5393 syslog servers act differently on log line length. All servers
5394 support the default value of 1024, but some servers simply drop
5395 larger lines while others do log them. If a server supports long
5396 lines, it may make sense to set this value here in order to avoid
5397 truncating long lines. Similarly, if a server drops long lines,
5398 it is preferable to truncate them before sending them. Accepted
5399 values are 80 to 65535 inclusive. The default value of 1024 is
5400 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005401 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005402
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005403 <ranges> A list of comma-separated ranges to identify the logs to sample.
5404 This is used to balance the load of the logs to send to the log
5405 server. The limits of the ranges cannot be null. They are numbered
5406 from 1. The size or period (in number of logs) of the sample must
5407 be set with <sample_size> parameter.
5408
5409 <sample_size>
5410 The size of the sample in number of logs to consider when balancing
5411 their logging loads. It is used to balance the load of the logs to
5412 send to the syslog server. This size must be greater or equal to the
5413 maximum of the high limits of the ranges.
5414 (see also <ranges> parameter).
5415
Willy Tarreauadb345d2018-11-12 07:56:13 +01005416 <format> is the log format used when generating syslog messages. It may be
5417 one of the following :
5418
5419 rfc3164 The RFC3164 syslog message format. This is the default.
5420 (https://tools.ietf.org/html/rfc3164)
5421
5422 rfc5424 The RFC5424 syslog message format.
5423 (https://tools.ietf.org/html/rfc5424)
5424
Willy Tarreaue8746a02018-11-12 08:45:00 +01005425 short A message containing only a level between angle brackets such as
5426 '<3>', followed by the text. The PID, date, time, process name
5427 and system name are omitted. This is designed to be used with a
5428 local log server. This format is compatible with what the
5429 systemd logger consumes.
5430
Willy Tarreauc1b06452018-11-12 11:57:56 +01005431 raw A message containing only the text. The level, PID, date, time,
5432 process name and system name are omitted. This is designed to
5433 be used in containers or during development, where the severity
5434 only depends on the file descriptor used (stdout/stderr).
5435
Willy Tarreau2769aa02007-12-27 18:26:09 +01005436 <facility> must be one of the 24 standard syslog facilities :
5437
Willy Tarreaue8746a02018-11-12 08:45:00 +01005438 kern user mail daemon auth syslog lpr news
5439 uucp cron auth2 ftp ntp audit alert cron2
5440 local0 local1 local2 local3 local4 local5 local6 local7
5441
Willy Tarreauc1b06452018-11-12 11:57:56 +01005442 Note that the facility is ignored for the "short" and "raw"
5443 formats, but still required as a positional field. It is
5444 recommended to use "daemon" in this case to make it clear that
5445 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005446
5447 <level> is optional and can be specified to filter outgoing messages. By
5448 default, all messages are sent. If a level is specified, only
5449 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005450 will be sent. An optional minimum level can be specified. If it
5451 is set, logs emitted with a more severe level than this one will
5452 be capped to this level. This is used to avoid sending "emerg"
5453 messages on all terminals on some default syslog configurations.
5454 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005455
5456 emerg alert crit err warning notice info debug
5457
William Lallemand0f99e342011-10-12 17:50:54 +02005458 It is important to keep in mind that it is the frontend which decides what to
5459 log from a connection, and that in case of content switching, the log entries
5460 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005461
5462 However, backend log declaration define how and where servers status changes
5463 will be logged. Level "notice" will be used to indicate a server going up,
5464 "warning" will be used for termination signals and definitive service
5465 termination, and "alert" will be used for when a server goes down.
5466
5467 Note : According to RFC3164, messages are truncated to 1024 bytes before
5468 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005469
5470 Example :
5471 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005472 log stdout format short daemon # send log to systemd
5473 log stdout format raw daemon # send everything to stdout
5474 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005475 log 127.0.0.1:514 local0 notice # only send important events
5476 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005477 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005478
Willy Tarreau2769aa02007-12-27 18:26:09 +01005479
William Lallemand48940402012-01-30 16:47:22 +01005480log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005481 Specifies the log format string to use for traffic logs
5482 May be used in sections: defaults | frontend | listen | backend
5483 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005484
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005485 This directive specifies the log format string that will be used for all logs
5486 resulting from traffic passing through the frontend using this line. If the
5487 directive is used in a defaults section, all subsequent frontends will use
5488 the same log format. Please see section 8.2.4 which covers the log format
5489 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005490
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005491 "log-format" directive overrides previous "option tcplog", "log-format" and
5492 "option httplog" directives.
5493
Dragan Dosen7ad31542015-09-28 17:16:47 +02005494log-format-sd <string>
5495 Specifies the RFC5424 structured-data log format string
5496 May be used in sections: defaults | frontend | listen | backend
5497 yes | yes | yes | no
5498
5499 This directive specifies the RFC5424 structured-data log format string that
5500 will be used for all logs resulting from traffic passing through the frontend
5501 using this line. If the directive is used in a defaults section, all
5502 subsequent frontends will use the same log format. Please see section 8.2.4
5503 which covers the log format string in depth.
5504
5505 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5506 about the RFC5424 structured-data part.
5507
5508 Note : This log format string will be used only for loggers that have set
5509 log format to "rfc5424".
5510
5511 Example :
5512 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5513
5514
Willy Tarreau094af4e2015-01-07 15:03:42 +01005515log-tag <string>
5516 Specifies the log tag to use for all outgoing logs
5517 May be used in sections: defaults | frontend | listen | backend
5518 yes | yes | yes | yes
5519
5520 Sets the tag field in the syslog header to this string. It defaults to the
5521 log-tag set in the global section, otherwise the program name as launched
5522 from the command line, which usually is "haproxy". Sometimes it can be useful
5523 to differentiate between multiple processes running on the same host, or to
5524 differentiate customer instances running in the same process. In the backend,
5525 logs about servers up/down will use this tag. As a hint, it can be convenient
5526 to set a log-tag related to a hosted customer in a defaults section then put
5527 all the frontends and backends for that customer, then start another customer
5528 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005529
Willy Tarreauc35362a2014-04-25 13:58:37 +02005530max-keep-alive-queue <value>
5531 Set the maximum server queue size for maintaining keep-alive connections
5532 May be used in sections: defaults | frontend | listen | backend
5533 yes | no | yes | yes
5534
5535 HTTP keep-alive tries to reuse the same server connection whenever possible,
5536 but sometimes it can be counter-productive, for example if a server has a lot
5537 of connections while other ones are idle. This is especially true for static
5538 servers.
5539
5540 The purpose of this setting is to set a threshold on the number of queued
5541 connections at which haproxy stops trying to reuse the same server and prefers
5542 to find another one. The default value, -1, means there is no limit. A value
5543 of zero means that keep-alive requests will never be queued. For very close
5544 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005545 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005546 use a value of 10 or less). For remote servers suffering from a high latency,
5547 higher values might be needed to cover for the latency and/or the cost of
5548 picking a different server.
5549
5550 Note that this has no impact on responses which are maintained to the same
5551 server consecutively to a 401 response. They will still go to the same server
5552 even if they have to be queued.
5553
5554 See also : "option http-server-close", "option prefer-last-server", server
5555 "maxconn" and cookie persistence.
5556
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01005557max-session-srv-conns <nb>
5558 Set the maximum number of outgoing connections we can keep idling for a given
5559 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
5560 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02005561
Willy Tarreau2769aa02007-12-27 18:26:09 +01005562maxconn <conns>
5563 Fix the maximum number of concurrent connections on a frontend
5564 May be used in sections : defaults | frontend | listen | backend
5565 yes | yes | yes | no
5566 Arguments :
5567 <conns> is the maximum number of concurrent connections the frontend will
5568 accept to serve. Excess connections will be queued by the system
5569 in the socket's listen queue and will be served once a connection
5570 closes.
5571
5572 If the system supports it, it can be useful on big sites to raise this limit
5573 very high so that haproxy manages connection queues, instead of leaving the
5574 clients with unanswered connection attempts. This value should not exceed the
5575 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005576 of tune.bufsize (16kB by default) each, as well as some other data resulting
5577 in about 33 kB of RAM being consumed per established connection. That means
5578 that a medium system equipped with 1GB of RAM can withstand around
5579 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005580
5581 Also, when <conns> is set to large values, it is possible that the servers
5582 are not sized to accept such loads, and for this reason it is generally wise
5583 to assign them some reasonable connection limits.
5584
Willy Tarreauc8d5b952019-02-27 17:25:52 +01005585 When this value is set to zero, which is the default, the global "maxconn"
5586 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02005587
Willy Tarreau2769aa02007-12-27 18:26:09 +01005588 See also : "server", global section's "maxconn", "fullconn"
5589
5590
5591mode { tcp|http|health }
5592 Set the running mode or protocol of the instance
5593 May be used in sections : defaults | frontend | listen | backend
5594 yes | yes | yes | yes
5595 Arguments :
5596 tcp The instance will work in pure TCP mode. A full-duplex connection
5597 will be established between clients and servers, and no layer 7
5598 examination will be performed. This is the default mode. It
5599 should be used for SSL, SSH, SMTP, ...
5600
5601 http The instance will work in HTTP mode. The client request will be
5602 analyzed in depth before connecting to any server. Any request
5603 which is not RFC-compliant will be rejected. Layer 7 filtering,
5604 processing and switching will be possible. This is the mode which
5605 brings HAProxy most of its value.
5606
5607 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005608 to incoming connections and close the connection. Alternatively,
5609 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5610 instead. Nothing will be logged in either case. This mode is used
5611 to reply to external components health checks. This mode is
5612 deprecated and should not be used anymore as it is possible to do
5613 the same and even better by combining TCP or HTTP modes with the
5614 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005615
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005616 When doing content switching, it is mandatory that the frontend and the
5617 backend are in the same mode (generally HTTP), otherwise the configuration
5618 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005619
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005620 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005621 defaults http_instances
5622 mode http
5623
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005624 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005625
Willy Tarreau0ba27502007-12-24 16:55:16 +01005626
Cyril Bontéf0c60612010-02-06 14:44:47 +01005627monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005628 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005629 May be used in sections : defaults | frontend | listen | backend
5630 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005631 Arguments :
5632 if <cond> the monitor request will fail if the condition is satisfied,
5633 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005634 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005635 are met, for instance a low number of servers both in a
5636 backend and its backup.
5637
5638 unless <cond> the monitor request will succeed only if the condition is
5639 satisfied, and will fail otherwise. Such a condition may be
5640 based on a test on the presence of a minimum number of active
5641 servers in a list of backends.
5642
5643 This statement adds a condition which can force the response to a monitor
5644 request to report a failure. By default, when an external component queries
5645 the URI dedicated to monitoring, a 200 response is returned. When one of the
5646 conditions above is met, haproxy will return 503 instead of 200. This is
5647 very useful to report a site failure to an external component which may base
5648 routing advertisements between multiple sites on the availability reported by
5649 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005650 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5651 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005652
5653 Example:
5654 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005655 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005656 acl site_dead nbsrv(dynamic) lt 2
5657 acl site_dead nbsrv(static) lt 2
5658 monitor-uri /site_alive
5659 monitor fail if site_dead
5660
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005661 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005662
5663
5664monitor-net <source>
5665 Declare a source network which is limited to monitor requests
5666 May be used in sections : defaults | frontend | listen | backend
5667 yes | yes | yes | no
5668 Arguments :
5669 <source> is the source IPv4 address or network which will only be able to
5670 get monitor responses to any request. It can be either an IPv4
5671 address, a host name, or an address followed by a slash ('/')
5672 followed by a mask.
5673
5674 In TCP mode, any connection coming from a source matching <source> will cause
5675 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005676 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005677 forwarding the connection to a remote server.
5678
5679 In HTTP mode, a connection coming from a source matching <source> will be
5680 accepted, the following response will be sent without waiting for a request,
5681 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5682 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005683 running without forwarding the request to a backend server. Note that this
5684 response is sent in raw format, without any transformation. This is important
5685 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005686
Willy Tarreau82569f92012-09-27 23:48:56 +02005687 Monitor requests are processed very early, just after tcp-request connection
5688 ACLs which are the only ones able to block them. These connections are short
5689 lived and never wait for any data from the client. They cannot be logged, and
5690 it is the intended purpose. They are only used to report HAProxy's health to
5691 an upper component, nothing more. Please note that "monitor fail" rules do
5692 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005693
Willy Tarreau95cd2832010-03-04 23:36:33 +01005694 Last, please note that only one "monitor-net" statement can be specified in
5695 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005696
Willy Tarreau2769aa02007-12-27 18:26:09 +01005697 Example :
5698 # addresses .252 and .253 are just probing us.
5699 frontend www
5700 monitor-net 192.168.0.252/31
5701
5702 See also : "monitor fail", "monitor-uri"
5703
5704
5705monitor-uri <uri>
5706 Intercept a URI used by external components' monitor requests
5707 May be used in sections : defaults | frontend | listen | backend
5708 yes | yes | yes | no
5709 Arguments :
5710 <uri> is the exact URI which we want to intercept to return HAProxy's
5711 health status instead of forwarding the request.
5712
5713 When an HTTP request referencing <uri> will be received on a frontend,
5714 HAProxy will not forward it nor log it, but instead will return either
5715 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5716 conditions defined with "monitor fail". This is normally enough for any
5717 front-end HTTP probe to detect that the service is UP and running without
5718 forwarding the request to a backend server. Note that the HTTP method, the
5719 version and all headers are ignored, but the request must at least be valid
5720 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5721
Willy Tarreau721d8e02017-12-01 18:25:08 +01005722 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005723 and even before any "http-request". The only rulesets applied before are the
5724 tcp-request ones. They cannot be logged either, and it is the intended
5725 purpose. They are only used to report HAProxy's health to an upper component,
5726 nothing more. However, it is possible to add any number of conditions using
5727 "monitor fail" and ACLs so that the result can be adjusted to whatever check
5728 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005729
5730 Example :
5731 # Use /haproxy_test to report haproxy's status
5732 frontend www
5733 mode http
5734 monitor-uri /haproxy_test
5735
5736 See also : "monitor fail", "monitor-net"
5737
Willy Tarreau0ba27502007-12-24 16:55:16 +01005738
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005739option abortonclose
5740no option abortonclose
5741 Enable or disable early dropping of aborted requests pending in queues.
5742 May be used in sections : defaults | frontend | listen | backend
5743 yes | no | yes | yes
5744 Arguments : none
5745
5746 In presence of very high loads, the servers will take some time to respond.
5747 The per-instance connection queue will inflate, and the response time will
5748 increase respective to the size of the queue times the average per-session
5749 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005750 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005751 the queue, and slowing down other users, and the servers as well, because the
5752 request will eventually be served, then aborted at the first error
5753 encountered while delivering the response.
5754
5755 As there is no way to distinguish between a full STOP and a simple output
5756 close on the client side, HTTP agents should be conservative and consider
5757 that the client might only have closed its output channel while waiting for
5758 the response. However, this introduces risks of congestion when lots of users
5759 do the same, and is completely useless nowadays because probably no client at
5760 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005761 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005762 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005763 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005764 of being the single component to break rare but valid traffic is extremely
5765 low, which adds to the temptation to be able to abort a session early while
5766 still not served and not pollute the servers.
5767
Davor Ocelice9ed2812017-12-25 17:49:28 +01005768 In HAProxy, the user can choose the desired behavior using the option
5769 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005770 compliant and aborted requests will be served. But when the option is
5771 specified, a session with an incoming channel closed will be aborted while
5772 it is still possible, either pending in the queue for a connection slot, or
5773 during the connection establishment if the server has not yet acknowledged
5774 the connection request. This considerably reduces the queue size and the load
5775 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005776 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005777
5778 If this option has been enabled in a "defaults" section, it can be disabled
5779 in a specific instance by prepending the "no" keyword before it.
5780
5781 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5782
5783
Willy Tarreau4076a152009-04-02 15:18:36 +02005784option accept-invalid-http-request
5785no option accept-invalid-http-request
5786 Enable or disable relaxing of HTTP request parsing
5787 May be used in sections : defaults | frontend | listen | backend
5788 yes | yes | yes | no
5789 Arguments : none
5790
Willy Tarreau91852eb2015-05-01 13:26:00 +02005791 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005792 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005793 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005794 forbidden characters are essentially used to build attacks exploiting server
5795 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5796 server will emit invalid header names for whatever reason (configuration,
5797 implementation) and the issue will not be immediately fixed. In such a case,
5798 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005799 even if that does not make sense, by specifying this option. Similarly, the
5800 list of characters allowed to appear in a URI is well defined by RFC3986, and
5801 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5802 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005803 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005804 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005805 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5806 to pass through (no version specified) and multiple digits for both the major
5807 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005808
5809 This option should never be enabled by default as it hides application bugs
5810 and open security breaches. It should only be deployed after a problem has
5811 been confirmed.
5812
5813 When this option is enabled, erroneous header names will still be accepted in
5814 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005815 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5816 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005817 also helps confirming that the issue has been solved.
5818
5819 If this option has been enabled in a "defaults" section, it can be disabled
5820 in a specific instance by prepending the "no" keyword before it.
5821
5822 See also : "option accept-invalid-http-response" and "show errors" on the
5823 stats socket.
5824
5825
5826option accept-invalid-http-response
5827no option accept-invalid-http-response
5828 Enable or disable relaxing of HTTP response parsing
5829 May be used in sections : defaults | frontend | listen | backend
5830 yes | no | yes | yes
5831 Arguments : none
5832
Willy Tarreau91852eb2015-05-01 13:26:00 +02005833 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005834 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005835 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005836 forbidden characters are essentially used to build attacks exploiting server
5837 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5838 server will emit invalid header names for whatever reason (configuration,
5839 implementation) and the issue will not be immediately fixed. In such a case,
5840 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005841 even if that does not make sense, by specifying this option. This option also
5842 relaxes the test on the HTTP version format, it allows multiple digits for
5843 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005844
5845 This option should never be enabled by default as it hides application bugs
5846 and open security breaches. It should only be deployed after a problem has
5847 been confirmed.
5848
5849 When this option is enabled, erroneous header names will still be accepted in
5850 responses, but the complete response will be captured in order to permit
5851 later analysis using the "show errors" request on the UNIX stats socket.
5852 Doing this also helps confirming that the issue has been solved.
5853
5854 If this option has been enabled in a "defaults" section, it can be disabled
5855 in a specific instance by prepending the "no" keyword before it.
5856
5857 See also : "option accept-invalid-http-request" and "show errors" on the
5858 stats socket.
5859
5860
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005861option allbackups
5862no option allbackups
5863 Use either all backup servers at a time or only the first one
5864 May be used in sections : defaults | frontend | listen | backend
5865 yes | no | yes | yes
5866 Arguments : none
5867
5868 By default, the first operational backup server gets all traffic when normal
5869 servers are all down. Sometimes, it may be preferred to use multiple backups
5870 at once, because one will not be enough. When "option allbackups" is enabled,
5871 the load balancing will be performed among all backup servers when all normal
5872 ones are unavailable. The same load balancing algorithm will be used and the
5873 servers' weights will be respected. Thus, there will not be any priority
5874 order between the backup servers anymore.
5875
5876 This option is mostly used with static server farms dedicated to return a
5877 "sorry" page when an application is completely offline.
5878
5879 If this option has been enabled in a "defaults" section, it can be disabled
5880 in a specific instance by prepending the "no" keyword before it.
5881
5882
5883option checkcache
5884no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005885 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005886 May be used in sections : defaults | frontend | listen | backend
5887 yes | no | yes | yes
5888 Arguments : none
5889
5890 Some high-level frameworks set application cookies everywhere and do not
5891 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005892 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005893 high risk of session crossing or stealing between users traversing the same
5894 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005895 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005896
5897 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005898 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005899 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005900 response to check if there's a risk of caching a cookie on a client-side
5901 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005902 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005903 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005904 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5905 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005906 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005907 - all those that result from a request using a method other than GET, HEAD,
5908 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005909 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005910 - those with a 'Pragma: no-cache' header
5911 - those with a 'Cache-control: private' header
5912 - those with a 'Cache-control: no-store' header
5913 - those with a 'Cache-control: max-age=0' header
5914 - those with a 'Cache-control: s-maxage=0' header
5915 - those with a 'Cache-control: no-cache' header
5916 - those with a 'Cache-control: no-cache="set-cookie"' header
5917 - those with a 'Cache-control: no-cache="set-cookie,' header
5918 (allowing other fields after set-cookie)
5919
5920 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005921 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
5922 gateway". The session state shows "PH--" meaning that the proxy blocked the
5923 response during headers processing. Additionally, an alert will be sent in
5924 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005925
5926 Due to the high impact on the application, the application should be tested
5927 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005928 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005929 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005930
5931 If this option has been enabled in a "defaults" section, it can be disabled
5932 in a specific instance by prepending the "no" keyword before it.
5933
5934
5935option clitcpka
5936no option clitcpka
5937 Enable or disable the sending of TCP keepalive packets on the client side
5938 May be used in sections : defaults | frontend | listen | backend
5939 yes | yes | yes | no
5940 Arguments : none
5941
5942 When there is a firewall or any session-aware component between a client and
5943 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005944 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005945 components decides to expire a session which has remained idle for too long.
5946
5947 Enabling socket-level TCP keep-alives makes the system regularly send packets
5948 to the other end of the connection, leaving it active. The delay between
5949 keep-alive probes is controlled by the system only and depends both on the
5950 operating system and its tuning parameters.
5951
5952 It is important to understand that keep-alive packets are neither emitted nor
5953 received at the application level. It is only the network stacks which sees
5954 them. For this reason, even if one side of the proxy already uses keep-alives
5955 to maintain its connection alive, those keep-alive packets will not be
5956 forwarded to the other side of the proxy.
5957
5958 Please note that this has nothing to do with HTTP keep-alive.
5959
5960 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5961 client side of a connection, which should help when session expirations are
5962 noticed between HAProxy and a client.
5963
5964 If this option has been enabled in a "defaults" section, it can be disabled
5965 in a specific instance by prepending the "no" keyword before it.
5966
5967 See also : "option srvtcpka", "option tcpka"
5968
5969
Willy Tarreau0ba27502007-12-24 16:55:16 +01005970option contstats
5971 Enable continuous traffic statistics updates
5972 May be used in sections : defaults | frontend | listen | backend
5973 yes | yes | yes | no
5974 Arguments : none
5975
5976 By default, counters used for statistics calculation are incremented
5977 only when a session finishes. It works quite well when serving small
5978 objects, but with big ones (for example large images or archives) or
5979 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01005980 a hedgehog. With this option enabled counters get incremented frequently
5981 along the session, typically every 5 seconds, which is often enough to
5982 produce clean graphs. Recounting touches a hotpath directly so it is not
5983 not enabled by default, as it can cause a lot of wakeups for very large
5984 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005985
5986
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005987option dontlog-normal
5988no option dontlog-normal
5989 Enable or disable logging of normal, successful connections
5990 May be used in sections : defaults | frontend | listen | backend
5991 yes | yes | yes | no
5992 Arguments : none
5993
5994 There are large sites dealing with several thousand connections per second
5995 and for which logging is a major pain. Some of them are even forced to turn
5996 logs off and cannot debug production issues. Setting this option ensures that
5997 normal connections, those which experience no error, no timeout, no retry nor
5998 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
5999 mode, the response status code is checked and return codes 5xx will still be
6000 logged.
6001
6002 It is strongly discouraged to use this option as most of the time, the key to
6003 complex issues is in the normal logs which will not be logged here. If you
6004 need to separate logs, see the "log-separate-errors" option instead.
6005
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006006 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006007 logging.
6008
6009
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006010option dontlognull
6011no option dontlognull
6012 Enable or disable logging of null connections
6013 May be used in sections : defaults | frontend | listen | backend
6014 yes | yes | yes | no
6015 Arguments : none
6016
6017 In certain environments, there are components which will regularly connect to
6018 various systems to ensure that they are still alive. It can be the case from
6019 another load balancer as well as from monitoring systems. By default, even a
6020 simple port probe or scan will produce a log. If those connections pollute
6021 the logs too much, it is possible to enable option "dontlognull" to indicate
6022 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02006023 which typically corresponds to those probes. Note that errors will still be
6024 returned to the client and accounted for in the stats. If this is not what is
6025 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006026
6027 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01006028 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006029 would not be logged.
6030
6031 If this option has been enabled in a "defaults" section, it can be disabled
6032 in a specific instance by prepending the "no" keyword before it.
6033
Willy Tarreau0f228a02015-05-01 15:37:53 +02006034 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
6035 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006036
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006037
Willy Tarreau87cf5142011-08-19 22:57:24 +02006038option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006039 Enable insertion of the X-Forwarded-For header to requests sent to servers
6040 May be used in sections : defaults | frontend | listen | backend
6041 yes | yes | yes | yes
6042 Arguments :
6043 <network> is an optional argument used to disable this option for sources
6044 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02006045 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01006046 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006047
6048 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
6049 their client address. This is sometimes annoying when the client's IP address
6050 is expected in server logs. To solve this problem, the well-known HTTP header
6051 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
6052 This header contains a value representing the client's IP address. Since this
6053 header is always appended at the end of the existing header list, the server
6054 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02006055 the server's manual to find how to enable use of this standard header. Note
6056 that only the last occurrence of the header must be used, since it is really
6057 possible that the client has already brought one.
6058
Willy Tarreaud72758d2010-01-12 10:42:19 +01006059 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02006060 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01006061 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01006062 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006063 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02006064 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01006065
6066 Sometimes, a same HAProxy instance may be shared between a direct client
6067 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6068 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6069 header for a known source address or network by adding the "except" keyword
6070 followed by the network address. In this case, any source IP matching the
6071 network will not cause an addition of this header. Most common uses are with
6072 private networks or 127.0.0.1.
6073
Willy Tarreau87cf5142011-08-19 22:57:24 +02006074 Alternatively, the keyword "if-none" states that the header will only be
6075 added if it is not present. This should only be used in perfectly trusted
6076 environment, as this might cause a security issue if headers reaching haproxy
6077 are under the control of the end-user.
6078
Willy Tarreauc27debf2008-01-06 08:57:02 +01006079 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02006080 least one of them uses it, the header will be added. Note that the backend's
6081 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02006082 both are defined. In the case of the "if-none" argument, if at least one of
6083 the frontend or the backend does not specify it, it wants the addition to be
6084 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006085
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006086 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01006087 # Public HTTP address also used by stunnel on the same machine
6088 frontend www
6089 mode http
6090 option forwardfor except 127.0.0.1 # stunnel already adds the header
6091
Ross Westaf72a1d2008-08-03 10:51:45 +02006092 # Those servers want the IP Address in X-Client
6093 backend www
6094 mode http
6095 option forwardfor header X-Client
6096
Willy Tarreau87cf5142011-08-19 22:57:24 +02006097 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006098 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01006099
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006100
Christopher Faulet98fbe952019-07-22 16:18:24 +02006101option h1-case-adjust-bogus-client
6102no option h1-case-adjust-bogus-client
6103 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
6104 May be used in sections : defaults | frontend | listen | backend
6105 yes | yes | yes | no
6106 Arguments : none
6107
6108 There is no standard case for header names because, as stated in RFC7230,
6109 they are case-insensitive. So applications must handle them in a case-
6110 insensitive manner. But some bogus applications violate the standards and
6111 erroneously rely on the cases most commonly used by browsers. This problem
6112 becomes critical with HTTP/2 because all header names must be exchanged in
6113 lower case, and HAProxy follows the same convention. All header names are
6114 sent in lower case to clients and servers, regardless of the HTTP version.
6115
6116 When HAProxy receives an HTTP/1 response, its header names are converted to
6117 lower case and manipulated and sent this way to the clients. If a client is
6118 known to violate the HTTP standards and to fail to process a response coming
6119 from HAProxy, it is possible to transform the lower case header names to a
6120 different format when the response is formatted and sent to the client, by
6121 enabling this option and specifying the list of headers to be reformatted
6122 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6123 must only be a temporary workaround for the time it takes the client to be
6124 fixed, because clients which require such workarounds might be vulnerable to
6125 content smuggling attacks and must absolutely be fixed.
6126
6127 Please note that this option will not affect standards-compliant clients.
6128
6129 If this option has been enabled in a "defaults" section, it can be disabled
6130 in a specific instance by prepending the "no" keyword before it.
6131
6132 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
6133 "h1-case-adjust-file".
6134
6135
6136option h1-case-adjust-bogus-server
6137no option h1-case-adjust-bogus-server
6138 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
6139 May be used in sections : defaults | frontend | listen | backend
6140 yes | no | yes | yes
6141 Arguments : none
6142
6143 There is no standard case for header names because, as stated in RFC7230,
6144 they are case-insensitive. So applications must handle them in a case-
6145 insensitive manner. But some bogus applications violate the standards and
6146 erroneously rely on the cases most commonly used by browsers. This problem
6147 becomes critical with HTTP/2 because all header names must be exchanged in
6148 lower case, and HAProxy follows the same convention. All header names are
6149 sent in lower case to clients and servers, regardless of the HTTP version.
6150
6151 When HAProxy receives an HTTP/1 request, its header names are converted to
6152 lower case and manipulated and sent this way to the servers. If a server is
6153 known to violate the HTTP standards and to fail to process a request coming
6154 from HAProxy, it is possible to transform the lower case header names to a
6155 different format when the request is formatted and sent to the server, by
6156 enabling this option and specifying the list of headers to be reformatted
6157 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6158 must only be a temporary workaround for the time it takes the server to be
6159 fixed, because servers which require such workarounds might be vulnerable to
6160 content smuggling attacks and must absolutely be fixed.
6161
6162 Please note that this option will not affect standards-compliant servers.
6163
6164 If this option has been enabled in a "defaults" section, it can be disabled
6165 in a specific instance by prepending the "no" keyword before it.
6166
6167 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
6168 "h1-case-adjust-file".
6169
6170
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006171option http-buffer-request
6172no option http-buffer-request
6173 Enable or disable waiting for whole HTTP request body before proceeding
6174 May be used in sections : defaults | frontend | listen | backend
6175 yes | yes | yes | yes
6176 Arguments : none
6177
6178 It is sometimes desirable to wait for the body of an HTTP request before
6179 taking a decision. This is what is being done by "balance url_param" for
6180 example. The first use case is to buffer requests from slow clients before
6181 connecting to the server. Another use case consists in taking the routing
6182 decision based on the request body's contents. This option placed in a
6183 frontend or backend forces the HTTP processing to wait until either the whole
6184 body is received, or the request buffer is full, or the first chunk is
6185 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01006186 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006187 the frontend and the backend, so this should definitely not be used by
6188 default.
6189
Baptiste Assmanneccdf432015-10-28 13:49:01 +01006190 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006191
6192
Willy Tarreau0f228a02015-05-01 15:37:53 +02006193option http-ignore-probes
6194no option http-ignore-probes
6195 Enable or disable logging of null connections and request timeouts
6196 May be used in sections : defaults | frontend | listen | backend
6197 yes | yes | yes | no
6198 Arguments : none
6199
6200 Recently some browsers started to implement a "pre-connect" feature
6201 consisting in speculatively connecting to some recently visited web sites
6202 just in case the user would like to visit them. This results in many
6203 connections being established to web sites, which end up in 408 Request
6204 Timeout if the timeout strikes first, or 400 Bad Request when the browser
6205 decides to close them first. These ones pollute the log and feed the error
6206 counters. There was already "option dontlognull" but it's insufficient in
6207 this case. Instead, this option does the following things :
6208 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01006209 was received over a connection before it was closed;
6210 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02006211 - prevent any error counter from being incremented
6212
6213 That way the empty connection is silently ignored. Note that it is better
6214 not to use this unless it is clear that it is needed, because it will hide
6215 real problems. The most common reason for not receiving a request and seeing
6216 a 408 is due to an MTU inconsistency between the client and an intermediary
6217 element such as a VPN, which blocks too large packets. These issues are
6218 generally seen with POST requests as well as GET with large cookies. The logs
6219 are often the only way to detect them.
6220
6221 If this option has been enabled in a "defaults" section, it can be disabled
6222 in a specific instance by prepending the "no" keyword before it.
6223
6224 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
6225
6226
Willy Tarreau16bfb022010-01-16 19:48:41 +01006227option http-keep-alive
6228no option http-keep-alive
6229 Enable or disable HTTP keep-alive from client to server
6230 May be used in sections : defaults | frontend | listen | backend
6231 yes | yes | yes | yes
6232 Arguments : none
6233
Willy Tarreau70dffda2014-01-30 03:07:23 +01006234 By default HAProxy operates in keep-alive mode with regards to persistent
6235 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02006236 leaves the connection idle on both sides between the end of a response and
6237 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006238 as "option http-server-close" or "option httpclose". This option allows to
6239 set back the keep-alive mode, which can be useful when another mode was used
6240 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01006241
6242 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
6243 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01006244 network) and the fastest session reuse on the server side at the expense
6245 of maintaining idle connections to the servers. In general, it is possible
6246 with this option to achieve approximately twice the request rate that the
6247 "http-server-close" option achieves on small objects. There are mainly two
6248 situations where this option may be useful :
6249
6250 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01006251 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01006252
6253 - when the cost of establishing the connection to the server is significant
6254 compared to the cost of retrieving the associated object from the server.
6255
6256 This last case can happen when the server is a fast static server of cache.
6257 In this case, the server will need to be properly tuned to support high enough
6258 connection counts because connections will last until the client sends another
6259 request.
6260
6261 If the client request has to go to another backend or another server due to
6262 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006263 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6264 available to try optimize server selection so that if the server currently
6265 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006266
Willy Tarreau16bfb022010-01-16 19:48:41 +01006267 At the moment, logs will not indicate whether requests came from the same
6268 session or not. The accept date reported in the logs corresponds to the end
6269 of the previous request, and the request time corresponds to the time spent
6270 waiting for a new request. The keep-alive request time is still bound to the
6271 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6272 not set.
6273
Christopher Faulet159e6672019-07-16 15:09:52 +02006274 This option disables and replaces any previous "option httpclose" or "option
6275 http-server-close". When backend and frontend options differ, all of these 4
6276 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006277
Christopher Faulet315b39c2018-09-21 16:26:19 +02006278 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006279 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01006280 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006281
6282
Willy Tarreau96e31212011-05-30 18:10:30 +02006283option http-no-delay
6284no option http-no-delay
6285 Instruct the system to favor low interactive delays over performance in HTTP
6286 May be used in sections : defaults | frontend | listen | backend
6287 yes | yes | yes | yes
6288 Arguments : none
6289
6290 In HTTP, each payload is unidirectional and has no notion of interactivity.
6291 Any agent is expected to queue data somewhat for a reasonably low delay.
6292 There are some very rare server-to-server applications that abuse the HTTP
6293 protocol and expect the payload phase to be highly interactive, with many
6294 interleaved data chunks in both directions within a single request. This is
6295 absolutely not supported by the HTTP specification and will not work across
6296 most proxies or servers. When such applications attempt to do this through
6297 haproxy, it works but they will experience high delays due to the network
6298 optimizations which favor performance by instructing the system to wait for
6299 enough data to be available in order to only send full packets. Typical
6300 delays are around 200 ms per round trip. Note that this only happens with
6301 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6302 affected.
6303
6304 When "option http-no-delay" is present in either the frontend or the backend
6305 used by a connection, all such optimizations will be disabled in order to
6306 make the exchanges as fast as possible. Of course this offers no guarantee on
6307 the functionality, as it may break at any other place. But if it works via
6308 HAProxy, it will work as fast as possible. This option should never be used
6309 by default, and should never be used at all unless such a buggy application
6310 is discovered. The impact of using this option is an increase of bandwidth
6311 usage and CPU usage, which may significantly lower performance in high
6312 latency environments.
6313
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006314 See also : "option http-buffer-request"
6315
Willy Tarreau96e31212011-05-30 18:10:30 +02006316
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006317option http-pretend-keepalive
6318no option http-pretend-keepalive
6319 Define whether haproxy will announce keepalive to the server or not
6320 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006321 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006322 Arguments : none
6323
Christopher Faulet315b39c2018-09-21 16:26:19 +02006324 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006325 adds a "Connection: close" header to the request forwarded to the server.
6326 Unfortunately, when some servers see this header, they automatically refrain
6327 from using the chunked encoding for responses of unknown length, while this
6328 is totally unrelated. The immediate effect is that this prevents haproxy from
6329 maintaining the client connection alive. A second effect is that a client or
6330 a cache could receive an incomplete response without being aware of it, and
6331 consider the response complete.
6332
6333 By setting "option http-pretend-keepalive", haproxy will make the server
6334 believe it will keep the connection alive. The server will then not fall back
6335 to the abnormal undesired above. When haproxy gets the whole response, it
6336 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006337 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006338 connection is correctly closed on the server side.
6339
6340 It is recommended not to enable this option by default, because most servers
6341 will more efficiently close the connection themselves after the last packet,
6342 and release its buffers slightly earlier. Also, the added packet on the
6343 network could slightly reduce the overall peak performance. However it is
6344 worth noting that when this option is enabled, haproxy will have slightly
6345 less work to do. So if haproxy is the bottleneck on the whole architecture,
6346 enabling this option might save a few CPU cycles.
6347
Christopher Faulet98db9762018-09-21 10:25:19 +02006348 This option may be set in backend and listen sections. Using it in a frontend
6349 section will be ignored and a warning will be reported during startup. It is
6350 a backend related option, so there is no real reason to set it on a
6351 frontend. This option may be combined with "option httpclose", which will
6352 cause keepalive to be announced to the server and close to be announced to
6353 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006354
6355 If this option has been enabled in a "defaults" section, it can be disabled
6356 in a specific instance by prepending the "no" keyword before it.
6357
Christopher Faulet315b39c2018-09-21 16:26:19 +02006358 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006359 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006360
Willy Tarreauc27debf2008-01-06 08:57:02 +01006361
Willy Tarreaub608feb2010-01-02 22:47:18 +01006362option http-server-close
6363no option http-server-close
6364 Enable or disable HTTP connection closing on the server side
6365 May be used in sections : defaults | frontend | listen | backend
6366 yes | yes | yes | yes
6367 Arguments : none
6368
Willy Tarreau70dffda2014-01-30 03:07:23 +01006369 By default HAProxy operates in keep-alive mode with regards to persistent
6370 connections: for each connection it processes each request and response, and
6371 leaves the connection idle on both sides between the end of a response and
6372 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006373 as "option http-server-close" or "option httpclose". Setting "option
6374 http-server-close" enables HTTP connection-close mode on the server side
6375 while keeping the ability to support HTTP keep-alive and pipelining on the
6376 client side. This provides the lowest latency on the client side (slow
6377 network) and the fastest session reuse on the server side to save server
6378 resources, similarly to "option httpclose". It also permits non-keepalive
6379 capable servers to be served in keep-alive mode to the clients if they
6380 conform to the requirements of RFC7230. Please note that some servers do not
6381 always conform to those requirements when they see "Connection: close" in the
6382 request. The effect will be that keep-alive will never be used. A workaround
6383 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006384
6385 At the moment, logs will not indicate whether requests came from the same
6386 session or not. The accept date reported in the logs corresponds to the end
6387 of the previous request, and the request time corresponds to the time spent
6388 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006389 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6390 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006391
6392 This option may be set both in a frontend and in a backend. It is enabled if
6393 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02006394 It disables and replaces any previous "option httpclose" or "option
6395 http-keep-alive". Please check section 4 ("Proxies") to see how this option
6396 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006397
6398 If this option has been enabled in a "defaults" section, it can be disabled
6399 in a specific instance by prepending the "no" keyword before it.
6400
Christopher Faulet315b39c2018-09-21 16:26:19 +02006401 See also : "option httpclose", "option http-pretend-keepalive",
6402 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006403
Willy Tarreau88d349d2010-01-25 12:15:43 +01006404option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006405no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006406 Make use of non-standard Proxy-Connection header instead of Connection
6407 May be used in sections : defaults | frontend | listen | backend
6408 yes | yes | yes | no
6409 Arguments : none
6410
Lukas Tribus23953682017-04-28 13:24:30 +00006411 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006412 Connection header to indicate their wish of persistent or non-persistent
6413 connections, both browsers and proxies ignore this header for proxied
6414 connections and make use of the undocumented, non-standard Proxy-Connection
6415 header instead. The issue begins when trying to put a load balancer between
6416 browsers and such proxies, because there will be a difference between what
6417 haproxy understands and what the client and the proxy agree on.
6418
6419 By setting this option in a frontend, haproxy can automatically switch to use
6420 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006421 defined here as one where the URI begins with neither a '/' nor a '*'. This
6422 is incompatible with the HTTP tunnel mode. Note that this option can only be
6423 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006424
Willy Tarreau844a7e72010-01-31 21:46:18 +01006425 Also, when this option is set, a request which requires authentication will
6426 automatically switch to use proxy authentication headers if it is itself a
6427 proxied request. That makes it possible to check or enforce authentication in
6428 front of an existing proxy.
6429
Willy Tarreau88d349d2010-01-25 12:15:43 +01006430 This option should normally never be used, except in front of a proxy.
6431
Christopher Faulet315b39c2018-09-21 16:26:19 +02006432 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006433
Willy Tarreaud63335a2010-02-26 12:56:52 +01006434option httpchk
6435option httpchk <uri>
6436option httpchk <method> <uri>
6437option httpchk <method> <uri> <version>
6438 Enable HTTP protocol to check on the servers health
6439 May be used in sections : defaults | frontend | listen | backend
6440 yes | no | yes | yes
6441 Arguments :
6442 <method> is the optional HTTP method used with the requests. When not set,
6443 the "OPTIONS" method is used, as it generally requires low server
6444 processing and is easy to filter out from the logs. Any method
6445 may be used, though it is not recommended to invent non-standard
6446 ones.
6447
6448 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6449 which is accessible by default on almost any server, but may be
6450 changed to any other URI. Query strings are permitted.
6451
6452 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6453 but some servers might behave incorrectly in HTTP 1.0, so turning
6454 it to HTTP/1.1 may sometimes help. Note that the Host field is
6455 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6456 after "\r\n" following the version string.
6457
6458 By default, server health checks only consist in trying to establish a TCP
6459 connection. When "option httpchk" is specified, a complete HTTP request is
6460 sent once the TCP connection is established, and responses 2xx and 3xx are
6461 considered valid, while all other ones indicate a server failure, including
6462 the lack of any response.
6463
6464 The port and interval are specified in the server configuration.
6465
6466 This option does not necessarily require an HTTP backend, it also works with
6467 plain TCP backends. This is particularly useful to check simple scripts bound
6468 to some dedicated ports using the inetd daemon.
6469
6470 Examples :
6471 # Relay HTTPS traffic to Apache instance and check service availability
6472 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6473 backend https_relay
6474 mode tcp
6475 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6476 server apache1 192.168.1.1:443 check port 80
6477
Simon Hormanafc47ee2013-11-25 10:46:35 +09006478 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6479 "option pgsql-check", "http-check" and the "check", "port" and
6480 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006481
6482
Willy Tarreauc27debf2008-01-06 08:57:02 +01006483option httpclose
6484no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006485 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006486 May be used in sections : defaults | frontend | listen | backend
6487 yes | yes | yes | yes
6488 Arguments : none
6489
Willy Tarreau70dffda2014-01-30 03:07:23 +01006490 By default HAProxy operates in keep-alive mode with regards to persistent
6491 connections: for each connection it processes each request and response, and
6492 leaves the connection idle on both sides between the end of a response and
6493 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006494 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006495
Christopher Faulet315b39c2018-09-21 16:26:19 +02006496 If "option httpclose" is set, HAProxy will close connections with the server
6497 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05006498 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02006499 and will add one if missing. Any "Connection" header different from "close"
6500 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006501
Christopher Faulet315b39c2018-09-21 16:26:19 +02006502 This option may also be combined with "option http-pretend-keepalive", which
6503 will disable sending of the "Connection: close" header, but will still cause
6504 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006505
6506 This option may be set both in a frontend and in a backend. It is enabled if
6507 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02006508 It disables and replaces any previous "option http-server-close" or "option
6509 http-keep-alive". Please check section 4 ("Proxies") to see how this option
6510 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006511
6512 If this option has been enabled in a "defaults" section, it can be disabled
6513 in a specific instance by prepending the "no" keyword before it.
6514
Christopher Faulet315b39c2018-09-21 16:26:19 +02006515 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006516
6517
Emeric Brun3a058f32009-06-30 18:26:00 +02006518option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006519 Enable logging of HTTP request, session state and timers
6520 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006521 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006522 Arguments :
6523 clf if the "clf" argument is added, then the output format will be
6524 the CLF format instead of HAProxy's default HTTP format. You can
6525 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006526 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006527 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006528
6529 By default, the log output format is very poor, as it only contains the
6530 source and destination addresses, and the instance name. By specifying
6531 "option httplog", each log line turns into a much richer format including,
6532 but not limited to, the HTTP request, the connection timers, the session
6533 status, the connections numbers, the captured headers and cookies, the
6534 frontend, backend and server name, and of course the source address and
6535 ports.
6536
PiBa-NLbd556bf2014-12-11 21:31:54 +01006537 Specifying only "option httplog" will automatically clear the 'clf' mode
6538 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006539
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006540 "option httplog" overrides any previous "log-format" directive.
6541
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006542 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006543
Willy Tarreau55165fe2009-05-10 12:02:55 +02006544
6545option http_proxy
6546no option http_proxy
6547 Enable or disable plain HTTP proxy mode
6548 May be used in sections : defaults | frontend | listen | backend
6549 yes | yes | yes | yes
6550 Arguments : none
6551
6552 It sometimes happens that people need a pure HTTP proxy which understands
6553 basic proxy requests without caching nor any fancy feature. In this case,
6554 it may be worth setting up an HAProxy instance with the "option http_proxy"
6555 set. In this mode, no server is declared, and the connection is forwarded to
6556 the IP address and port found in the URL after the "http://" scheme.
6557
6558 No host address resolution is performed, so this only works when pure IP
6559 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006560 it will probably be used only by experts who know they need exactly it. This
6561 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006562
6563 If this option has been enabled in a "defaults" section, it can be disabled
6564 in a specific instance by prepending the "no" keyword before it.
6565
6566 Example :
6567 # this backend understands HTTP proxy requests and forwards them directly.
6568 backend direct_forward
6569 option httpclose
6570 option http_proxy
6571
6572 See also : "option httpclose"
6573
Willy Tarreau211ad242009-10-03 21:45:07 +02006574
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006575option independent-streams
6576no option independent-streams
6577 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006578 May be used in sections : defaults | frontend | listen | backend
6579 yes | yes | yes | yes
6580 Arguments : none
6581
6582 By default, when data is sent over a socket, both the write timeout and the
6583 read timeout for that socket are refreshed, because we consider that there is
6584 activity on that socket, and we have no other means of guessing if we should
6585 receive data or not.
6586
Davor Ocelice9ed2812017-12-25 17:49:28 +01006587 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006588 exists a situation where it is desirable to disable it, and only refresh the
6589 read timeout if there are incoming data. This happens on sessions with large
6590 timeouts and low amounts of exchanged data such as telnet session. If the
6591 server suddenly disappears, the output data accumulates in the system's
6592 socket buffers, both timeouts are correctly refreshed, and there is no way
6593 to know the server does not receive them, so we don't timeout. However, when
6594 the underlying protocol always echoes sent data, it would be enough by itself
6595 to detect the issue using the read timeout. Note that this problem does not
6596 happen with more verbose protocols because data won't accumulate long in the
6597 socket buffers.
6598
6599 When this option is set on the frontend, it will disable read timeout updates
6600 on data sent to the client. There probably is little use of this case. When
6601 the option is set on the backend, it will disable read timeout updates on
6602 data sent to the server. Doing so will typically break large HTTP posts from
6603 slow lines, so use it with caution.
6604
Willy Tarreauce887fd2012-05-12 12:50:00 +02006605 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006606
6607
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006608option ldap-check
6609 Use LDAPv3 health checks for server testing
6610 May be used in sections : defaults | frontend | listen | backend
6611 yes | no | yes | yes
6612 Arguments : none
6613
6614 It is possible to test that the server correctly talks LDAPv3 instead of just
6615 testing that it accepts the TCP connection. When this option is set, an
6616 LDAPv3 anonymous simple bind message is sent to the server, and the response
6617 is analyzed to find an LDAPv3 bind response message.
6618
6619 The server is considered valid only when the LDAP response contains success
6620 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6621
6622 Logging of bind requests is server dependent see your documentation how to
6623 configure it.
6624
6625 Example :
6626 option ldap-check
6627
6628 See also : "option httpchk"
6629
6630
Simon Horman98637e52014-06-20 12:30:16 +09006631option external-check
6632 Use external processes for server health checks
6633 May be used in sections : defaults | frontend | listen | backend
6634 yes | no | yes | yes
6635
6636 It is possible to test the health of a server using an external command.
6637 This is achieved by running the executable set using "external-check
6638 command".
6639
6640 Requires the "external-check" global to be set.
6641
6642 See also : "external-check", "external-check command", "external-check path"
6643
6644
Willy Tarreau211ad242009-10-03 21:45:07 +02006645option log-health-checks
6646no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006647 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006648 May be used in sections : defaults | frontend | listen | backend
6649 yes | no | yes | yes
6650 Arguments : none
6651
Willy Tarreaubef1b322014-05-13 21:01:39 +02006652 By default, failed health check are logged if server is UP and successful
6653 health checks are logged if server is DOWN, so the amount of additional
6654 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006655
Willy Tarreaubef1b322014-05-13 21:01:39 +02006656 When this option is enabled, any change of the health check status or to
6657 the server's health will be logged, so that it becomes possible to know
6658 that a server was failing occasional checks before crashing, or exactly when
6659 it failed to respond a valid HTTP status, then when the port started to
6660 reject connections, then when the server stopped responding at all.
6661
Davor Ocelice9ed2812017-12-25 17:49:28 +01006662 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006663 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006664
Willy Tarreaubef1b322014-05-13 21:01:39 +02006665 See also: "option httpchk", "option ldap-check", "option mysql-check",
6666 "option pgsql-check", "option redis-check", "option smtpchk",
6667 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006668
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006669
6670option log-separate-errors
6671no option log-separate-errors
6672 Change log level for non-completely successful connections
6673 May be used in sections : defaults | frontend | listen | backend
6674 yes | yes | yes | no
6675 Arguments : none
6676
6677 Sometimes looking for errors in logs is not easy. This option makes haproxy
6678 raise the level of logs containing potentially interesting information such
6679 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6680 level changes from "info" to "err". This makes it possible to log them
6681 separately to a different file with most syslog daemons. Be careful not to
6682 remove them from the original file, otherwise you would lose ordering which
6683 provides very important information.
6684
6685 Using this option, large sites dealing with several thousand connections per
6686 second may log normal traffic to a rotating buffer and only archive smaller
6687 error logs.
6688
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006689 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006690 logging.
6691
Willy Tarreauc27debf2008-01-06 08:57:02 +01006692
6693option logasap
6694no option logasap
6695 Enable or disable early logging of HTTP requests
6696 May be used in sections : defaults | frontend | listen | backend
6697 yes | yes | yes | no
6698 Arguments : none
6699
6700 By default, HTTP requests are logged upon termination so that the total
6701 transfer time and the number of bytes appear in the logs. When large objects
6702 are being transferred, it may take a while before the request appears in the
6703 logs. Using "option logasap", the request gets logged as soon as the server
6704 sends the complete headers. The only missing information in the logs will be
6705 the total number of bytes which will indicate everything except the amount
6706 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006707 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006708 "Content-Length" response header so that the logs at least indicate how many
6709 bytes are expected to be transferred.
6710
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006711 Examples :
6712 listen http_proxy 0.0.0.0:80
6713 mode http
6714 option httplog
6715 option logasap
6716 log 192.168.2.200 local3
6717
6718 >>> Feb 6 12:14:14 localhost \
6719 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6720 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6721 "GET /image.iso HTTP/1.0"
6722
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006723 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006724 logging.
6725
6726
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006727option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006728 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006729 May be used in sections : defaults | frontend | listen | backend
6730 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006731 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006732 <username> This is the username which will be used when connecting to MySQL
6733 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006734 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006735
6736 If you specify a username, the check consists of sending two MySQL packet,
6737 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006738 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006739 Error packet. It is a basic but useful test which does not produce error nor
6740 aborted connect on the server. However, it requires adding an authorization
6741 in the MySQL table, like this :
6742
6743 USE mysql;
6744 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6745 FLUSH PRIVILEGES;
6746
6747 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006748 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006749 Error packet, we don't send anything in this mode. It was reported that it
6750 can generate lockout if check is too frequent and/or if there is not enough
6751 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6752 value as if a connection is established successfully within fewer than MySQL
6753 "max_connect_errors" attempts after a previous connection was interrupted,
6754 the error count for the host is cleared to zero. If HAProxy's server get
6755 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6756
6757 Remember that this does not check database presence nor database consistency.
6758 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006759
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006760 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006761
6762 Most often, an incoming MySQL server needs to see the client's IP address for
6763 various purposes, including IP privilege matching and connection logging.
6764 When possible, it is often wise to masquerade the client's IP address when
6765 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006766 which requires the transparent proxy feature to be compiled in, and the MySQL
6767 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006768
6769 See also: "option httpchk"
6770
6771
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006772option nolinger
6773no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006774 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006775 May be used in sections: defaults | frontend | listen | backend
6776 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006777 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006778
Davor Ocelice9ed2812017-12-25 17:49:28 +01006779 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006780 physically disconnected), the session timeouts triggers and the session is
6781 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6782 using some resources and possibly limiting the ability to establish newer
6783 connections.
6784
6785 When this happens, it is possible to activate "option nolinger" which forces
6786 the system to immediately remove any socket's pending data on close. Thus,
6787 the session is instantly purged from the system's tables. This usually has
6788 side effects such as increased number of TCP resets due to old retransmits
6789 getting immediately rejected. Some firewalls may sometimes complain about
6790 this too.
6791
6792 For this reason, it is not recommended to use this option when not absolutely
6793 needed. You know that you need it when you have thousands of FIN_WAIT1
6794 sessions on your system (TIME_WAIT ones do not count).
6795
6796 This option may be used both on frontends and backends, depending on the side
6797 where it is required. Use it on the frontend for clients, and on the backend
6798 for servers.
6799
6800 If this option has been enabled in a "defaults" section, it can be disabled
6801 in a specific instance by prepending the "no" keyword before it.
6802
6803
Willy Tarreau55165fe2009-05-10 12:02:55 +02006804option originalto [ except <network> ] [ header <name> ]
6805 Enable insertion of the X-Original-To header to requests sent to servers
6806 May be used in sections : defaults | frontend | listen | backend
6807 yes | yes | yes | yes
6808 Arguments :
6809 <network> is an optional argument used to disable this option for sources
6810 matching <network>
6811 <name> an optional argument to specify a different "X-Original-To"
6812 header name.
6813
6814 Since HAProxy can work in transparent mode, every request from a client can
6815 be redirected to the proxy and HAProxy itself can proxy every request to a
6816 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6817 be lost. This is annoying when you want access rules based on destination ip
6818 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6819 added by HAProxy to all requests sent to the server. This header contains a
6820 value representing the original destination IP address. Since this must be
6821 configured to always use the last occurrence of this header only. Note that
6822 only the last occurrence of the header must be used, since it is really
6823 possible that the client has already brought one.
6824
6825 The keyword "header" may be used to supply a different header name to replace
6826 the default "X-Original-To". This can be useful where you might already
6827 have a "X-Original-To" header from a different application, and you need
6828 preserve it. Also if your backend server doesn't use the "X-Original-To"
6829 header and requires different one.
6830
6831 Sometimes, a same HAProxy instance may be shared between a direct client
6832 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6833 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6834 header for a known source address or network by adding the "except" keyword
6835 followed by the network address. In this case, any source IP matching the
6836 network will not cause an addition of this header. Most common uses are with
6837 private networks or 127.0.0.1.
6838
6839 This option may be specified either in the frontend or in the backend. If at
6840 least one of them uses it, the header will be added. Note that the backend's
6841 setting of the header subargument takes precedence over the frontend's if
6842 both are defined.
6843
Willy Tarreau55165fe2009-05-10 12:02:55 +02006844 Examples :
6845 # Original Destination address
6846 frontend www
6847 mode http
6848 option originalto except 127.0.0.1
6849
6850 # Those servers want the IP Address in X-Client-Dst
6851 backend www
6852 mode http
6853 option originalto header X-Client-Dst
6854
Christopher Faulet315b39c2018-09-21 16:26:19 +02006855 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02006856
6857
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006858option persist
6859no option persist
6860 Enable or disable forced persistence on down servers
6861 May be used in sections: defaults | frontend | listen | backend
6862 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006863 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006864
6865 When an HTTP request reaches a backend with a cookie which references a dead
6866 server, by default it is redispatched to another server. It is possible to
6867 force the request to be sent to the dead server first using "option persist"
6868 if absolutely needed. A common use case is when servers are under extreme
6869 load and spend their time flapping. In this case, the users would still be
6870 directed to the server they opened the session on, in the hope they would be
6871 correctly served. It is recommended to use "option redispatch" in conjunction
6872 with this option so that in the event it would not be possible to connect to
6873 the server at all (server definitely dead), the client would finally be
6874 redirected to another valid server.
6875
6876 If this option has been enabled in a "defaults" section, it can be disabled
6877 in a specific instance by prepending the "no" keyword before it.
6878
Willy Tarreau4de91492010-01-22 19:10:05 +01006879 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006880
6881
Willy Tarreau0c122822013-12-15 18:49:01 +01006882option pgsql-check [ user <username> ]
6883 Use PostgreSQL health checks for server testing
6884 May be used in sections : defaults | frontend | listen | backend
6885 yes | no | yes | yes
6886 Arguments :
6887 <username> This is the username which will be used when connecting to
6888 PostgreSQL server.
6889
6890 The check sends a PostgreSQL StartupMessage and waits for either
6891 Authentication request or ErrorResponse message. It is a basic but useful
6892 test which does not produce error nor aborted connect on the server.
6893 This check is identical with the "mysql-check".
6894
6895 See also: "option httpchk"
6896
6897
Willy Tarreau9420b122013-12-15 18:58:25 +01006898option prefer-last-server
6899no option prefer-last-server
6900 Allow multiple load balanced requests to remain on the same server
6901 May be used in sections: defaults | frontend | listen | backend
6902 yes | no | yes | yes
6903 Arguments : none
6904
6905 When the load balancing algorithm in use is not deterministic, and a previous
6906 request was sent to a server to which haproxy still holds a connection, it is
6907 sometimes desirable that subsequent requests on a same session go to the same
6908 server as much as possible. Note that this is different from persistence, as
6909 we only indicate a preference which haproxy tries to apply without any form
6910 of warranty. The real use is for keep-alive connections sent to servers. When
6911 this option is used, haproxy will try to reuse the same connection that is
6912 attached to the server instead of rebalancing to another server, causing a
6913 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006914 not make much sense to use this in combination with hashing algorithms. Note,
6915 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02006916 to a proxy which sends a 407 (authentication required), when the load
6917 balancing algorithm is not deterministic. This is mandatory for use with the
6918 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01006919 troubleshooting some faulty applications. Option prefer-last-server might be
6920 desirable in these environments as well, to avoid redistributing the traffic
6921 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006922
6923 If this option has been enabled in a "defaults" section, it can be disabled
6924 in a specific instance by prepending the "no" keyword before it.
6925
6926 See also: "option http-keep-alive"
6927
6928
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006929option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006930option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006931no option redispatch
6932 Enable or disable session redistribution in case of connection failure
6933 May be used in sections: defaults | frontend | listen | backend
6934 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006935 Arguments :
6936 <interval> The optional integer value that controls how often redispatches
6937 occur when retrying connections. Positive value P indicates a
6938 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006939 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006940 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006941 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006942 positive value of 1 would indicate a redispatch on every retry,
6943 and a positive value of 3 would indicate a redispatch on every
6944 third retry. You can disable redispatches with a value of 0.
6945
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006946
6947 In HTTP mode, if a server designated by a cookie is down, clients may
6948 definitely stick to it because they cannot flush the cookie, so they will not
6949 be able to access the service anymore.
6950
Willy Tarreau59884a62019-01-02 14:48:31 +01006951 Specifying "option redispatch" will allow the proxy to break cookie or
6952 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006953
Joseph Lynch726ab712015-05-11 23:25:34 -07006954 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006955 connection failures. Of course, it requires having "retries" set to a nonzero
6956 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006957
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006958 If this option has been enabled in a "defaults" section, it can be disabled
6959 in a specific instance by prepending the "no" keyword before it.
6960
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006961 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006962
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006963
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006964option redis-check
6965 Use redis health checks for server testing
6966 May be used in sections : defaults | frontend | listen | backend
6967 yes | no | yes | yes
6968 Arguments : none
6969
6970 It is possible to test that the server correctly talks REDIS protocol instead
6971 of just testing that it accepts the TCP connection. When this option is set,
6972 a PING redis command is sent to the server, and the response is analyzed to
6973 find the "+PONG" response message.
6974
6975 Example :
6976 option redis-check
6977
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006978 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006979
6980
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006981option smtpchk
6982option smtpchk <hello> <domain>
6983 Use SMTP health checks for server testing
6984 May be used in sections : defaults | frontend | listen | backend
6985 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01006986 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006987 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02006988 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006989 values will be turned into the default command ("HELO").
6990
6991 <domain> is the domain name to present to the server. It may only be
6992 specified (and is mandatory) if the hello command has been
6993 specified. By default, "localhost" is used.
6994
6995 When "option smtpchk" is set, the health checks will consist in TCP
6996 connections followed by an SMTP command. By default, this command is
6997 "HELO localhost". The server's return code is analyzed and only return codes
6998 starting with a "2" will be considered as valid. All other responses,
6999 including a lack of response will constitute an error and will indicate a
7000 dead server.
7001
7002 This test is meant to be used with SMTP servers or relays. Depending on the
7003 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01007004 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007005 25 is often easier than adjusting the configuration.
7006
7007 Most often, an incoming SMTP server needs to see the client's IP address for
7008 various purposes, including spam filtering, anti-spoofing and logging. When
7009 possible, it is often wise to masquerade the client's IP address when
7010 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007011 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007012
7013 Example :
7014 option smtpchk HELO mydomain.org
7015
7016 See also : "option httpchk", "source"
7017
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007018
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02007019option socket-stats
7020no option socket-stats
7021
7022 Enable or disable collecting & providing separate statistics for each socket.
7023 May be used in sections : defaults | frontend | listen | backend
7024 yes | yes | yes | no
7025
7026 Arguments : none
7027
7028
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007029option splice-auto
7030no option splice-auto
7031 Enable or disable automatic kernel acceleration on sockets in both directions
7032 May be used in sections : defaults | frontend | listen | backend
7033 yes | yes | yes | yes
7034 Arguments : none
7035
7036 When this option is enabled either on a frontend or on a backend, haproxy
7037 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007038 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007039 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007040 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007041 are not much aggressive in order to limit excessive use of splicing. This
7042 option requires splicing to be enabled at compile time, and may be globally
7043 disabled with the global option "nosplice". Since splice uses pipes, using it
7044 requires that there are enough spare pipes.
7045
7046 Important note: kernel-based TCP splicing is a Linux-specific feature which
7047 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
7048 transfer data between sockets without copying these data to user-space, thus
7049 providing noticeable performance gains and CPU cycles savings. Since many
7050 early implementations are buggy, corrupt data and/or are inefficient, this
7051 feature is not enabled by default, and it should be used with extreme care.
7052 While it is not possible to detect the correctness of an implementation,
7053 2.6.29 is the first version offering a properly working implementation. In
7054 case of doubt, splicing may be globally disabled using the global "nosplice"
7055 keyword.
7056
7057 Example :
7058 option splice-auto
7059
7060 If this option has been enabled in a "defaults" section, it can be disabled
7061 in a specific instance by prepending the "no" keyword before it.
7062
7063 See also : "option splice-request", "option splice-response", and global
7064 options "nosplice" and "maxpipes"
7065
7066
7067option splice-request
7068no option splice-request
7069 Enable or disable automatic kernel acceleration on sockets for requests
7070 May be used in sections : defaults | frontend | listen | backend
7071 yes | yes | yes | yes
7072 Arguments : none
7073
7074 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007075 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007076 the client to the server. It might still use the recv/send scheme if there
7077 are no spare pipes left. This option requires splicing to be enabled at
7078 compile time, and may be globally disabled with the global option "nosplice".
7079 Since splice uses pipes, using it requires that there are enough spare pipes.
7080
7081 Important note: see "option splice-auto" for usage limitations.
7082
7083 Example :
7084 option splice-request
7085
7086 If this option has been enabled in a "defaults" section, it can be disabled
7087 in a specific instance by prepending the "no" keyword before it.
7088
7089 See also : "option splice-auto", "option splice-response", and global options
7090 "nosplice" and "maxpipes"
7091
7092
7093option splice-response
7094no option splice-response
7095 Enable or disable automatic kernel acceleration on sockets for responses
7096 May be used in sections : defaults | frontend | listen | backend
7097 yes | yes | yes | yes
7098 Arguments : none
7099
7100 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007101 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007102 the server to the client. It might still use the recv/send scheme if there
7103 are no spare pipes left. This option requires splicing to be enabled at
7104 compile time, and may be globally disabled with the global option "nosplice".
7105 Since splice uses pipes, using it requires that there are enough spare pipes.
7106
7107 Important note: see "option splice-auto" for usage limitations.
7108
7109 Example :
7110 option splice-response
7111
7112 If this option has been enabled in a "defaults" section, it can be disabled
7113 in a specific instance by prepending the "no" keyword before it.
7114
7115 See also : "option splice-auto", "option splice-request", and global options
7116 "nosplice" and "maxpipes"
7117
7118
Christopher Fauletba7bc162016-11-07 21:07:38 +01007119option spop-check
7120 Use SPOP health checks for server testing
7121 May be used in sections : defaults | frontend | listen | backend
7122 no | no | no | yes
7123 Arguments : none
7124
7125 It is possible to test that the server correctly talks SPOP protocol instead
7126 of just testing that it accepts the TCP connection. When this option is set,
7127 a HELLO handshake is performed between HAProxy and the server, and the
7128 response is analyzed to check no error is reported.
7129
7130 Example :
7131 option spop-check
7132
7133 See also : "option httpchk"
7134
7135
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007136option srvtcpka
7137no option srvtcpka
7138 Enable or disable the sending of TCP keepalive packets on the server side
7139 May be used in sections : defaults | frontend | listen | backend
7140 yes | no | yes | yes
7141 Arguments : none
7142
7143 When there is a firewall or any session-aware component between a client and
7144 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007145 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007146 components decides to expire a session which has remained idle for too long.
7147
7148 Enabling socket-level TCP keep-alives makes the system regularly send packets
7149 to the other end of the connection, leaving it active. The delay between
7150 keep-alive probes is controlled by the system only and depends both on the
7151 operating system and its tuning parameters.
7152
7153 It is important to understand that keep-alive packets are neither emitted nor
7154 received at the application level. It is only the network stacks which sees
7155 them. For this reason, even if one side of the proxy already uses keep-alives
7156 to maintain its connection alive, those keep-alive packets will not be
7157 forwarded to the other side of the proxy.
7158
7159 Please note that this has nothing to do with HTTP keep-alive.
7160
7161 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
7162 server side of a connection, which should help when session expirations are
7163 noticed between HAProxy and a server.
7164
7165 If this option has been enabled in a "defaults" section, it can be disabled
7166 in a specific instance by prepending the "no" keyword before it.
7167
7168 See also : "option clitcpka", "option tcpka"
7169
7170
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007171option ssl-hello-chk
7172 Use SSLv3 client hello health checks for server testing
7173 May be used in sections : defaults | frontend | listen | backend
7174 yes | no | yes | yes
7175 Arguments : none
7176
7177 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
7178 possible to test that the server correctly talks SSL instead of just testing
7179 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
7180 SSLv3 client hello messages are sent once the connection is established to
7181 the server, and the response is analyzed to find an SSL server hello message.
7182 The server is considered valid only when the response contains this server
7183 hello message.
7184
7185 All servers tested till there correctly reply to SSLv3 client hello messages,
7186 and most servers tested do not even log the requests containing only hello
7187 messages, which is appreciable.
7188
Willy Tarreau763a95b2012-10-04 23:15:39 +02007189 Note that this check works even when SSL support was not built into haproxy
7190 because it forges the SSL message. When SSL support is available, it is best
7191 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007192
Willy Tarreau763a95b2012-10-04 23:15:39 +02007193 See also: "option httpchk", "check-ssl"
7194
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007195
Willy Tarreaued179852013-12-16 01:07:00 +01007196option tcp-check
7197 Perform health checks using tcp-check send/expect sequences
7198 May be used in sections: defaults | frontend | listen | backend
7199 yes | no | yes | yes
7200
7201 This health check method is intended to be combined with "tcp-check" command
7202 lists in order to support send/expect types of health check sequences.
7203
7204 TCP checks currently support 4 modes of operations :
7205 - no "tcp-check" directive : the health check only consists in a connection
7206 attempt, which remains the default mode.
7207
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007208 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007209 used to send a string along with a connection opening. With some
7210 protocols, it helps sending a "QUIT" message for example that prevents
7211 the server from logging a connection error for each health check. The
7212 check result will still be based on the ability to open the connection
7213 only.
7214
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007215 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007216 The connection is opened and haproxy waits for the server to present some
7217 contents which must validate some rules. The check result will be based
7218 on the matching between the contents and the rules. This is suited for
7219 POP, IMAP, SMTP, FTP, SSH, TELNET.
7220
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007221 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007222 used to test a hello-type protocol. HAProxy sends a message, the server
7223 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007224 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007225 suited for protocols which require a binding or a request/response model.
7226 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7227 already all have their dedicated checks with a deeper understanding of
7228 the respective protocols.
7229 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007230 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007231
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007232 A fifth mode can be used to insert comments in different steps of the
7233 script.
7234
7235 For each tcp-check rule you create, you can add a "comment" directive,
7236 followed by a string. This string will be reported in the log and stderr
7237 in debug mode. It is useful to make user-friendly error reporting.
7238 The "comment" is of course optional.
7239
7240
Willy Tarreaued179852013-12-16 01:07:00 +01007241 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007242 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007243 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007244 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007245
Davor Ocelice9ed2812017-12-25 17:49:28 +01007246 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007247 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007248 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007249
7250 # look for the redis master server after ensuring it speaks well
7251 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007252 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007253 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007254 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007255 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007256 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007257 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007258 tcp-check send info\ replication\r\n
7259 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007260 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007261 tcp-check send QUIT\r\n
7262 tcp-check expect string +OK
7263
Davor Ocelice9ed2812017-12-25 17:49:28 +01007264 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007265 (send many headers before analyzing)
7266 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007267 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007268 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7269 tcp-check send Host:\ www.mydomain.com\r\n
7270 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7271 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007272 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007273
7274
7275 See also : "tcp-check expect", "tcp-check send"
7276
7277
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007278option tcp-smart-accept
7279no option tcp-smart-accept
7280 Enable or disable the saving of one ACK packet during the accept sequence
7281 May be used in sections : defaults | frontend | listen | backend
7282 yes | yes | yes | no
7283 Arguments : none
7284
7285 When an HTTP connection request comes in, the system acknowledges it on
7286 behalf of HAProxy, then the client immediately sends its request, and the
7287 system acknowledges it too while it is notifying HAProxy about the new
7288 connection. HAProxy then reads the request and responds. This means that we
7289 have one TCP ACK sent by the system for nothing, because the request could
7290 very well be acknowledged by HAProxy when it sends its response.
7291
7292 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7293 sending this useless ACK on platforms which support it (currently at least
7294 Linux). It must not cause any problem, because the system will send it anyway
7295 after 40 ms if the response takes more time than expected to come.
7296
7297 During complex network debugging sessions, it may be desirable to disable
7298 this optimization because delayed ACKs can make troubleshooting more complex
7299 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007300 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007301
7302 It is also possible to force it for non-HTTP proxies by simply specifying
7303 "option tcp-smart-accept". For instance, it can make sense with some services
7304 such as SMTP where the server speaks first.
7305
7306 It is recommended to avoid forcing this option in a defaults section. In case
7307 of doubt, consider setting it back to automatic values by prepending the
7308 "default" keyword before it, or disabling it using the "no" keyword.
7309
Willy Tarreaud88edf22009-06-14 15:48:17 +02007310 See also : "option tcp-smart-connect"
7311
7312
7313option tcp-smart-connect
7314no option tcp-smart-connect
7315 Enable or disable the saving of one ACK packet during the connect sequence
7316 May be used in sections : defaults | frontend | listen | backend
7317 yes | no | yes | yes
7318 Arguments : none
7319
7320 On certain systems (at least Linux), HAProxy can ask the kernel not to
7321 immediately send an empty ACK upon a connection request, but to directly
7322 send the buffer request instead. This saves one packet on the network and
7323 thus boosts performance. It can also be useful for some servers, because they
7324 immediately get the request along with the incoming connection.
7325
7326 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7327 It is not enabled by default because it makes network troubleshooting more
7328 complex.
7329
7330 It only makes sense to enable it with protocols where the client speaks first
7331 such as HTTP. In other situations, if there is no data to send in place of
7332 the ACK, a normal ACK is sent.
7333
7334 If this option has been enabled in a "defaults" section, it can be disabled
7335 in a specific instance by prepending the "no" keyword before it.
7336
7337 See also : "option tcp-smart-accept"
7338
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007339
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007340option tcpka
7341 Enable or disable the sending of TCP keepalive packets on both sides
7342 May be used in sections : defaults | frontend | listen | backend
7343 yes | yes | yes | yes
7344 Arguments : none
7345
7346 When there is a firewall or any session-aware component between a client and
7347 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007348 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007349 components decides to expire a session which has remained idle for too long.
7350
7351 Enabling socket-level TCP keep-alives makes the system regularly send packets
7352 to the other end of the connection, leaving it active. The delay between
7353 keep-alive probes is controlled by the system only and depends both on the
7354 operating system and its tuning parameters.
7355
7356 It is important to understand that keep-alive packets are neither emitted nor
7357 received at the application level. It is only the network stacks which sees
7358 them. For this reason, even if one side of the proxy already uses keep-alives
7359 to maintain its connection alive, those keep-alive packets will not be
7360 forwarded to the other side of the proxy.
7361
7362 Please note that this has nothing to do with HTTP keep-alive.
7363
7364 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7365 the client and server sides of a connection. Note that this is meaningful
7366 only in "defaults" or "listen" sections. If this option is used in a
7367 frontend, only the client side will get keep-alives, and if this option is
7368 used in a backend, only the server side will get keep-alives. For this
7369 reason, it is strongly recommended to explicitly use "option clitcpka" and
7370 "option srvtcpka" when the configuration is split between frontends and
7371 backends.
7372
7373 See also : "option clitcpka", "option srvtcpka"
7374
Willy Tarreau844e3c52008-01-11 16:28:18 +01007375
7376option tcplog
7377 Enable advanced logging of TCP connections with session state and timers
7378 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007379 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007380 Arguments : none
7381
7382 By default, the log output format is very poor, as it only contains the
7383 source and destination addresses, and the instance name. By specifying
7384 "option tcplog", each log line turns into a much richer format including, but
7385 not limited to, the connection timers, the session status, the connections
7386 numbers, the frontend, backend and server name, and of course the source
7387 address and ports. This option is useful for pure TCP proxies in order to
7388 find which of the client or server disconnects or times out. For normal HTTP
7389 proxies, it's better to use "option httplog" which is even more complete.
7390
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007391 "option tcplog" overrides any previous "log-format" directive.
7392
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007393 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007394
7395
Willy Tarreau844e3c52008-01-11 16:28:18 +01007396option transparent
7397no option transparent
7398 Enable client-side transparent proxying
7399 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007400 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007401 Arguments : none
7402
7403 This option was introduced in order to provide layer 7 persistence to layer 3
7404 load balancers. The idea is to use the OS's ability to redirect an incoming
7405 connection for a remote address to a local process (here HAProxy), and let
7406 this process know what address was initially requested. When this option is
7407 used, sessions without cookies will be forwarded to the original destination
7408 IP address of the incoming request (which should match that of another
7409 equipment), while requests with cookies will still be forwarded to the
7410 appropriate server.
7411
7412 Note that contrary to a common belief, this option does NOT make HAProxy
7413 present the client's IP to the server when establishing the connection.
7414
Willy Tarreaua1146052011-03-01 09:51:54 +01007415 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007416 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007417
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007418
Simon Horman98637e52014-06-20 12:30:16 +09007419external-check command <command>
7420 Executable to run when performing an external-check
7421 May be used in sections : defaults | frontend | listen | backend
7422 yes | no | yes | yes
7423
7424 Arguments :
7425 <command> is the external command to run
7426
Simon Horman98637e52014-06-20 12:30:16 +09007427 The arguments passed to the to the command are:
7428
Cyril Bonté777be862014-12-02 21:21:35 +01007429 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007430
Cyril Bonté777be862014-12-02 21:21:35 +01007431 The <proxy_address> and <proxy_port> are derived from the first listener
7432 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7433 listener the proxy_address will be the path of the socket and the
7434 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7435 possible to determine a listener, and both <proxy_address> and <proxy_port>
7436 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007437
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007438 Some values are also provided through environment variables.
7439
7440 Environment variables :
7441 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7442 applicable, for example in a "backend" section).
7443
7444 HAPROXY_PROXY_ID The backend id.
7445
7446 HAPROXY_PROXY_NAME The backend name.
7447
7448 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7449 applicable, for example in a "backend" section or
7450 for a UNIX socket).
7451
7452 HAPROXY_SERVER_ADDR The server address.
7453
7454 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7455
7456 HAPROXY_SERVER_ID The server id.
7457
7458 HAPROXY_SERVER_MAXCONN The server max connections.
7459
7460 HAPROXY_SERVER_NAME The server name.
7461
7462 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7463 socket).
7464
7465 PATH The PATH environment variable used when executing
7466 the command may be set using "external-check path".
7467
William Lallemand4d03e432019-06-14 15:35:37 +02007468 See also "2.3. Environment variables" for other variables.
7469
Simon Horman98637e52014-06-20 12:30:16 +09007470 If the command executed and exits with a zero status then the check is
7471 considered to have passed, otherwise the check is considered to have
7472 failed.
7473
7474 Example :
7475 external-check command /bin/true
7476
7477 See also : "external-check", "option external-check", "external-check path"
7478
7479
7480external-check path <path>
7481 The value of the PATH environment variable used when running an external-check
7482 May be used in sections : defaults | frontend | listen | backend
7483 yes | no | yes | yes
7484
7485 Arguments :
7486 <path> is the path used when executing external command to run
7487
7488 The default path is "".
7489
7490 Example :
7491 external-check path "/usr/bin:/bin"
7492
7493 See also : "external-check", "option external-check",
7494 "external-check command"
7495
7496
Emeric Brun647caf12009-06-30 17:57:00 +02007497persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007498persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007499 Enable RDP cookie-based persistence
7500 May be used in sections : defaults | frontend | listen | backend
7501 yes | no | yes | yes
7502 Arguments :
7503 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007504 default cookie name "msts" will be used. There currently is no
7505 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007506
7507 This statement enables persistence based on an RDP cookie. The RDP cookie
7508 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007509 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007510 and if an RDP cookie is found, it is decoded. If it matches a known server
7511 which is still UP (or if "option persist" is set), then the connection is
7512 forwarded to this server.
7513
7514 Note that this only makes sense in a TCP backend, but for this to work, the
7515 frontend must have waited long enough to ensure that an RDP cookie is present
7516 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007517 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007518 a single "listen" section.
7519
Willy Tarreau61e28f22010-05-16 22:31:05 +02007520 Also, it is important to understand that the terminal server will emit this
7521 RDP cookie only if it is configured for "token redirection mode", which means
7522 that the "IP address redirection" option is disabled.
7523
Emeric Brun647caf12009-06-30 17:57:00 +02007524 Example :
7525 listen tse-farm
7526 bind :3389
7527 # wait up to 5s for an RDP cookie in the request
7528 tcp-request inspect-delay 5s
7529 tcp-request content accept if RDP_COOKIE
7530 # apply RDP cookie persistence
7531 persist rdp-cookie
7532 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007533 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007534 balance rdp-cookie
7535 server srv1 1.1.1.1:3389
7536 server srv2 1.1.1.2:3389
7537
Simon Hormanab814e02011-06-24 14:50:20 +09007538 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7539 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007540
7541
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007542rate-limit sessions <rate>
7543 Set a limit on the number of new sessions accepted per second on a frontend
7544 May be used in sections : defaults | frontend | listen | backend
7545 yes | yes | yes | no
7546 Arguments :
7547 <rate> The <rate> parameter is an integer designating the maximum number
7548 of new sessions per second to accept on the frontend.
7549
7550 When the frontend reaches the specified number of new sessions per second, it
7551 stops accepting new connections until the rate drops below the limit again.
7552 During this time, the pending sessions will be kept in the socket's backlog
7553 (in system buffers) and haproxy will not even be aware that sessions are
7554 pending. When applying very low limit on a highly loaded service, it may make
7555 sense to increase the socket's backlog using the "backlog" keyword.
7556
7557 This feature is particularly efficient at blocking connection-based attacks
7558 or service abuse on fragile servers. Since the session rate is measured every
7559 millisecond, it is extremely accurate. Also, the limit applies immediately,
7560 no delay is needed at all to detect the threshold.
7561
7562 Example : limit the connection rate on SMTP to 10 per second max
7563 listen smtp
7564 mode tcp
7565 bind :25
7566 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007567 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007568
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007569 Note : when the maximum rate is reached, the frontend's status is not changed
7570 but its sockets appear as "WAITING" in the statistics if the
7571 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007572
7573 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7574
7575
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007576redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7577redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7578redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007579 Return an HTTP redirection if/unless a condition is matched
7580 May be used in sections : defaults | frontend | listen | backend
7581 no | yes | yes | yes
7582
7583 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007584 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007585
Willy Tarreau0140f252008-11-19 21:07:09 +01007586 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007587 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007588 the HTTP "Location" header. When used in an "http-request" rule,
7589 <loc> value follows the log-format rules and can include some
7590 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007591
7592 <pfx> With "redirect prefix", the "Location" header is built from the
7593 concatenation of <pfx> and the complete URI path, including the
7594 query string, unless the "drop-query" option is specified (see
7595 below). As a special case, if <pfx> equals exactly "/", then
7596 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007597 redirect to the same URL (for instance, to insert a cookie). When
7598 used in an "http-request" rule, <pfx> value follows the log-format
7599 rules and can include some dynamic values (see Custom Log Format
7600 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007601
7602 <sch> With "redirect scheme", then the "Location" header is built by
7603 concatenating <sch> with "://" then the first occurrence of the
7604 "Host" header, and then the URI path, including the query string
7605 unless the "drop-query" option is specified (see below). If no
7606 path is found or if the path is "*", then "/" is used instead. If
7607 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007608 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007609 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007610 HTTPS. When used in an "http-request" rule, <sch> value follows
7611 the log-format rules and can include some dynamic values (see
7612 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007613
7614 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007615 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7616 with 302 used by default if no code is specified. 301 means
7617 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007618 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007619 cache the redirection. 303 is equivalent to 302 except that the
7620 browser will fetch the location with a GET method. 307 is just
7621 like 302 but makes it clear that the same method must be reused.
7622 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007623
7624 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007625 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007626
7627 - "drop-query"
7628 When this keyword is used in a prefix-based redirection, then the
7629 location will be set without any possible query-string, which is useful
7630 for directing users to a non-secure page for instance. It has no effect
7631 with a location-type redirect.
7632
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007633 - "append-slash"
7634 This keyword may be used in conjunction with "drop-query" to redirect
7635 users who use a URL not ending with a '/' to the same one with the '/'.
7636 It can be useful to ensure that search engines will only see one URL.
7637 For this, a return code 301 is preferred.
7638
Willy Tarreau0140f252008-11-19 21:07:09 +01007639 - "set-cookie NAME[=value]"
7640 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7641 to the response. This is sometimes used to indicate that a user has
7642 been seen, for instance to protect against some types of DoS. No other
7643 cookie option is added, so the cookie will be a session cookie. Note
7644 that for a browser, a sole cookie name without an equal sign is
7645 different from a cookie with an equal sign.
7646
7647 - "clear-cookie NAME[=]"
7648 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7649 with the "Max-Age" attribute set to zero. This will tell the browser to
7650 delete this cookie. It is useful for instance on logout pages. It is
7651 important to note that clearing the cookie "NAME" will not remove a
7652 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7653 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007654
7655 Example: move the login URL only to HTTPS.
7656 acl clear dst_port 80
7657 acl secure dst_port 8080
7658 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007659 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007660 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007661 acl cookie_set hdr_sub(cookie) SEEN=1
7662
7663 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007664 redirect prefix https://mysite.com if login_page !secure
7665 redirect prefix http://mysite.com drop-query if login_page !uid_given
7666 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007667 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007668
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007669 Example: send redirects for request for articles without a '/'.
7670 acl missing_slash path_reg ^/article/[^/]*$
7671 redirect code 301 prefix / drop-query append-slash if missing_slash
7672
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007673 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007674 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007675
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007676 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007677 http-request redirect code 301 location \
7678 http://www.%[hdr(host)]%[capture.req.uri] \
7679 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007680
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007681 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007682
Willy Tarreau303c0352008-01-17 19:01:39 +01007683
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007684retries <value>
7685 Set the number of retries to perform on a server after a connection failure
7686 May be used in sections: defaults | frontend | listen | backend
7687 yes | no | yes | yes
7688 Arguments :
7689 <value> is the number of times a connection attempt should be retried on
7690 a server when a connection either is refused or times out. The
7691 default value is 3.
7692
7693 It is important to understand that this value applies to the number of
7694 connection attempts, not full requests. When a connection has effectively
7695 been established to a server, there will be no more retry.
7696
7697 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007698 a turn-around timer of min("timeout connect", one second) is applied before
7699 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007700
7701 When "option redispatch" is set, the last retry may be performed on another
7702 server even if a cookie references a different server.
7703
7704 See also : "option redispatch"
7705
7706
Olivier Houcharda254a372019-04-05 15:30:12 +02007707retry-on [list of keywords]
7708 Specify when to attempt to automatically retry a failed request
7709 May be used in sections: defaults | frontend | listen | backend
7710 yes | no | yes | yes
7711 Arguments :
7712 <keywords> is a list of keywords or HTTP status codes, each representing a
7713 type of failure event on which an attempt to retry the request
7714 is desired. Please read the notes at the bottom before changing
7715 this setting. The following keywords are supported :
7716
7717 none never retry
7718
7719 conn-failure retry when the connection or the SSL handshake failed
7720 and the request could not be sent. This is the default.
7721
7722 empty-response retry when the server connection was closed after part
7723 of the request was sent, and nothing was received from
7724 the server. This type of failure may be caused by the
7725 request timeout on the server side, poor network
7726 condition, or a server crash or restart while
7727 processing the request.
7728
Olivier Houcharde3249a92019-05-03 23:01:47 +02007729 junk-response retry when the server returned something not looking
7730 like a complete HTTP response. This includes partial
7731 responses headers as well as non-HTTP contents. It
7732 usually is a bad idea to retry on such events, which
7733 may be caused a configuration issue (wrong server port)
7734 or by the request being harmful to the server (buffer
7735 overflow attack for example).
7736
Olivier Houcharda254a372019-04-05 15:30:12 +02007737 response-timeout the server timeout stroke while waiting for the server
7738 to respond to the request. This may be caused by poor
7739 network condition, the reuse of an idle connection
7740 which has expired on the path, or by the request being
7741 extremely expensive to process. It generally is a bad
7742 idea to retry on such events on servers dealing with
7743 heavy database processing (full scans, etc) as it may
7744 amplify denial of service attacks.
7745
Olivier Houchard865d8392019-05-03 22:46:27 +02007746 0rtt-rejected retry requests which were sent over early data and were
7747 rejected by the server. These requests are generally
7748 considered to be safe to retry.
7749
Olivier Houcharda254a372019-04-05 15:30:12 +02007750 <status> any HTTP status code among "404" (Not Found), "408"
7751 (Request Timeout), "425" (Too Early), "500" (Server
7752 Error), "501" (Not Implemented), "502" (Bad Gateway),
7753 "503" (Service Unavailable), "504" (Gateway Timeout).
7754
Olivier Houchardddf0e032019-05-10 18:05:40 +02007755 all-retryable-errors
7756 retry request for any error that are considered
7757 retryable. This currently activates "conn-failure",
7758 "empty-response", "junk-response", "response-timeout",
7759 "0rtt-rejected", "500", "502", "503", and "504".
7760
Olivier Houcharda254a372019-04-05 15:30:12 +02007761 Using this directive replaces any previous settings with the new ones; it is
7762 not cumulative.
7763
7764 Please note that using anything other than "none" and "conn-failure" requires
7765 to allocate a buffer and copy the whole request into it, so it has memory and
7766 performance impacts. Requests not fitting in a single buffer will never be
7767 retried (see the global tune.bufsize setting).
7768
7769 You have to make sure the application has a replay protection mechanism built
7770 in such as a unique transaction IDs passed in requests, or that replaying the
7771 same request has no consequence, or it is very dangerous to use any retry-on
7772 value beside "conn-failure" and "none". Static file servers and caches are
7773 generally considered safe against any type of retry. Using a status code can
7774 be useful to quickly leave a server showing an abnormal behavior (out of
7775 memory, file system issues, etc), but in this case it may be a good idea to
7776 immediately redispatch the connection to another server (please see "option
7777 redispatch" for this). Last, it is important to understand that most causes
7778 of failures are the requests themselves and that retrying a request causing a
7779 server to misbehave will often make the situation even worse for this server,
7780 or for the whole service in case of redispatch.
7781
7782 Unless you know exactly how the application deals with replayed requests, you
7783 should not use this directive.
7784
7785 The default is "conn-failure".
7786
7787 See also: "retries", "option redispatch", "tune.bufsize"
7788
David du Colombier486df472011-03-17 10:40:26 +01007789server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007790 Declare a server in a backend
7791 May be used in sections : defaults | frontend | listen | backend
7792 no | no | yes | yes
7793 Arguments :
7794 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007795 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007796 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007797
David du Colombier486df472011-03-17 10:40:26 +01007798 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7799 resolvable hostname is supported, but this name will be resolved
7800 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7801 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007802 address as the one from the client connection. This is useful in
7803 transparent proxy architectures where the client's connection is
7804 intercepted and haproxy must forward to the original destination
7805 address. This is more or less what the "transparent" keyword does
7806 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007807 to report statistics. Optionally, an address family prefix may be
7808 used before the address to force the family regardless of the
7809 address format, which can be useful to specify a path to a unix
7810 socket with no slash ('/'). Currently supported prefixes are :
7811 - 'ipv4@' -> address is always IPv4
7812 - 'ipv6@' -> address is always IPv6
7813 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007814 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02007815 - 'sockpair@' -> address is the FD of a connected unix
7816 socket or of a socketpair. During a connection, the
7817 backend creates a pair of connected sockets, and passes
7818 one of them over the FD. The bind part will use the
7819 received socket as the client FD. Should be used
7820 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02007821 You may want to reference some environment variables in the
7822 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007823 variables. The "init-addr" setting can be used to modify the way
7824 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007825
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007826 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007827 be sent to this port. If unset, the same port the client
7828 connected to will be used. The port may also be prefixed by a "+"
7829 or a "-". In this case, the server's port will be determined by
7830 adding this value to the client's port.
7831
7832 <param*> is a list of parameters for this server. The "server" keywords
7833 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007834 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007835
7836 Examples :
7837 server first 10.1.1.1:1080 cookie first check inter 1000
7838 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007839 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007840 server backup "${SRV_BACKUP}:1080" backup
7841 server www1_dc1 "${LAN_DC1}.101:80"
7842 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007843
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007844 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7845 sun_path length is used for the address length. Some other programs
7846 such as socat use the string length only by default. Pass the option
7847 ",unix-tightsocklen=0" to any abstract socket definition in socat to
7848 make it compatible with HAProxy's.
7849
Mark Lamourinec2247f02012-01-04 13:02:01 -05007850 See also: "default-server", "http-send-name-header" and section 5 about
7851 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007852
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007853server-state-file-name [<file>]
7854 Set the server state file to read, load and apply to servers available in
7855 this backend. It only applies when the directive "load-server-state-from-file"
7856 is set to "local". When <file> is not provided or if this directive is not
7857 set, then backend name is used. If <file> starts with a slash '/', then it is
7858 considered as an absolute path. Otherwise, <file> is concatenated to the
7859 global directive "server-state-file-base".
7860
7861 Example: the minimal configuration below would make HAProxy look for the
7862 state server file '/etc/haproxy/states/bk':
7863
7864 global
7865 server-state-file-base /etc/haproxy/states
7866
7867 backend bk
7868 load-server-state-from-file
7869
7870 See also: "server-state-file-base", "load-server-state-from-file", and
7871 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007872
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02007873server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
7874 Set a template to initialize servers with shared parameters.
7875 The names of these servers are built from <prefix> and <num | range> parameters.
7876 May be used in sections : defaults | frontend | listen | backend
7877 no | no | yes | yes
7878
7879 Arguments:
7880 <prefix> A prefix for the server names to be built.
7881
7882 <num | range>
7883 If <num> is provided, this template initializes <num> servers
7884 with 1 up to <num> as server name suffixes. A range of numbers
7885 <num_low>-<num_high> may also be used to use <num_low> up to
7886 <num_high> as server name suffixes.
7887
7888 <fqdn> A FQDN for all the servers this template initializes.
7889
7890 <port> Same meaning as "server" <port> argument (see "server" keyword).
7891
7892 <params*>
7893 Remaining server parameters among all those supported by "server"
7894 keyword.
7895
7896 Examples:
7897 # Initializes 3 servers with srv1, srv2 and srv3 as names,
7898 # google.com as FQDN, and health-check enabled.
7899 server-template srv 1-3 google.com:80 check
7900
7901 # or
7902 server-template srv 3 google.com:80 check
7903
7904 # would be equivalent to:
7905 server srv1 google.com:80 check
7906 server srv2 google.com:80 check
7907 server srv3 google.com:80 check
7908
7909
7910
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007911source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02007912source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007913source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007914 Set the source address for outgoing connections
7915 May be used in sections : defaults | frontend | listen | backend
7916 yes | no | yes | yes
7917 Arguments :
7918 <addr> is the IPv4 address HAProxy will bind to before connecting to a
7919 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01007920
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007921 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01007922 the most appropriate address to reach its destination. Optionally
7923 an address family prefix may be used before the address to force
7924 the family regardless of the address format, which can be useful
7925 to specify a path to a unix socket with no slash ('/'). Currently
7926 supported prefixes are :
7927 - 'ipv4@' -> address is always IPv4
7928 - 'ipv6@' -> address is always IPv6
7929 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007930 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02007931 You may want to reference some environment variables in the
7932 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007933
7934 <port> is an optional port. It is normally not needed but may be useful
7935 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02007936 the system will select a free port. Note that port ranges are not
7937 supported in the backend. If you want to force port ranges, you
7938 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007939
7940 <addr2> is the IP address to present to the server when connections are
7941 forwarded in full transparent proxy mode. This is currently only
7942 supported on some patched Linux kernels. When this address is
7943 specified, clients connecting to the server will be presented
7944 with this address, while health checks will still use the address
7945 <addr>.
7946
7947 <port2> is the optional port to present to the server when connections
7948 are forwarded in full transparent proxy mode (see <addr2> above).
7949 The default value of zero means the system will select a free
7950 port.
7951
Willy Tarreaubce70882009-09-07 11:51:47 +02007952 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
7953 This is the name of a comma-separated header list which can
7954 contain multiple IP addresses. By default, the last occurrence is
7955 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01007956 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02007957 by previous proxy, typically Stunnel. In order to use another
7958 occurrence from the last one, please see the <occ> parameter
7959 below. When the header (or occurrence) is not found, no binding
7960 is performed so that the proxy's default IP address is used. Also
7961 keep in mind that the header name is case insensitive, as for any
7962 HTTP header.
7963
7964 <occ> is the occurrence number of a value to be used in a multi-value
7965 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007966 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02007967 address. Positive values indicate a position from the first
7968 occurrence, 1 being the first one. Negative values indicate
7969 positions relative to the last one, -1 being the last one. This
7970 is helpful for situations where an X-Forwarded-For header is set
7971 at the entry point of an infrastructure and must be used several
7972 proxy layers away. When this value is not specified, -1 is
7973 assumed. Passing a zero here disables the feature.
7974
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007975 <name> is an optional interface name to which to bind to for outgoing
7976 traffic. On systems supporting this features (currently, only
7977 Linux), this allows one to bind all traffic to the server to
7978 this interface even if it is not the one the system would select
7979 based on routing tables. This should be used with extreme care.
7980 Note that using this option requires root privileges.
7981
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007982 The "source" keyword is useful in complex environments where a specific
7983 address only is allowed to connect to the servers. It may be needed when a
7984 private address must be used through a public gateway for instance, and it is
7985 known that the system cannot determine the adequate source address by itself.
7986
7987 An extension which is available on certain patched Linux kernels may be used
7988 through the "usesrc" optional keyword. It makes it possible to connect to the
7989 servers with an IP address which does not belong to the system itself. This
7990 is called "full transparent proxy mode". For this to work, the destination
7991 servers have to route their traffic back to this address through the machine
7992 running HAProxy, and IP forwarding must generally be enabled on this machine.
7993
7994 In this "full transparent proxy" mode, it is possible to force a specific IP
7995 address to be presented to the servers. This is not much used in fact. A more
7996 common use is to tell HAProxy to present the client's IP address. For this,
7997 there are two methods :
7998
7999 - present the client's IP and port addresses. This is the most transparent
8000 mode, but it can cause problems when IP connection tracking is enabled on
8001 the machine, because a same connection may be seen twice with different
8002 states. However, this solution presents the huge advantage of not
8003 limiting the system to the 64k outgoing address+port couples, because all
8004 of the client ranges may be used.
8005
8006 - present only the client's IP address and select a spare port. This
8007 solution is still quite elegant but slightly less transparent (downstream
8008 firewalls logs will not match upstream's). It also presents the downside
8009 of limiting the number of concurrent connections to the usual 64k ports.
8010 However, since the upstream and downstream ports are different, local IP
8011 connection tracking on the machine will not be upset by the reuse of the
8012 same session.
8013
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008014 This option sets the default source for all servers in the backend. It may
8015 also be specified in a "defaults" section. Finer source address specification
8016 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008017 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008018
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008019 In order to work, "usesrc" requires root privileges.
8020
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008021 Examples :
8022 backend private
8023 # Connect to the servers using our 192.168.1.200 source address
8024 source 192.168.1.200
8025
8026 backend transparent_ssl1
8027 # Connect to the SSL farm from the client's source address
8028 source 192.168.1.200 usesrc clientip
8029
8030 backend transparent_ssl2
8031 # Connect to the SSL farm from the client's source address and port
8032 # not recommended if IP conntrack is present on the local machine.
8033 source 192.168.1.200 usesrc client
8034
8035 backend transparent_ssl3
8036 # Connect to the SSL farm from the client's source address. It
8037 # is more conntrack-friendly.
8038 source 192.168.1.200 usesrc clientip
8039
8040 backend transparent_smtp
8041 # Connect to the SMTP farm from the client's source address/port
8042 # with Tproxy version 4.
8043 source 0.0.0.0 usesrc clientip
8044
Willy Tarreaubce70882009-09-07 11:51:47 +02008045 backend transparent_http
8046 # Connect to the servers using the client's IP as seen by previous
8047 # proxy.
8048 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8049
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008050 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008051 the Linux kernel on www.balabit.com, the "bind" keyword.
8052
Willy Tarreau844e3c52008-01-11 16:28:18 +01008053
Cyril Bonté66c327d2010-10-12 00:14:37 +02008054stats admin { if | unless } <cond>
8055 Enable statistics admin level if/unless a condition is matched
8056 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008057 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008058
8059 This statement enables the statistics admin level if/unless a condition is
8060 matched.
8061
8062 The admin level allows to enable/disable servers from the web interface. By
8063 default, statistics page is read-only for security reasons.
8064
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008065 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8066 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008067 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008068
Cyril Bonté23b39d92011-02-10 22:54:44 +01008069 Currently, the POST request is limited to the buffer size minus the reserved
8070 buffer space, which means that if the list of servers is too long, the
8071 request won't be processed. It is recommended to alter few servers at a
8072 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008073
8074 Example :
8075 # statistics admin level only for localhost
8076 backend stats_localhost
8077 stats enable
8078 stats admin if LOCALHOST
8079
8080 Example :
8081 # statistics admin level always enabled because of the authentication
8082 backend stats_auth
8083 stats enable
8084 stats auth admin:AdMiN123
8085 stats admin if TRUE
8086
8087 Example :
8088 # statistics admin level depends on the authenticated user
8089 userlist stats-auth
8090 group admin users admin
8091 user admin insecure-password AdMiN123
8092 group readonly users haproxy
8093 user haproxy insecure-password haproxy
8094
8095 backend stats_auth
8096 stats enable
8097 acl AUTH http_auth(stats-auth)
8098 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8099 stats http-request auth unless AUTH
8100 stats admin if AUTH_ADMIN
8101
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008102 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8103 "bind-process", section 3.4 about userlists and section 7 about
8104 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008105
8106
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008107stats auth <user>:<passwd>
8108 Enable statistics with authentication and grant access to an account
8109 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008110 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008111 Arguments :
8112 <user> is a user name to grant access to
8113
8114 <passwd> is the cleartext password associated to this user
8115
8116 This statement enables statistics with default settings, and restricts access
8117 to declared users only. It may be repeated as many times as necessary to
8118 allow as many users as desired. When a user tries to access the statistics
8119 without a valid account, a "401 Forbidden" response will be returned so that
8120 the browser asks the user to provide a valid user and password. The real
8121 which will be returned to the browser is configurable using "stats realm".
8122
8123 Since the authentication method is HTTP Basic Authentication, the passwords
8124 circulate in cleartext on the network. Thus, it was decided that the
8125 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008126 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008127
8128 It is also possible to reduce the scope of the proxies which appear in the
8129 report using "stats scope".
8130
8131 Though this statement alone is enough to enable statistics reporting, it is
8132 recommended to set all other settings in order to avoid relying on default
8133 unobvious parameters.
8134
8135 Example :
8136 # public access (limited to this backend only)
8137 backend public_www
8138 server srv1 192.168.0.1:80
8139 stats enable
8140 stats hide-version
8141 stats scope .
8142 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008143 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008144 stats auth admin1:AdMiN123
8145 stats auth admin2:AdMiN321
8146
8147 # internal monitoring access (unlimited)
8148 backend private_monitoring
8149 stats enable
8150 stats uri /admin?stats
8151 stats refresh 5s
8152
8153 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8154
8155
8156stats enable
8157 Enable statistics reporting with default settings
8158 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008159 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008160 Arguments : none
8161
8162 This statement enables statistics reporting with default settings defined
8163 at build time. Unless stated otherwise, these settings are used :
8164 - stats uri : /haproxy?stats
8165 - stats realm : "HAProxy Statistics"
8166 - stats auth : no authentication
8167 - stats scope : no restriction
8168
8169 Though this statement alone is enough to enable statistics reporting, it is
8170 recommended to set all other settings in order to avoid relying on default
8171 unobvious parameters.
8172
8173 Example :
8174 # public access (limited to this backend only)
8175 backend public_www
8176 server srv1 192.168.0.1:80
8177 stats enable
8178 stats hide-version
8179 stats scope .
8180 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008181 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008182 stats auth admin1:AdMiN123
8183 stats auth admin2:AdMiN321
8184
8185 # internal monitoring access (unlimited)
8186 backend private_monitoring
8187 stats enable
8188 stats uri /admin?stats
8189 stats refresh 5s
8190
8191 See also : "stats auth", "stats realm", "stats uri"
8192
8193
Willy Tarreaud63335a2010-02-26 12:56:52 +01008194stats hide-version
8195 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008196 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008197 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008198 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008199
Willy Tarreaud63335a2010-02-26 12:56:52 +01008200 By default, the stats page reports some useful status information along with
8201 the statistics. Among them is HAProxy's version. However, it is generally
8202 considered dangerous to report precise version to anyone, as it can help them
8203 target known weaknesses with specific attacks. The "stats hide-version"
8204 statement removes the version from the statistics report. This is recommended
8205 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008206
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008207 Though this statement alone is enough to enable statistics reporting, it is
8208 recommended to set all other settings in order to avoid relying on default
8209 unobvious parameters.
8210
Willy Tarreaud63335a2010-02-26 12:56:52 +01008211 Example :
8212 # public access (limited to this backend only)
8213 backend public_www
8214 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008215 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008216 stats hide-version
8217 stats scope .
8218 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008219 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008220 stats auth admin1:AdMiN123
8221 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008222
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008223 # internal monitoring access (unlimited)
8224 backend private_monitoring
8225 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008226 stats uri /admin?stats
8227 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008228
Willy Tarreaud63335a2010-02-26 12:56:52 +01008229 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008230
Willy Tarreau983e01e2010-01-11 18:42:06 +01008231
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008232stats http-request { allow | deny | auth [realm <realm>] }
8233 [ { if | unless } <condition> ]
8234 Access control for statistics
8235
8236 May be used in sections: defaults | frontend | listen | backend
8237 no | no | yes | yes
8238
8239 As "http-request", these set of options allow to fine control access to
8240 statistics. Each option may be followed by if/unless and acl.
8241 First option with matched condition (or option without condition) is final.
8242 For "deny" a 403 error will be returned, for "allow" normal processing is
8243 performed, for "auth" a 401/407 error code is returned so the client
8244 should be asked to enter a username and password.
8245
8246 There is no fixed limit to the number of http-request statements per
8247 instance.
8248
8249 See also : "http-request", section 3.4 about userlists and section 7
8250 about ACL usage.
8251
8252
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008253stats realm <realm>
8254 Enable statistics and set authentication realm
8255 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008256 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008257 Arguments :
8258 <realm> is the name of the HTTP Basic Authentication realm reported to
8259 the browser. The browser uses it to display it in the pop-up
8260 inviting the user to enter a valid username and password.
8261
8262 The realm is read as a single word, so any spaces in it should be escaped
8263 using a backslash ('\').
8264
8265 This statement is useful only in conjunction with "stats auth" since it is
8266 only related to authentication.
8267
8268 Though this statement alone is enough to enable statistics reporting, it is
8269 recommended to set all other settings in order to avoid relying on default
8270 unobvious parameters.
8271
8272 Example :
8273 # public access (limited to this backend only)
8274 backend public_www
8275 server srv1 192.168.0.1:80
8276 stats enable
8277 stats hide-version
8278 stats scope .
8279 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008280 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008281 stats auth admin1:AdMiN123
8282 stats auth admin2:AdMiN321
8283
8284 # internal monitoring access (unlimited)
8285 backend private_monitoring
8286 stats enable
8287 stats uri /admin?stats
8288 stats refresh 5s
8289
8290 See also : "stats auth", "stats enable", "stats uri"
8291
8292
8293stats refresh <delay>
8294 Enable statistics with automatic refresh
8295 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008296 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008297 Arguments :
8298 <delay> is the suggested refresh delay, specified in seconds, which will
8299 be returned to the browser consulting the report page. While the
8300 browser is free to apply any delay, it will generally respect it
8301 and refresh the page this every seconds. The refresh interval may
8302 be specified in any other non-default time unit, by suffixing the
8303 unit after the value, as explained at the top of this document.
8304
8305 This statement is useful on monitoring displays with a permanent page
8306 reporting the load balancer's activity. When set, the HTML report page will
8307 include a link "refresh"/"stop refresh" so that the user can select whether
8308 he wants automatic refresh of the page or not.
8309
8310 Though this statement alone is enough to enable statistics reporting, it is
8311 recommended to set all other settings in order to avoid relying on default
8312 unobvious parameters.
8313
8314 Example :
8315 # public access (limited to this backend only)
8316 backend public_www
8317 server srv1 192.168.0.1:80
8318 stats enable
8319 stats hide-version
8320 stats scope .
8321 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008322 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008323 stats auth admin1:AdMiN123
8324 stats auth admin2:AdMiN321
8325
8326 # internal monitoring access (unlimited)
8327 backend private_monitoring
8328 stats enable
8329 stats uri /admin?stats
8330 stats refresh 5s
8331
8332 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8333
8334
8335stats scope { <name> | "." }
8336 Enable statistics and limit access scope
8337 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008338 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008339 Arguments :
8340 <name> is the name of a listen, frontend or backend section to be
8341 reported. The special name "." (a single dot) designates the
8342 section in which the statement appears.
8343
8344 When this statement is specified, only the sections enumerated with this
8345 statement will appear in the report. All other ones will be hidden. This
8346 statement may appear as many times as needed if multiple sections need to be
8347 reported. Please note that the name checking is performed as simple string
8348 comparisons, and that it is never checked that a give section name really
8349 exists.
8350
8351 Though this statement alone is enough to enable statistics reporting, it is
8352 recommended to set all other settings in order to avoid relying on default
8353 unobvious parameters.
8354
8355 Example :
8356 # public access (limited to this backend only)
8357 backend public_www
8358 server srv1 192.168.0.1:80
8359 stats enable
8360 stats hide-version
8361 stats scope .
8362 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008363 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008364 stats auth admin1:AdMiN123
8365 stats auth admin2:AdMiN321
8366
8367 # internal monitoring access (unlimited)
8368 backend private_monitoring
8369 stats enable
8370 stats uri /admin?stats
8371 stats refresh 5s
8372
8373 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8374
Willy Tarreaud63335a2010-02-26 12:56:52 +01008375
Willy Tarreauc9705a12010-07-27 20:05:50 +02008376stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008377 Enable reporting of a description on the statistics page.
8378 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008379 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008380
Willy Tarreauc9705a12010-07-27 20:05:50 +02008381 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008382 description from global section is automatically used instead.
8383
8384 This statement is useful for users that offer shared services to their
8385 customers, where node or description should be different for each customer.
8386
8387 Though this statement alone is enough to enable statistics reporting, it is
8388 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008389 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008390
8391 Example :
8392 # internal monitoring access (unlimited)
8393 backend private_monitoring
8394 stats enable
8395 stats show-desc Master node for Europe, Asia, Africa
8396 stats uri /admin?stats
8397 stats refresh 5s
8398
8399 See also: "show-node", "stats enable", "stats uri" and "description" in
8400 global section.
8401
8402
8403stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008404 Enable reporting additional information on the statistics page
8405 May be used in sections : defaults | frontend | listen | backend
8406 yes | yes | yes | yes
8407 Arguments : none
8408
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008409 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008410 - cap: capabilities (proxy)
8411 - mode: one of tcp, http or health (proxy)
8412 - id: SNMP ID (proxy, socket, server)
8413 - IP (socket, server)
8414 - cookie (backend, server)
8415
8416 Though this statement alone is enough to enable statistics reporting, it is
8417 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008418 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008419
8420 See also: "stats enable", "stats uri".
8421
8422
8423stats show-node [ <name> ]
8424 Enable reporting of a host name on the statistics page.
8425 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008426 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008427 Arguments:
8428 <name> is an optional name to be reported. If unspecified, the
8429 node name from global section is automatically used instead.
8430
8431 This statement is useful for users that offer shared services to their
8432 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008433 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008434
8435 Though this statement alone is enough to enable statistics reporting, it is
8436 recommended to set all other settings in order to avoid relying on default
8437 unobvious parameters.
8438
8439 Example:
8440 # internal monitoring access (unlimited)
8441 backend private_monitoring
8442 stats enable
8443 stats show-node Europe-1
8444 stats uri /admin?stats
8445 stats refresh 5s
8446
8447 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8448 section.
8449
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008450
8451stats uri <prefix>
8452 Enable statistics and define the URI prefix to access them
8453 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008454 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008455 Arguments :
8456 <prefix> is the prefix of any URI which will be redirected to stats. This
8457 prefix may contain a question mark ('?') to indicate part of a
8458 query string.
8459
8460 The statistics URI is intercepted on the relayed traffic, so it appears as a
8461 page within the normal application. It is strongly advised to ensure that the
8462 selected URI will never appear in the application, otherwise it will never be
8463 possible to reach it in the application.
8464
8465 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008466 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008467 It is generally a good idea to include a question mark in the URI so that
8468 intermediate proxies refrain from caching the results. Also, since any string
8469 beginning with the prefix will be accepted as a stats request, the question
8470 mark helps ensuring that no valid URI will begin with the same words.
8471
8472 It is sometimes very convenient to use "/" as the URI prefix, and put that
8473 statement in a "listen" instance of its own. That makes it easy to dedicate
8474 an address or a port to statistics only.
8475
8476 Though this statement alone is enough to enable statistics reporting, it is
8477 recommended to set all other settings in order to avoid relying on default
8478 unobvious parameters.
8479
8480 Example :
8481 # public access (limited to this backend only)
8482 backend public_www
8483 server srv1 192.168.0.1:80
8484 stats enable
8485 stats hide-version
8486 stats scope .
8487 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008488 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008489 stats auth admin1:AdMiN123
8490 stats auth admin2:AdMiN321
8491
8492 # internal monitoring access (unlimited)
8493 backend private_monitoring
8494 stats enable
8495 stats uri /admin?stats
8496 stats refresh 5s
8497
8498 See also : "stats auth", "stats enable", "stats realm"
8499
8500
Willy Tarreaud63335a2010-02-26 12:56:52 +01008501stick match <pattern> [table <table>] [{if | unless} <cond>]
8502 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008503 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008504 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008505
8506 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008507 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008508 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008509 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008510 stickiness table. This rule is mandatory.
8511
8512 <table> is an optional stickiness table name. If unspecified, the same
8513 backend's table is used. A stickiness table is declared using
8514 the "stick-table" statement.
8515
8516 <cond> is an optional matching condition. It makes it possible to match
8517 on a certain criterion only when other conditions are met (or
8518 not met). For instance, it could be used to match on a source IP
8519 address except when a request passes through a known proxy, in
8520 which case we'd match on a header containing that IP address.
8521
8522 Some protocols or applications require complex stickiness rules and cannot
8523 always simply rely on cookies nor hashing. The "stick match" statement
8524 describes a rule to extract the stickiness criterion from an incoming request
8525 or connection. See section 7 for a complete list of possible patterns and
8526 transformation rules.
8527
8528 The table has to be declared using the "stick-table" statement. It must be of
8529 a type compatible with the pattern. By default it is the one which is present
8530 in the same backend. It is possible to share a table with other backends by
8531 referencing it using the "table" keyword. If another table is referenced,
8532 the server's ID inside the backends are used. By default, all server IDs
8533 start at 1 in each backend, so the server ordering is enough. But in case of
8534 doubt, it is highly recommended to force server IDs using their "id" setting.
8535
8536 It is possible to restrict the conditions where a "stick match" statement
8537 will apply, using "if" or "unless" followed by a condition. See section 7 for
8538 ACL based conditions.
8539
8540 There is no limit on the number of "stick match" statements. The first that
8541 applies and matches will cause the request to be directed to the same server
8542 as was used for the request which created the entry. That way, multiple
8543 matches can be used as fallbacks.
8544
8545 The stick rules are checked after the persistence cookies, so they will not
8546 affect stickiness if a cookie has already been used to select a server. That
8547 way, it becomes very easy to insert cookies and match on IP addresses in
8548 order to maintain stickiness between HTTP and HTTPS.
8549
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008550 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8551 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008552 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008553
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008554 Example :
8555 # forward SMTP users to the same server they just used for POP in the
8556 # last 30 minutes
8557 backend pop
8558 mode tcp
8559 balance roundrobin
8560 stick store-request src
8561 stick-table type ip size 200k expire 30m
8562 server s1 192.168.1.1:110
8563 server s2 192.168.1.1:110
8564
8565 backend smtp
8566 mode tcp
8567 balance roundrobin
8568 stick match src table pop
8569 server s1 192.168.1.1:25
8570 server s2 192.168.1.1:25
8571
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008572 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008573 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008574
8575
8576stick on <pattern> [table <table>] [{if | unless} <condition>]
8577 Define a request pattern to associate a user to a server
8578 May be used in sections : defaults | frontend | listen | backend
8579 no | no | yes | yes
8580
8581 Note : This form is exactly equivalent to "stick match" followed by
8582 "stick store-request", all with the same arguments. Please refer
8583 to both keywords for details. It is only provided as a convenience
8584 for writing more maintainable configurations.
8585
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008586 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8587 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008588 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008589
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008590 Examples :
8591 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008592 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008593
8594 # ...is strictly equivalent to this one :
8595 stick match src table pop if !localhost
8596 stick store-request src table pop if !localhost
8597
8598
8599 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8600 # well as HTTP without cookie. Share the same table between both accesses.
8601 backend http
8602 mode http
8603 balance roundrobin
8604 stick on src table https
8605 cookie SRV insert indirect nocache
8606 server s1 192.168.1.1:80 cookie s1
8607 server s2 192.168.1.1:80 cookie s2
8608
8609 backend https
8610 mode tcp
8611 balance roundrobin
8612 stick-table type ip size 200k expire 30m
8613 stick on src
8614 server s1 192.168.1.1:443
8615 server s2 192.168.1.1:443
8616
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008617 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008618
8619
8620stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8621 Define a request pattern used to create an entry in a stickiness table
8622 May be used in sections : defaults | frontend | listen | backend
8623 no | no | yes | yes
8624
8625 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008626 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008627 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008628 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008629 server is selected.
8630
8631 <table> is an optional stickiness table name. If unspecified, the same
8632 backend's table is used. A stickiness table is declared using
8633 the "stick-table" statement.
8634
8635 <cond> is an optional storage condition. It makes it possible to store
8636 certain criteria only when some conditions are met (or not met).
8637 For instance, it could be used to store the source IP address
8638 except when the request passes through a known proxy, in which
8639 case we'd store a converted form of a header containing that IP
8640 address.
8641
8642 Some protocols or applications require complex stickiness rules and cannot
8643 always simply rely on cookies nor hashing. The "stick store-request" statement
8644 describes a rule to decide what to extract from the request and when to do
8645 it, in order to store it into a stickiness table for further requests to
8646 match it using the "stick match" statement. Obviously the extracted part must
8647 make sense and have a chance to be matched in a further request. Storing a
8648 client's IP address for instance often makes sense. Storing an ID found in a
8649 URL parameter also makes sense. Storing a source port will almost never make
8650 any sense because it will be randomly matched. See section 7 for a complete
8651 list of possible patterns and transformation rules.
8652
8653 The table has to be declared using the "stick-table" statement. It must be of
8654 a type compatible with the pattern. By default it is the one which is present
8655 in the same backend. It is possible to share a table with other backends by
8656 referencing it using the "table" keyword. If another table is referenced,
8657 the server's ID inside the backends are used. By default, all server IDs
8658 start at 1 in each backend, so the server ordering is enough. But in case of
8659 doubt, it is highly recommended to force server IDs using their "id" setting.
8660
8661 It is possible to restrict the conditions where a "stick store-request"
8662 statement will apply, using "if" or "unless" followed by a condition. This
8663 condition will be evaluated while parsing the request, so any criteria can be
8664 used. See section 7 for ACL based conditions.
8665
8666 There is no limit on the number of "stick store-request" statements, but
8667 there is a limit of 8 simultaneous stores per request or response. This
8668 makes it possible to store up to 8 criteria, all extracted from either the
8669 request or the response, regardless of the number of rules. Only the 8 first
8670 ones which match will be kept. Using this, it is possible to feed multiple
8671 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008672 another protocol or access method. Using multiple store-request rules with
8673 the same table is possible and may be used to find the best criterion to rely
8674 on, by arranging the rules by decreasing preference order. Only the first
8675 extracted criterion for a given table will be stored. All subsequent store-
8676 request rules referencing the same table will be skipped and their ACLs will
8677 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008678
8679 The "store-request" rules are evaluated once the server connection has been
8680 established, so that the table will contain the real server that processed
8681 the request.
8682
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008683 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8684 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008685 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008686
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008687 Example :
8688 # forward SMTP users to the same server they just used for POP in the
8689 # last 30 minutes
8690 backend pop
8691 mode tcp
8692 balance roundrobin
8693 stick store-request src
8694 stick-table type ip size 200k expire 30m
8695 server s1 192.168.1.1:110
8696 server s2 192.168.1.1:110
8697
8698 backend smtp
8699 mode tcp
8700 balance roundrobin
8701 stick match src table pop
8702 server s1 192.168.1.1:25
8703 server s2 192.168.1.1:25
8704
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008705 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008706 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008707
8708
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008709stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008710 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8711 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008712 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008713 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008714 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008715
8716 Arguments :
8717 ip a table declared with "type ip" will only store IPv4 addresses.
8718 This form is very compact (about 50 bytes per entry) and allows
8719 very fast entry lookup and stores with almost no overhead. This
8720 is mainly used to store client source IP addresses.
8721
David du Colombier9a6d3c92011-03-17 10:40:24 +01008722 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8723 This form is very compact (about 60 bytes per entry) and allows
8724 very fast entry lookup and stores with almost no overhead. This
8725 is mainly used to store client source IP addresses.
8726
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008727 integer a table declared with "type integer" will store 32bit integers
8728 which can represent a client identifier found in a request for
8729 instance.
8730
8731 string a table declared with "type string" will store substrings of up
8732 to <len> characters. If the string provided by the pattern
8733 extractor is larger than <len>, it will be truncated before
8734 being stored. During matching, at most <len> characters will be
8735 compared between the string in the table and the extracted
8736 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008737 to 32 characters.
8738
8739 binary a table declared with "type binary" will store binary blocks
8740 of <len> bytes. If the block provided by the pattern
8741 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008742 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008743 is shorter than <len>, it will be padded by 0. When not
8744 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008745
8746 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008747 "string" type table (See type "string" above). Or the number
8748 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008749 changing this parameter as memory usage will proportionally
8750 increase.
8751
8752 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008753 value directly impacts memory usage. Count approximately
8754 50 bytes per entry, plus the size of a string if any. The size
8755 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008756
8757 [nopurge] indicates that we refuse to purge older entries when the table
8758 is full. When not specified and the table is full when haproxy
8759 wants to store an entry in it, it will flush a few of the oldest
8760 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008761 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008762 be desirable to refuse new entries instead of purging the older
8763 ones. That may be the case when the amount of data to store is
8764 far above the hardware limits and we prefer not to offer access
8765 to new clients than to reject the ones already connected. When
8766 using this parameter, be sure to properly set the "expire"
8767 parameter (see below).
8768
Emeric Brunf099e792010-09-27 12:05:28 +02008769 <peersect> is the name of the peers section to use for replication. Entries
8770 which associate keys to server IDs are kept synchronized with
8771 the remote peers declared in this section. All entries are also
8772 automatically learned from the local peer (old process) during a
8773 soft restart.
8774
Willy Tarreau1abc6732015-05-01 19:21:02 +02008775 NOTE : each peers section may be referenced only by tables
8776 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008777
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008778 <expire> defines the maximum duration of an entry in the table since it
8779 was last created, refreshed or matched. The expiration delay is
8780 defined using the standard time format, similarly as the various
8781 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008782 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008783 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008784 be removed once full. Be sure not to use the "nopurge" parameter
8785 if not expiration delay is specified.
8786
Willy Tarreau08d5f982010-06-06 13:34:54 +02008787 <data_type> is used to store additional information in the stick-table. This
8788 may be used by ACLs in order to control various criteria related
8789 to the activity of the client matching the stick-table. For each
8790 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008791 that the additional data can fit. Several data types may be
8792 stored with an entry. Multiple data types may be specified after
8793 the "store" keyword, as a comma-separated list. Alternatively,
8794 it is possible to repeat the "store" keyword followed by one or
8795 several data types. Except for the "server_id" type which is
8796 automatically detected and enabled, all data types must be
8797 explicitly declared to be stored. If an ACL references a data
8798 type which is not stored, the ACL will simply not match. Some
8799 data types require an argument which must be passed just after
8800 the type between parenthesis. See below for the supported data
8801 types and their arguments.
8802
8803 The data types that can be stored with an entry are the following :
8804 - server_id : this is an integer which holds the numeric ID of the server a
8805 request was assigned to. It is used by the "stick match", "stick store",
8806 and "stick on" rules. It is automatically enabled when referenced.
8807
8808 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
8809 integer which may be used for anything. Most of the time it will be used
8810 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01008811 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02008812
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008813 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
8814 over a period. It is a positive 32-bit integer integer which may be used
8815 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01008816 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008817 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01008818 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008819
Frédéric Lécaille6778b272018-01-29 15:22:53 +01008820 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
8821 integer which may be used for anything. Most of the time it will be used
8822 to put a special tag on some entries, for instance to note that a
8823 specific behavior was detected and must be known for future matches.
8824
8825 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
8826 over a period. It is a positive 32-bit integer integer which may be used
8827 for anything. Just like <gpc1>, it counts events, but instead of keeping
8828 a cumulative number, it maintains the rate at which the counter is
8829 incremented. Most of the time it will be used to measure the frequency of
8830 occurrence of certain events (e.g. requests to a specific URL).
8831
Willy Tarreauc9705a12010-07-27 20:05:50 +02008832 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
8833 the absolute number of connections received from clients which matched
8834 this entry. It does not mean the connections were accepted, just that
8835 they were received.
8836
8837 - conn_cur : Current Connections. It is a positive 32-bit integer which
8838 stores the concurrent connection counts for the entry. It is incremented
8839 once an incoming connection matches the entry, and decremented once the
8840 connection leaves. That way it is possible to know at any time the exact
8841 number of concurrent connections for an entry.
8842
8843 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8844 integer parameter <period> which indicates in milliseconds the length
8845 of the period over which the average is measured. It reports the average
8846 incoming connection rate over that period, in connections per period. The
8847 result is an integer which can be matched using ACLs.
8848
8849 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
8850 the absolute number of sessions received from clients which matched this
8851 entry. A session is a connection that was accepted by the layer 4 rules.
8852
8853 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8854 integer parameter <period> which indicates in milliseconds the length
8855 of the period over which the average is measured. It reports the average
8856 incoming session rate over that period, in sessions per period. The
8857 result is an integer which can be matched using ACLs.
8858
8859 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
8860 counts the absolute number of HTTP requests received from clients which
8861 matched this entry. It does not matter whether they are valid requests or
8862 not. Note that this is different from sessions when keep-alive is used on
8863 the client side.
8864
8865 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8866 integer parameter <period> which indicates in milliseconds the length
8867 of the period over which the average is measured. It reports the average
8868 HTTP request rate over that period, in requests per period. The result is
8869 an integer which can be matched using ACLs. It does not matter whether
8870 they are valid requests or not. Note that this is different from sessions
8871 when keep-alive is used on the client side.
8872
8873 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
8874 counts the absolute number of HTTP requests errors induced by clients
8875 which matched this entry. Errors are counted on invalid and truncated
8876 requests, as well as on denied or tarpitted requests, and on failed
8877 authentications. If the server responds with 4xx, then the request is
8878 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01008879 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02008880
8881 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8882 integer parameter <period> which indicates in milliseconds the length
8883 of the period over which the average is measured. It reports the average
8884 HTTP request error rate over that period, in requests per period (see
8885 http_err_cnt above for what is accounted as an error). The result is an
8886 integer which can be matched using ACLs.
8887
8888 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008889 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02008890 which matched this entry. Headers are included in the count. This may be
8891 used to limit abuse of upload features on photo or video servers.
8892
8893 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8894 integer parameter <period> which indicates in milliseconds the length
8895 of the period over which the average is measured. It reports the average
8896 incoming bytes rate over that period, in bytes per period. It may be used
8897 to detect users which upload too much and too fast. Warning: with large
8898 uploads, it is possible that the amount of uploaded data will be counted
8899 once upon termination, thus causing spikes in the average transfer speed
8900 instead of having a smooth one. This may partially be smoothed with
8901 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
8902 recommended for better fairness.
8903
8904 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008905 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02008906 matched this entry. Headers are included in the count. This may be used
8907 to limit abuse of bots sucking the whole site.
8908
8909 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
8910 an integer parameter <period> which indicates in milliseconds the length
8911 of the period over which the average is measured. It reports the average
8912 outgoing bytes rate over that period, in bytes per period. It may be used
8913 to detect users which download too much and too fast. Warning: with large
8914 transfers, it is possible that the amount of transferred data will be
8915 counted once upon termination, thus causing spikes in the average
8916 transfer speed instead of having a smooth one. This may partially be
8917 smoothed with "option contstats" though this is not perfect yet. Use of
8918 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02008919
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008920 There is only one stick-table per proxy. At the moment of writing this doc,
8921 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008922 to be required, simply create a dummy backend with a stick-table in it and
8923 reference it.
8924
8925 It is important to understand that stickiness based on learning information
8926 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01008927 lost upon restart unless peers are properly configured to transfer such
8928 information upon restart (recommended). In general it can be good as a
8929 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008930
Willy Tarreauc9705a12010-07-27 20:05:50 +02008931 Last, memory requirements may be important when storing many data types.
8932 Indeed, storing all indicators above at once in each entry requires 116 bytes
8933 per entry, or 116 MB for a 1-million entries table. This is definitely not
8934 something that can be ignored.
8935
8936 Example:
8937 # Keep track of counters of up to 1 million IP addresses over 5 minutes
8938 # and store a general purpose counter and the average connection rate
8939 # computed over a sliding window of 30 seconds.
8940 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
8941
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008942 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01008943 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008944
8945
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008946stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01008947 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008948 May be used in sections : defaults | frontend | listen | backend
8949 no | no | yes | yes
8950
8951 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008952 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008953 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01008954 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008955 server is selected.
8956
8957 <table> is an optional stickiness table name. If unspecified, the same
8958 backend's table is used. A stickiness table is declared using
8959 the "stick-table" statement.
8960
8961 <cond> is an optional storage condition. It makes it possible to store
8962 certain criteria only when some conditions are met (or not met).
8963 For instance, it could be used to store the SSL session ID only
8964 when the response is a SSL server hello.
8965
8966 Some protocols or applications require complex stickiness rules and cannot
8967 always simply rely on cookies nor hashing. The "stick store-response"
8968 statement describes a rule to decide what to extract from the response and
8969 when to do it, in order to store it into a stickiness table for further
8970 requests to match it using the "stick match" statement. Obviously the
8971 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008972 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008973 See section 7 for a complete list of possible patterns and transformation
8974 rules.
8975
8976 The table has to be declared using the "stick-table" statement. It must be of
8977 a type compatible with the pattern. By default it is the one which is present
8978 in the same backend. It is possible to share a table with other backends by
8979 referencing it using the "table" keyword. If another table is referenced,
8980 the server's ID inside the backends are used. By default, all server IDs
8981 start at 1 in each backend, so the server ordering is enough. But in case of
8982 doubt, it is highly recommended to force server IDs using their "id" setting.
8983
8984 It is possible to restrict the conditions where a "stick store-response"
8985 statement will apply, using "if" or "unless" followed by a condition. This
8986 condition will be evaluated while parsing the response, so any criteria can
8987 be used. See section 7 for ACL based conditions.
8988
8989 There is no limit on the number of "stick store-response" statements, but
8990 there is a limit of 8 simultaneous stores per request or response. This
8991 makes it possible to store up to 8 criteria, all extracted from either the
8992 request or the response, regardless of the number of rules. Only the 8 first
8993 ones which match will be kept. Using this, it is possible to feed multiple
8994 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008995 another protocol or access method. Using multiple store-response rules with
8996 the same table is possible and may be used to find the best criterion to rely
8997 on, by arranging the rules by decreasing preference order. Only the first
8998 extracted criterion for a given table will be stored. All subsequent store-
8999 response rules referencing the same table will be skipped and their ACLs will
9000 not be evaluated. However, even if a store-request rule references a table, a
9001 store-response rule may also use the same table. This means that each table
9002 may learn exactly one element from the request and one element from the
9003 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009004
9005 The table will contain the real server that processed the request.
9006
9007 Example :
9008 # Learn SSL session ID from both request and response and create affinity.
9009 backend https
9010 mode tcp
9011 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009012 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009013 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009014
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009015 acl clienthello req_ssl_hello_type 1
9016 acl serverhello rep_ssl_hello_type 2
9017
9018 # use tcp content accepts to detects ssl client and server hello.
9019 tcp-request inspect-delay 5s
9020 tcp-request content accept if clienthello
9021
9022 # no timeout on response inspect delay by default.
9023 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009024
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009025 # SSL session ID (SSLID) may be present on a client or server hello.
9026 # Its length is coded on 1 byte at offset 43 and its value starts
9027 # at offset 44.
9028
9029 # Match and learn on request if client hello.
9030 stick on payload_lv(43,1) if clienthello
9031
9032 # Learn on response if server hello.
9033 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009034
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009035 server s1 192.168.1.1:443
9036 server s2 192.168.1.1:443
9037
9038 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9039 extraction.
9040
9041
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009042tcp-check connect [params*]
9043 Opens a new connection
9044 May be used in sections: defaults | frontend | listen | backend
9045 no | no | yes | yes
9046
9047 When an application lies on more than a single TCP port or when HAProxy
9048 load-balance many services in a single backend, it makes sense to probe all
9049 the services individually before considering a server as operational.
9050
9051 When there are no TCP port configured on the server line neither server port
9052 directive, then the 'tcp-check connect port <port>' must be the first step
9053 of the sequence.
9054
9055 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9056 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9057 do.
9058
9059 Parameters :
9060 They are optional and can be used to describe how HAProxy should open and
9061 use the TCP connection.
9062
9063 port if not set, check port or server port is used.
9064 It tells HAProxy where to open the connection to.
9065 <port> must be a valid TCP port source integer, from 1 to 65535.
9066
9067 send-proxy send a PROXY protocol string
9068
9069 ssl opens a ciphered connection
9070
9071 Examples:
9072 # check HTTP and HTTPs services on a server.
9073 # first open port 80 thanks to server line port directive, then
9074 # tcp-check opens port 443, ciphered and run a request on it:
9075 option tcp-check
9076 tcp-check connect
9077 tcp-check send GET\ /\ HTTP/1.0\r\n
9078 tcp-check send Host:\ haproxy.1wt.eu\r\n
9079 tcp-check send \r\n
9080 tcp-check expect rstring (2..|3..)
9081 tcp-check connect port 443 ssl
9082 tcp-check send GET\ /\ HTTP/1.0\r\n
9083 tcp-check send Host:\ haproxy.1wt.eu\r\n
9084 tcp-check send \r\n
9085 tcp-check expect rstring (2..|3..)
9086 server www 10.0.0.1 check port 80
9087
9088 # check both POP and IMAP from a single server:
9089 option tcp-check
9090 tcp-check connect port 110
9091 tcp-check expect string +OK\ POP3\ ready
9092 tcp-check connect port 143
9093 tcp-check expect string *\ OK\ IMAP4\ ready
9094 server mail 10.0.0.1 check
9095
9096 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9097
9098
9099tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009100 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009101 May be used in sections: defaults | frontend | listen | backend
9102 no | no | yes | yes
9103
9104 Arguments :
9105 <match> is a keyword indicating how to look for a specific pattern in the
9106 response. The keyword may be one of "string", "rstring" or
9107 binary.
9108 The keyword may be preceded by an exclamation mark ("!") to negate
9109 the match. Spaces are allowed between the exclamation mark and the
9110 keyword. See below for more details on the supported keywords.
9111
9112 <pattern> is the pattern to look for. It may be a string or a regular
9113 expression. If the pattern contains spaces, they must be escaped
9114 with the usual backslash ('\').
9115 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009116 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009117 two digits will represent a byte. The hexadecimal digits may be
9118 used upper or lower case.
9119
9120
9121 The available matches are intentionally similar to their http-check cousins :
9122
9123 string <string> : test the exact string matches in the response buffer.
9124 A health check response will be considered valid if the
9125 response's buffer contains this exact string. If the
9126 "string" keyword is prefixed with "!", then the response
9127 will be considered invalid if the body contains this
9128 string. This can be used to look for a mandatory pattern
9129 in a protocol response, or to detect a failure when a
9130 specific error appears in a protocol banner.
9131
9132 rstring <regex> : test a regular expression on the response buffer.
9133 A health check response will be considered valid if the
9134 response's buffer matches this expression. If the
9135 "rstring" keyword is prefixed with "!", then the response
9136 will be considered invalid if the body matches the
9137 expression.
9138
9139 binary <hexstring> : test the exact string in its hexadecimal form matches
9140 in the response buffer. A health check response will
9141 be considered valid if the response's buffer contains
9142 this exact hexadecimal string.
9143 Purpose is to match data on binary protocols.
9144
9145 It is important to note that the responses will be limited to a certain size
9146 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9147 Thus, too large responses may not contain the mandatory pattern when using
9148 "string", "rstring" or binary. If a large response is absolutely required, it
9149 is possible to change the default max size by setting the global variable.
9150 However, it is worth keeping in mind that parsing very large responses can
9151 waste some CPU cycles, especially when regular expressions are used, and that
9152 it is always better to focus the checks on smaller resources. Also, in its
9153 current state, the check will not find any string nor regex past a null
9154 character in the response. Similarly it is not possible to request matching
9155 the null character.
9156
9157 Examples :
9158 # perform a POP check
9159 option tcp-check
9160 tcp-check expect string +OK\ POP3\ ready
9161
9162 # perform an IMAP check
9163 option tcp-check
9164 tcp-check expect string *\ OK\ IMAP4\ ready
9165
9166 # look for the redis master server
9167 option tcp-check
9168 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009169 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009170 tcp-check send info\ replication\r\n
9171 tcp-check expect string role:master
9172 tcp-check send QUIT\r\n
9173 tcp-check expect string +OK
9174
9175
9176 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9177 "tcp-check send-binary", "http-check expect", tune.chksize
9178
9179
9180tcp-check send <data>
9181 Specify a string to be sent as a question during a generic health check
9182 May be used in sections: defaults | frontend | listen | backend
9183 no | no | yes | yes
9184
9185 <data> : the data to be sent as a question during a generic health check
9186 session. For now, <data> must be a string.
9187
9188 Examples :
9189 # look for the redis master server
9190 option tcp-check
9191 tcp-check send info\ replication\r\n
9192 tcp-check expect string role:master
9193
9194 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9195 "tcp-check send-binary", tune.chksize
9196
9197
Davor Ocelice9ed2812017-12-25 17:49:28 +01009198tcp-check send-binary <hexstring>
9199 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009200 tcp health check
9201 May be used in sections: defaults | frontend | listen | backend
9202 no | no | yes | yes
9203
9204 <data> : the data to be sent as a question during a generic health check
9205 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009206 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009207 response buffer. A health check response will be considered
9208 valid if the response's buffer contains this exact
9209 hexadecimal string.
9210 Purpose is to send binary data to ask on binary protocols.
9211
9212 Examples :
9213 # redis check in binary
9214 option tcp-check
9215 tcp-check send-binary 50494e470d0a # PING\r\n
9216 tcp-check expect binary 2b504F4e47 # +PONG
9217
9218
9219 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9220 "tcp-check send", tune.chksize
9221
9222
Willy Tarreaue9656522010-08-17 15:40:09 +02009223tcp-request connection <action> [{if | unless} <condition>]
9224 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009225 May be used in sections : defaults | frontend | listen | backend
9226 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009227 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009228 <action> defines the action to perform if the condition applies. See
9229 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009230
Willy Tarreaue9656522010-08-17 15:40:09 +02009231 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009232
9233 Immediately after acceptance of a new incoming connection, it is possible to
9234 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009235 or dropped or have its counters tracked. Those conditions cannot make use of
9236 any data contents because the connection has not been read from yet, and the
9237 buffers are not yet allocated. This is used to selectively and very quickly
9238 accept or drop connections from various sources with a very low overhead. If
9239 some contents need to be inspected in order to take the decision, the
9240 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009241
Willy Tarreaue9656522010-08-17 15:40:09 +02009242 The "tcp-request connection" rules are evaluated in their exact declaration
9243 order. If no rule matches or if there is no rule, the default action is to
9244 accept the incoming connection. There is no specific limit to the number of
9245 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009246
Willy Tarreaua9083d02015-05-08 15:27:59 +02009247 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009248 - accept :
9249 accepts the connection if the condition is true (when used with "if")
9250 or false (when used with "unless"). The first such rule executed ends
9251 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009252
Willy Tarreaue9656522010-08-17 15:40:09 +02009253 - reject :
9254 rejects the connection if the condition is true (when used with "if")
9255 or false (when used with "unless"). The first such rule executed ends
9256 the rules evaluation. Rejected connections do not even become a
9257 session, which is why they are accounted separately for in the stats,
9258 as "denied connections". They are not considered for the session
9259 rate-limit and are not logged either. The reason is that these rules
9260 should only be used to filter extremely high connection rates such as
9261 the ones encountered during a massive DDoS attack. Under these extreme
9262 conditions, the simple action of logging each event would make the
9263 system collapse and would considerably lower the filtering capacity. If
9264 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009265 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009266
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009267 - expect-proxy layer4 :
9268 configures the client-facing connection to receive a PROXY protocol
9269 header before any byte is read from the socket. This is equivalent to
9270 having the "accept-proxy" keyword on the "bind" line, except that using
9271 the TCP rule allows the PROXY protocol to be accepted only for certain
9272 IP address ranges using an ACL. This is convenient when multiple layers
9273 of load balancers are passed through by traffic coming from public
9274 hosts.
9275
Bertrand Jacquin90759682016-06-06 15:35:39 +01009276 - expect-netscaler-cip layer4 :
9277 configures the client-facing connection to receive a NetScaler Client
9278 IP insertion protocol header before any byte is read from the socket.
9279 This is equivalent to having the "accept-netscaler-cip" keyword on the
9280 "bind" line, except that using the TCP rule allows the PROXY protocol
9281 to be accepted only for certain IP address ranges using an ACL. This
9282 is convenient when multiple layers of load balancers are passed
9283 through by traffic coming from public hosts.
9284
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009285 - capture <sample> len <length> :
9286 This only applies to "tcp-request content" rules. It captures sample
9287 expression <sample> from the request buffer, and converts it to a
9288 string of at most <len> characters. The resulting string is stored into
9289 the next request "capture" slot, so it will possibly appear next to
9290 some captured HTTP headers. It will then automatically appear in the
9291 logs, and it will be possible to extract it using sample fetch rules to
9292 feed it into headers or anything. The length should be limited given
9293 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009294 session life. Please check section 7.3 (Fetching samples) and "capture
9295 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009296
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009297 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009298 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009299 rules do not stop evaluation and do not change default action. The
9300 number of counters that may be simultaneously tracked by the same
9301 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -05009302 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009303 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9304 tracking of the counters of the specified table as the first set. The
9305 first "track-sc1" rule executed enables tracking of the counters of the
9306 specified table as the second set. The first "track-sc2" rule executed
9307 enables tracking of the counters of the specified table as the third
9308 set. It is a recommended practice to use the first set of counters for
9309 the per-frontend counters and the second set for the per-backend ones.
9310 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009311
Willy Tarreaue9656522010-08-17 15:40:09 +02009312 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009313 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009314 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009315 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009316 and used to select which table entry to update the counters.
9317 Note that "tcp-request connection" cannot use content-based
9318 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009319
Willy Tarreaue9656522010-08-17 15:40:09 +02009320 <table> is an optional table to be used instead of the default one,
9321 which is the stick-table declared in the current proxy. All
9322 the counters for the matches and updates for the key will
9323 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009324
Willy Tarreaue9656522010-08-17 15:40:09 +02009325 Once a "track-sc*" rule is executed, the key is looked up in the table
9326 and if it is not found, an entry is allocated for it. Then a pointer to
9327 that entry is kept during all the session's life, and this entry's
9328 counters are updated as often as possible, every time the session's
9329 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009330 Counters are only updated for events that happen after the tracking has
9331 been started. For example, connection counters will not be updated when
9332 tracking layer 7 information, since the connection event happens before
9333 layer7 information is extracted.
9334
Willy Tarreaue9656522010-08-17 15:40:09 +02009335 If the entry tracks concurrent connection counters, one connection is
9336 counted for as long as the entry is tracked, and the entry will not
9337 expire during that time. Tracking counters also provides a performance
9338 advantage over just checking the keys, because only one table lookup is
9339 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009340
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009341 - sc-inc-gpc0(<sc-id>):
9342 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9343 counter designated by <sc-id>. If an error occurs, this action silently
9344 fails and the actions evaluation continues.
9345
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009346 - sc-inc-gpc1(<sc-id>):
9347 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9348 counter designated by <sc-id>. If an error occurs, this action silently
9349 fails and the actions evaluation continues.
9350
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009351 - sc-set-gpt0(<sc-id>) <int>:
9352 This action sets the GPT0 tag according to the sticky counter designated
9353 by <sc-id> and the value of <int>. The expected result is a boolean. If
9354 an error occurs, this action silently fails and the actions evaluation
9355 continues.
9356
William Lallemand2e785f22016-05-25 01:48:42 +02009357 - set-src <expr> :
9358 Is used to set the source IP address to the value of specified
9359 expression. Useful if you want to mask source IP for privacy.
9360 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009361 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +02009362
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009363 Arguments:
9364 <expr> Is a standard HAProxy expression formed by a sample-fetch
9365 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +02009366
9367 Example:
William Lallemand2e785f22016-05-25 01:48:42 +02009368 tcp-request connection set-src src,ipmask(24)
9369
Willy Tarreau0c630532016-10-21 17:52:58 +02009370 When possible, set-src preserves the original source port as long as the
9371 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009372
William Lallemand44be6402016-05-25 01:51:35 +02009373 - set-src-port <expr> :
9374 Is used to set the source port address to the value of specified
9375 expression.
9376
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009377 Arguments:
9378 <expr> Is a standard HAProxy expression formed by a sample-fetch
9379 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +02009380
9381 Example:
William Lallemand44be6402016-05-25 01:51:35 +02009382 tcp-request connection set-src-port int(4000)
9383
Willy Tarreau0c630532016-10-21 17:52:58 +02009384 When possible, set-src-port preserves the original source address as long
9385 as the address family supports a port, otherwise it forces the source
9386 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009387
William Lallemand13e9b0c2016-05-25 02:34:07 +02009388 - set-dst <expr> :
9389 Is used to set the destination IP address to the value of specified
9390 expression. Useful if you want to mask IP for privacy in log.
9391 If you want to provide an IP from a HTTP header use "http-request
9392 set-dst". If you want to connect to the new address/port, use
9393 '0.0.0.0:0' as a server address in the backend.
9394
9395 <expr> Is a standard HAProxy expression formed by a sample-fetch
9396 followed by some converters.
9397
9398 Example:
9399
9400 tcp-request connection set-dst dst,ipmask(24)
9401 tcp-request connection set-dst ipv4(10.0.0.1)
9402
Willy Tarreau0c630532016-10-21 17:52:58 +02009403 When possible, set-dst preserves the original destination port as long as
9404 the address family allows it, otherwise the destination port is set to 0.
9405
William Lallemand13e9b0c2016-05-25 02:34:07 +02009406 - set-dst-port <expr> :
9407 Is used to set the destination port address to the value of specified
9408 expression. If you want to connect to the new address/port, use
9409 '0.0.0.0:0' as a server address in the backend.
9410
9411
9412 <expr> Is a standard HAProxy expression formed by a sample-fetch
9413 followed by some converters.
9414
9415 Example:
9416
9417 tcp-request connection set-dst-port int(4000)
9418
Willy Tarreau0c630532016-10-21 17:52:58 +02009419 When possible, set-dst-port preserves the original destination address as
9420 long as the address family supports a port, otherwise it forces the
9421 destination address to IPv4 "0.0.0.0" before rewriting the port.
9422
Willy Tarreau2d392c22015-08-24 01:43:45 +02009423 - "silent-drop" :
9424 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009425 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009426 to prevent the client from being notified. The effect it then that the
9427 client still sees an established connection while there's none on
9428 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9429 except that it doesn't use any local resource at all on the machine
9430 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009431 slow down stronger attackers. It is important to understand the impact
9432 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009433 client and HAProxy (firewalls, proxies, load balancers) will also keep
9434 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009435 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009436 TCP_REPAIR socket option is used to block the emission of a TCP
9437 reset. On other systems, the socket's TTL is reduced to 1 so that the
9438 TCP reset doesn't pass the first router, though it's still delivered to
9439 local networks. Do not use it unless you fully understand how it works.
9440
Willy Tarreaue9656522010-08-17 15:40:09 +02009441 Note that the "if/unless" condition is optional. If no condition is set on
9442 the action, it is simply performed unconditionally. That can be useful for
9443 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009444
Willy Tarreaue9656522010-08-17 15:40:09 +02009445 Example: accept all connections from white-listed hosts, reject too fast
9446 connection without counting them, and track accepted connections.
9447 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009448
Willy Tarreaue9656522010-08-17 15:40:09 +02009449 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009450 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009451 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009452
Willy Tarreaue9656522010-08-17 15:40:09 +02009453 Example: accept all connections from white-listed hosts, count all other
9454 connections and reject too fast ones. This results in abusive ones
9455 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009456
Willy Tarreaue9656522010-08-17 15:40:09 +02009457 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009458 tcp-request connection track-sc0 src
9459 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009460
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009461 Example: enable the PROXY protocol for traffic coming from all known proxies.
9462
9463 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9464
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009465 See section 7 about ACL usage.
9466
Willy Tarreau4f614292016-10-21 17:49:36 +02009467 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009468
9469
Willy Tarreaue9656522010-08-17 15:40:09 +02009470tcp-request content <action> [{if | unless} <condition>]
9471 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009472 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009473 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009474 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009475 <action> defines the action to perform if the condition applies. See
9476 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009477
Willy Tarreaue9656522010-08-17 15:40:09 +02009478 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009479
Davor Ocelice9ed2812017-12-25 17:49:28 +01009480 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009481 called "TCP content inspection". During this stage, ACL-based rules are
9482 evaluated every time the request contents are updated, until either an
9483 "accept" or a "reject" rule matches, or the TCP request inspection delay
9484 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009485
Willy Tarreaue9656522010-08-17 15:40:09 +02009486 The first difference between these rules and "tcp-request connection" rules
9487 is that "tcp-request content" rules can make use of contents to take a
9488 decision. Most often, these decisions will consider a protocol recognition or
9489 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009490 both frontends and backends. In case of HTTP keep-alive with the client, all
9491 tcp-request content rules are evaluated again, so haproxy keeps a record of
9492 what sticky counters were assigned by a "tcp-request connection" versus a
9493 "tcp-request content" rule, and flushes all the content-related ones after
9494 processing an HTTP request, so that they may be evaluated again by the rules
9495 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009496 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009497 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009498
Willy Tarreaue9656522010-08-17 15:40:09 +02009499 Content-based rules are evaluated in their exact declaration order. If no
9500 rule matches or if there is no rule, the default action is to accept the
9501 contents. There is no specific limit to the number of rules which may be
9502 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009503
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009504 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009505 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +01009506 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009507 - reject : the request is rejected and the connection is closed
9508 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009509 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009510 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009511 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009512 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009513 - sc-set-gpt0(<sc-id>) <int>
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +02009514 - set-dst <expr>
9515 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009516 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009517 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009518 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009519 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009520
Willy Tarreaue9656522010-08-17 15:40:09 +02009521 They have the same meaning as their counter-parts in "tcp-request connection"
9522 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +01009523 For "do-resolve" action, please check the "http-request do-resolve"
9524 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +02009525
Willy Tarreauf3338342014-01-28 21:40:28 +01009526 While there is nothing mandatory about it, it is recommended to use the
9527 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9528 content" rules in the frontend, and track-sc2 for "tcp-request content"
9529 rules in the backend, because that makes the configuration more readable
9530 and easier to troubleshoot, but this is just a guideline and all counters
9531 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009532
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009533 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009534 the action, it is simply performed unconditionally. That can be useful for
9535 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009536
Willy Tarreaue9656522010-08-17 15:40:09 +02009537 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009538 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9539 contents of a buffer before extracting the required data. If the buffered
9540 contents do not parse as a valid HTTP message, then the ACL does not match.
9541 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009542 processing, so there is no risk of parsing something differently. In an HTTP
9543 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9544 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009545
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009546 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009547 are present when the rule is processed. The rule processing engine is able to
9548 wait until the inspect delay expires when the data to be tracked is not yet
9549 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009550
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +02009551 The "set-dst" and "set-dst-port" are used to set respectively the destination
9552 IP and port. More information on how to use it at "http-request set-dst".
9553
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009554 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009555 declared inline. For "tcp-request session" rules, only session-level
9556 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009557
Daniel Schneller0b547052016-03-21 20:46:57 +01009558 <var-name> The name of the variable starts with an indication about
9559 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009560 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009561 "sess" : the variable is shared with the whole session
9562 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009563 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009564 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009565 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009566 "res" : the variable is shared only during response
9567 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009568 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009569 The name may only contain characters 'a-z', 'A-Z', '0-9',
9570 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009571
9572 <expr> Is a standard HAProxy expression formed by a sample-fetch
9573 followed by some converters.
9574
Christopher Faulet85d79c92016-11-09 16:54:56 +01009575 The "unset-var" is used to unset a variable. See above for details about
9576 <var-name>.
9577
Patrick Hemmer268a7072018-05-11 12:52:31 -04009578 The "set-priority-class" is used to set the queue priority class of the
9579 current request. The value must be a sample expression which converts to an
9580 integer in the range -2047..2047. Results outside this range will be
9581 truncated. The priority class determines the order in which queued requests
9582 are processed. Lower values have higher priority.
9583
9584 The "set-priority-offset" is used to set the queue priority timestamp offset
9585 of the current request. The value must be a sample expression which converts
9586 to an integer in the range -524287..524287. Results outside this range will be
9587 truncated. When a request is queued, it is ordered first by the priority
9588 class, then by the current timestamp adjusted by the given offset in
9589 milliseconds. Lower values have higher priority.
9590 Note that the resulting timestamp is is only tracked with enough precision for
9591 524,287ms (8m44s287ms). If the request is queued long enough to where the
9592 adjusted timestamp exceeds this value, it will be misidentified as highest
9593 priority. Thus it is important to set "timeout queue" to a value, where when
9594 combined with the offset, does not exceed this limit.
9595
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009596 The "send-spoe-group" is used to trigger sending of a group of SPOE
9597 messages. To do so, the SPOE engine used to send messages must be defined, as
9598 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9599 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9600 the SPOE agent name must be used.
9601
9602 <engine-name> The SPOE engine name.
9603
9604 <group-name> The SPOE group name as specified in the engine configuration.
9605
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009606 Example:
9607
9608 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009609 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009610
Willy Tarreau62644772008-07-16 18:36:06 +02009611 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009612 # Accept HTTP requests containing a Host header saying "example.com"
9613 # and reject everything else.
9614 acl is_host_com hdr(Host) -i example.com
9615 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009616 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009617 tcp-request content reject
9618
9619 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009620 # reject SMTP connection if client speaks first
9621 tcp-request inspect-delay 30s
9622 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009623 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009624
9625 # Forward HTTPS connection only if client speaks
9626 tcp-request inspect-delay 30s
9627 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009628 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009629 tcp-request content reject
9630
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009631 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009632 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009633 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009634 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009635 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9636 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009637
9638 Example:
9639 # track request counts per "base" (concatenation of Host+URL)
9640 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009641 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009642
Willy Tarreaue9656522010-08-17 15:40:09 +02009643 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009644 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009645
9646 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009647 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009648 # protecting all our sites
9649 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009650 tcp-request connection track-sc0 src
9651 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009652 ...
9653 use_backend http_dynamic if { path_end .php }
9654
9655 backend http_dynamic
9656 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009657 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009658 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009659 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009660 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009661 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009662 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009663
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009664 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009665
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009666 See also : "tcp-request connection", "tcp-request session",
9667 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009668
9669
9670tcp-request inspect-delay <timeout>
9671 Set the maximum allowed time to wait for data during content inspection
9672 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009673 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009674 Arguments :
9675 <timeout> is the timeout value specified in milliseconds by default, but
9676 can be in any other unit if the number is suffixed by the unit,
9677 as explained at the top of this document.
9678
9679 People using haproxy primarily as a TCP relay are often worried about the
9680 risk of passing any type of protocol to a server without any analysis. In
9681 order to be able to analyze the request contents, we must first withhold
9682 the data then analyze them. This statement simply enables withholding of
9683 data for at most the specified amount of time.
9684
Willy Tarreaufb356202010-08-03 14:02:05 +02009685 TCP content inspection applies very early when a connection reaches a
9686 frontend, then very early when the connection is forwarded to a backend. This
9687 means that a connection may experience a first delay in the frontend and a
9688 second delay in the backend if both have tcp-request rules.
9689
Willy Tarreau62644772008-07-16 18:36:06 +02009690 Note that when performing content inspection, haproxy will evaluate the whole
9691 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009692 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009693 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009694 contents are definitive. If no delay is set, haproxy will not wait at all
9695 and will immediately apply a verdict based on the available information.
9696 Obviously this is unlikely to be very useful and might even be racy, so such
9697 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009698
9699 As soon as a rule matches, the request is released and continues as usual. If
9700 the timeout is reached and no rule matches, the default policy will be to let
9701 it pass through unaffected.
9702
9703 For most protocols, it is enough to set it to a few seconds, as most clients
9704 send the full request immediately upon connection. Add 3 or more seconds to
9705 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009706 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009707 before the server (e.g. SMTP), or to wait for a client to talk before passing
9708 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009709 least the inspection delay, otherwise it will expire first. If the client
9710 closes the connection or if the buffer is full, the delay immediately expires
9711 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009712
Willy Tarreau55165fe2009-05-10 12:02:55 +02009713 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009714 "timeout client".
9715
9716
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009717tcp-response content <action> [{if | unless} <condition>]
9718 Perform an action on a session response depending on a layer 4-7 condition
9719 May be used in sections : defaults | frontend | listen | backend
9720 no | no | yes | yes
9721 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009722 <action> defines the action to perform if the condition applies. See
9723 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009724
9725 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9726
Davor Ocelice9ed2812017-12-25 17:49:28 +01009727 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009728 called "TCP content inspection". During this stage, ACL-based rules are
9729 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009730 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9731 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009732
9733 Most often, these decisions will consider a protocol recognition or validity.
9734
9735 Content-based rules are evaluated in their exact declaration order. If no
9736 rule matches or if there is no rule, the default action is to accept the
9737 contents. There is no specific limit to the number of rules which may be
9738 inserted.
9739
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009740 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009741 - accept :
9742 accepts the response if the condition is true (when used with "if")
9743 or false (when used with "unless"). The first such rule executed ends
9744 the rules evaluation.
9745
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009746 - close :
9747 immediately closes the connection with the server if the condition is
9748 true (when used with "if"), or false (when used with "unless"). The
9749 first such rule executed ends the rules evaluation. The main purpose of
9750 this action is to force a connection to be finished between a client
9751 and a server after an exchange when the application protocol expects
9752 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009753 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009754 protocols.
9755
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009756 - reject :
9757 rejects the response if the condition is true (when used with "if")
9758 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009759 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009760
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009761 - set-var(<var-name>) <expr>
9762 Sets a variable.
9763
Christopher Faulet85d79c92016-11-09 16:54:56 +01009764 - unset-var(<var-name>)
9765 Unsets a variable.
9766
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009767 - sc-inc-gpc0(<sc-id>):
9768 This action increments the GPC0 counter according to the sticky
9769 counter designated by <sc-id>. If an error occurs, this action fails
9770 silently and the actions evaluation continues.
9771
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009772 - sc-inc-gpc1(<sc-id>):
9773 This action increments the GPC1 counter according to the sticky
9774 counter designated by <sc-id>. If an error occurs, this action fails
9775 silently and the actions evaluation continues.
9776
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009777 - sc-set-gpt0(<sc-id>) <int> :
9778 This action sets the GPT0 tag according to the sticky counter designated
9779 by <sc-id> and the value of <int>. The expected result is a boolean. If
9780 an error occurs, this action silently fails and the actions evaluation
9781 continues.
9782
Willy Tarreau2d392c22015-08-24 01:43:45 +02009783 - "silent-drop" :
9784 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009785 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009786 to prevent the client from being notified. The effect it then that the
9787 client still sees an established connection while there's none on
9788 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9789 except that it doesn't use any local resource at all on the machine
9790 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009791 slow down stronger attackers. It is important to understand the impact
9792 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009793 client and HAProxy (firewalls, proxies, load balancers) will also keep
9794 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009795 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009796 TCP_REPAIR socket option is used to block the emission of a TCP
9797 reset. On other systems, the socket's TTL is reduced to 1 so that the
9798 TCP reset doesn't pass the first router, though it's still delivered to
9799 local networks. Do not use it unless you fully understand how it works.
9800
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009801 - send-spoe-group <engine-name> <group-name>
9802 Send a group of SPOE messages.
9803
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009804 Note that the "if/unless" condition is optional. If no condition is set on
9805 the action, it is simply performed unconditionally. That can be useful for
9806 for changing the default action to a reject.
9807
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009808 It is perfectly possible to match layer 7 contents with "tcp-response
9809 content" rules, but then it is important to ensure that a full response has
9810 been buffered, otherwise no contents will match. In order to achieve this,
9811 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009812 period.
9813
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009814 The "set-var" is used to set the content of a variable. The variable is
9815 declared inline.
9816
Daniel Schneller0b547052016-03-21 20:46:57 +01009817 <var-name> The name of the variable starts with an indication about
9818 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009819 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009820 "sess" : the variable is shared with the whole session
9821 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009822 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009823 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009824 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009825 "res" : the variable is shared only during response
9826 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009827 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009828 The name may only contain characters 'a-z', 'A-Z', '0-9',
9829 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009830
9831 <expr> Is a standard HAProxy expression formed by a sample-fetch
9832 followed by some converters.
9833
9834 Example:
9835
9836 tcp-request content set-var(sess.my_var) src
9837
Christopher Faulet85d79c92016-11-09 16:54:56 +01009838 The "unset-var" is used to unset a variable. See above for details about
9839 <var-name>.
9840
9841 Example:
9842
9843 tcp-request content unset-var(sess.my_var)
9844
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009845 The "send-spoe-group" is used to trigger sending of a group of SPOE
9846 messages. To do so, the SPOE engine used to send messages must be defined, as
9847 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9848 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9849 the SPOE agent name must be used.
9850
9851 <engine-name> The SPOE engine name.
9852
9853 <group-name> The SPOE group name as specified in the engine configuration.
9854
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009855 See section 7 about ACL usage.
9856
9857 See also : "tcp-request content", "tcp-response inspect-delay"
9858
9859
Willy Tarreau4f614292016-10-21 17:49:36 +02009860tcp-request session <action> [{if | unless} <condition>]
9861 Perform an action on a validated session depending on a layer 5 condition
9862 May be used in sections : defaults | frontend | listen | backend
9863 no | yes | yes | no
9864 Arguments :
9865 <action> defines the action to perform if the condition applies. See
9866 below.
9867
9868 <condition> is a standard layer5-only ACL-based condition (see section 7).
9869
Davor Ocelice9ed2812017-12-25 17:49:28 +01009870 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +02009871 it is possible to evaluate some conditions to decide whether this session
9872 must be accepted or dropped or have its counters tracked. Those conditions
9873 cannot make use of any data contents because no buffers are allocated yet and
9874 the processing cannot wait at this stage. The main use case it to copy some
9875 early information into variables (since variables are accessible in the
9876 session), or to keep track of some information collected after the handshake,
9877 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +01009878 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +02009879 extracted information can thus be copied to a variable or tracked using
9880 "track-sc" rules. Of course it is also possible to decide to accept/reject as
9881 with other rulesets. Most operations performed here could also be performed
9882 in "tcp-request content" rules, except that in HTTP these rules are evaluated
9883 for each new request, and that might not always be acceptable. For example a
9884 rule might increment a counter on each evaluation. It would also be possible
9885 that a country is resolved by geolocation from the source IP address,
9886 assigned to a session-wide variable, then the source address rewritten from
9887 an HTTP header for all requests. If some contents need to be inspected in
9888 order to take the decision, the "tcp-request content" statements must be used
9889 instead.
9890
9891 The "tcp-request session" rules are evaluated in their exact declaration
9892 order. If no rule matches or if there is no rule, the default action is to
9893 accept the incoming session. There is no specific limit to the number of
9894 rules which may be inserted.
9895
9896 Several types of actions are supported :
9897 - accept : the request is accepted
9898 - reject : the request is rejected and the connection is closed
9899 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
9900 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009901 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009902 - sc-set-gpt0(<sc-id>) <int>
9903 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009904 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009905 - silent-drop
9906
9907 These actions have the same meaning as their respective counter-parts in
9908 "tcp-request connection" and "tcp-request content", so please refer to these
9909 sections for a complete description.
9910
9911 Note that the "if/unless" condition is optional. If no condition is set on
9912 the action, it is simply performed unconditionally. That can be useful for
9913 "track-sc*" actions as well as for changing the default action to a reject.
9914
9915 Example: track the original source address by default, or the one advertised
9916 in the PROXY protocol header for connection coming from the local
9917 proxies. The first connection-level rule enables receipt of the
9918 PROXY protocol for these ones, the second rule tracks whatever
9919 address we decide to keep after optional decoding.
9920
9921 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9922 tcp-request session track-sc0 src
9923
9924 Example: accept all sessions from white-listed hosts, reject too fast
9925 sessions without counting them, and track accepted sessions.
9926 This results in session rate being capped from abusive sources.
9927
9928 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9929 tcp-request session reject if { src_sess_rate gt 10 }
9930 tcp-request session track-sc0 src
9931
9932 Example: accept all sessions from white-listed hosts, count all other
9933 sessions and reject too fast ones. This results in abusive ones
9934 being blocked as long as they don't slow down.
9935
9936 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9937 tcp-request session track-sc0 src
9938 tcp-request session reject if { sc0_sess_rate gt 10 }
9939
9940 See section 7 about ACL usage.
9941
9942 See also : "tcp-request connection", "tcp-request content", "stick-table"
9943
9944
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009945tcp-response inspect-delay <timeout>
9946 Set the maximum allowed time to wait for a response during content inspection
9947 May be used in sections : defaults | frontend | listen | backend
9948 no | no | yes | yes
9949 Arguments :
9950 <timeout> is the timeout value specified in milliseconds by default, but
9951 can be in any other unit if the number is suffixed by the unit,
9952 as explained at the top of this document.
9953
9954 See also : "tcp-response content", "tcp-request inspect-delay".
9955
9956
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009957timeout check <timeout>
9958 Set additional check timeout, but only after a connection has been already
9959 established.
9960
9961 May be used in sections: defaults | frontend | listen | backend
9962 yes | no | yes | yes
9963 Arguments:
9964 <timeout> is the timeout value specified in milliseconds by default, but
9965 can be in any other unit if the number is suffixed by the unit,
9966 as explained at the top of this document.
9967
9968 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
9969 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009970 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009971 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01009972 (Please also note that there is no valid reason to have such long connect
9973 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
9974 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009975
9976 If "timeout check" is not set haproxy uses "inter" for complete check
9977 timeout (connect + read) exactly like all <1.3.15 version.
9978
9979 In most cases check request is much simpler and faster to handle than normal
9980 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +01009981 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009982
9983 This parameter is specific to backends, but can be specified once for all in
9984 "defaults" sections. This is in fact one of the easiest solutions not to
9985 forget about it.
9986
Willy Tarreau41a340d2008-01-22 12:25:31 +01009987 See also: "timeout connect", "timeout queue", "timeout server",
9988 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009989
9990
Willy Tarreau0ba27502007-12-24 16:55:16 +01009991timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +01009992 Set the maximum inactivity time on the client side.
9993 May be used in sections : defaults | frontend | listen | backend
9994 yes | yes | yes | no
9995 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01009996 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01009997 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 The inactivity timeout applies when the client is expected to acknowledge or
10001 send data. In HTTP mode, this timeout is particularly important to consider
10002 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010003 response while it is reading data sent by the server. That said, for the
10004 first phase, it is preferable to set the "timeout http-request" to better
10005 protect HAProxy from Slowloris like attacks. The value is specified in
10006 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010007 suffixed by the unit, as specified at the top of this document. In TCP mode
10008 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10009 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010010 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010011 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010012 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10013 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010014 which overrides "timeout client" and "timeout server" for tunnels, as well as
10015 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010016
10017 This parameter is specific to frontends, but can be specified once for all in
10018 "defaults" sections. This is in fact one of the easiest solutions not to
10019 forget about it. An unspecified timeout results in an infinite timeout, which
10020 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010021 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010022 the system if the system's timeouts are not configured either.
10023
Willy Tarreau95c4e142017-11-26 12:18:55 +010010024 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010025
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010026 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010027
Willy Tarreau0ba27502007-12-24 16:55:16 +010010028
Willy Tarreau05cdd962014-05-10 14:30:07 +020010029timeout client-fin <timeout>
10030 Set the inactivity timeout on the client side for half-closed connections.
10031 May be used in sections : defaults | frontend | listen | backend
10032 yes | yes | yes | no
10033 Arguments :
10034 <timeout> is the timeout value specified in milliseconds by default, but
10035 can be in any other unit if the number is suffixed by the unit,
10036 as explained at the top of this document.
10037
10038 The inactivity timeout applies when the client is expected to acknowledge or
10039 send data while one direction is already shut down. This timeout is different
10040 from "timeout client" in that it only applies to connections which are closed
10041 in one direction. This is particularly useful to avoid keeping connections in
10042 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10043 problem is particularly common long connections such as RDP or WebSocket.
10044 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010045 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10046 frame was sent, often indicating an expectation that the connection quickly
10047 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010048
10049 This parameter is specific to frontends, but can be specified once for all in
10050 "defaults" sections. By default it is not set, so half-closed connections
10051 will use the other timeouts (timeout.client or timeout.tunnel).
10052
10053 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10054
10055
Willy Tarreau0ba27502007-12-24 16:55:16 +010010056timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010057 Set the maximum time to wait for a connection attempt to a server to succeed.
10058 May be used in sections : defaults | frontend | listen | backend
10059 yes | no | yes | yes
10060 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010061 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010062 can be in any other unit if the number is suffixed by the unit,
10063 as explained at the top of this document.
10064
10065 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010066 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010067 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010068 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010069 connect timeout also presets both queue and tarpit timeouts to the same value
10070 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010071
10072 This parameter is specific to backends, but can be specified once for all in
10073 "defaults" sections. This is in fact one of the easiest solutions not to
10074 forget about it. An unspecified timeout results in an infinite timeout, which
10075 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010076 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010077 the system if the system's timeouts are not configured either.
10078
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010079 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010080
Willy Tarreau0ba27502007-12-24 16:55:16 +010010081
Willy Tarreaub16a5742010-01-10 14:46:16 +010010082timeout http-keep-alive <timeout>
10083 Set the maximum allowed time to wait for a new HTTP request to appear
10084 May be used in sections : defaults | frontend | listen | backend
10085 yes | yes | yes | yes
10086 Arguments :
10087 <timeout> is the timeout value specified in milliseconds by default, but
10088 can be in any other unit if the number is suffixed by the unit,
10089 as explained at the top of this document.
10090
10091 By default, the time to wait for a new request in case of keep-alive is set
10092 by "timeout http-request". However this is not always convenient because some
10093 people want very short keep-alive timeouts in order to release connections
10094 faster, and others prefer to have larger ones but still have short timeouts
10095 once the request has started to present itself.
10096
10097 The "http-keep-alive" timeout covers these needs. It will define how long to
10098 wait for a new HTTP request to start coming after a response was sent. Once
10099 the first byte of request has been seen, the "http-request" timeout is used
10100 to wait for the complete request to come. Note that empty lines prior to a
10101 new request do not refresh the timeout and are not counted as a new request.
10102
10103 There is also another difference between the two timeouts : when a connection
10104 expires during timeout http-keep-alive, no error is returned, the connection
10105 just closes. If the connection expires in "http-request" while waiting for a
10106 connection to complete, a HTTP 408 error is returned.
10107
10108 In general it is optimal to set this value to a few tens to hundreds of
10109 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010110 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010111 1 millisecond) it will probably only accept pipelined requests but not the
10112 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010113 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010114
10115 If this parameter is not set, the "http-request" timeout applies, and if both
10116 are not set, "timeout client" still applies at the lower level. It should be
10117 set in the frontend to take effect, unless the frontend is in TCP mode, in
10118 which case the HTTP backend's timeout will be used.
10119
Willy Tarreau95c4e142017-11-26 12:18:55 +010010120 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10121 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010122 (where we only have one connection per client and a connection setup).
10123
Willy Tarreaub16a5742010-01-10 14:46:16 +010010124 See also : "timeout http-request", "timeout client".
10125
10126
Willy Tarreau036fae02008-01-06 13:24:40 +010010127timeout http-request <timeout>
10128 Set the maximum allowed time to wait for a complete HTTP request
10129 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010130 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010131 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010132 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010133 can be in any other unit if the number is suffixed by the unit,
10134 as explained at the top of this document.
10135
10136 In order to offer DoS protection, it may be required to lower the maximum
10137 accepted time to receive a complete HTTP request without affecting the client
10138 timeout. This helps protecting against established connections on which
10139 nothing is sent. The client timeout cannot offer a good protection against
10140 this abuse because it is an inactivity timeout, which means that if the
10141 attacker sends one character every now and then, the timeout will not
10142 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010143 types, the request will be aborted if it does not complete in time. When the
10144 timeout expires, an HTTP 408 response is sent to the client to inform it
10145 about the problem, and the connection is closed. The logs will report
10146 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010147 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010148 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10149 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010150
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010151 By default, this timeout only applies to the header part of the request,
10152 and not to any data. As soon as the empty line is received, this timeout is
10153 not used anymore. When combined with "option http-buffer-request", this
10154 timeout also applies to the body of the request..
10155 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010156 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010157
10158 Generally it is enough to set it to a few seconds, as most clients send the
10159 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010160 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010161 generally work on local networks as long as there are no packet losses. This
10162 will prevent people from sending bare HTTP requests using telnet.
10163
10164 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010165 chunk of the incoming request. It should be set in the frontend to take
10166 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10167 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010168
Willy Tarreau0f228a02015-05-01 15:37:53 +020010169 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010170 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010171
Willy Tarreau844e3c52008-01-11 16:28:18 +010010172
10173timeout queue <timeout>
10174 Set the maximum time to wait in the queue for a connection slot to be free
10175 May be used in sections : defaults | frontend | listen | backend
10176 yes | no | yes | yes
10177 Arguments :
10178 <timeout> is the timeout value specified in milliseconds by default, but
10179 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 When a server's maxconn is reached, connections are left pending in a queue
10183 which may be server-specific or global to the backend. In order not to wait
10184 indefinitely, a timeout is applied to requests pending in the queue. If the
10185 timeout is reached, it is considered that the request will almost never be
10186 served, so it is dropped and a 503 error is returned to the client.
10187
10188 The "timeout queue" statement allows to fix the maximum time for a request to
10189 be left pending in a queue. If unspecified, the same value as the backend's
10190 connection timeout ("timeout connect") is used, for backwards compatibility
10191 with older versions with no "timeout queue" parameter.
10192
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010193 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010194
10195
10196timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010010197 Set the maximum inactivity time on the server side.
10198 May be used in sections : defaults | frontend | listen | backend
10199 yes | no | yes | yes
10200 Arguments :
10201 <timeout> is the timeout value specified in milliseconds by default, but
10202 can be in any other unit if the number is suffixed by the unit,
10203 as explained at the top of this document.
10204
10205 The inactivity timeout applies when the server is expected to acknowledge or
10206 send data. In HTTP mode, this timeout is particularly important to consider
10207 during the first phase of the server's response, when it has to send the
10208 headers, as it directly represents the server's processing time for the
10209 request. To find out what value to put there, it's often good to start with
10210 what would be considered as unacceptable response times, then check the logs
10211 to observe the response time distribution, and adjust the value accordingly.
10212
10213 The value is specified in milliseconds by default, but can be in any other
10214 unit if the number is suffixed by the unit, as specified at the top of this
10215 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10216 recommended that the client timeout remains equal to the server timeout in
10217 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010218 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010219 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010220 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10221 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010222 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10223 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010224
10225 This parameter is specific to backends, but can be specified once for all in
10226 "defaults" sections. This is in fact one of the easiest solutions not to
10227 forget about it. An unspecified timeout results in an infinite timeout, which
10228 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010229 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010010230 the system if the system's timeouts are not configured either.
10231
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010232 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010233
Willy Tarreau05cdd962014-05-10 14:30:07 +020010234
10235timeout server-fin <timeout>
10236 Set the inactivity timeout on the server side for half-closed connections.
10237 May be used in sections : defaults | frontend | listen | backend
10238 yes | no | yes | yes
10239 Arguments :
10240 <timeout> is the timeout value specified in milliseconds by default, but
10241 can be in any other unit if the number is suffixed by the unit,
10242 as explained at the top of this document.
10243
10244 The inactivity timeout applies when the server is expected to acknowledge or
10245 send data while one direction is already shut down. This timeout is different
10246 from "timeout server" in that it only applies to connections which are closed
10247 in one direction. This is particularly useful to avoid keeping connections in
10248 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10249 This problem is particularly common long connections such as RDP or WebSocket.
10250 Note that this timeout can override "timeout tunnel" when a connection shuts
10251 down in one direction. This setting was provided for completeness, but in most
10252 situations, it should not be needed.
10253
10254 This parameter is specific to backends, but can be specified once for all in
10255 "defaults" sections. By default it is not set, so half-closed connections
10256 will use the other timeouts (timeout.server or timeout.tunnel).
10257
10258 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10259
Willy Tarreau844e3c52008-01-11 16:28:18 +010010260
10261timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010262 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010263 May be used in sections : defaults | frontend | listen | backend
10264 yes | yes | yes | yes
10265 Arguments :
10266 <timeout> is the tarpit duration specified in milliseconds by default, but
10267 can be in any other unit if the number is suffixed by the unit,
10268 as explained at the top of this document.
10269
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020010270 When a connection is tarpitted using "http-request tarpit", it is maintained
10271 open with no activity for a certain amount of time, then closed. "timeout
10272 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010273
10274 The value is specified in milliseconds by default, but can be in any other
10275 unit if the number is suffixed by the unit, as specified at the top of this
10276 document. If unspecified, the same value as the backend's connection timeout
10277 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010278 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010279
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010280 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010281
10282
Willy Tarreauce887fd2012-05-12 12:50:00 +020010283timeout tunnel <timeout>
10284 Set the maximum inactivity time on the client and server side for tunnels.
10285 May be used in sections : defaults | frontend | listen | backend
10286 yes | no | yes | yes
10287 Arguments :
10288 <timeout> is the timeout value specified in milliseconds by default, but
10289 can be in any other unit if the number is suffixed by the unit,
10290 as explained at the top of this document.
10291
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010292 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010293 between a client and a server, and the connection remains inactive in both
10294 directions. This timeout supersedes both the client and server timeouts once
10295 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010296 analyzer remains attached to either connection (e.g. tcp content rules are
10297 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010298 when switching to the WebSocket protocol, or forwarding a CONNECT request
10299 to a proxy), or after the first response when no keepalive/close option is
10300 specified.
10301
Willy Tarreau05cdd962014-05-10 14:30:07 +020010302 Since this timeout is usually used in conjunction with long-lived connections,
10303 it usually is a good idea to also set "timeout client-fin" to handle the
10304 situation where a client suddenly disappears from the net and does not
10305 acknowledge a close, or sends a shutdown and does not acknowledge pending
10306 data anymore. This can happen in lossy networks where firewalls are present,
10307 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10308 state.
10309
Willy Tarreauce887fd2012-05-12 12:50:00 +020010310 The value is specified in milliseconds by default, but can be in any other
10311 unit if the number is suffixed by the unit, as specified at the top of this
10312 document. Whatever the expected normal idle time, it is a good practice to
10313 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010314 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010315
10316 This parameter is specific to backends, but can be specified once for all in
10317 "defaults" sections. This is in fact one of the easiest solutions not to
10318 forget about it.
10319
10320 Example :
10321 defaults http
10322 option http-server-close
10323 timeout connect 5s
10324 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010325 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010326 timeout server 30s
10327 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10328
Willy Tarreau05cdd962014-05-10 14:30:07 +020010329 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010330
10331
Willy Tarreau844e3c52008-01-11 16:28:18 +010010332transparent (deprecated)
10333 Enable client-side transparent proxying
10334 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010335 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010336 Arguments : none
10337
10338 This keyword was introduced in order to provide layer 7 persistence to layer
10339 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10340 connection for a remote address to a local process (here HAProxy), and let
10341 this process know what address was initially requested. When this option is
10342 used, sessions without cookies will be forwarded to the original destination
10343 IP address of the incoming request (which should match that of another
10344 equipment), while requests with cookies will still be forwarded to the
10345 appropriate server.
10346
10347 The "transparent" keyword is deprecated, use "option transparent" instead.
10348
10349 Note that contrary to a common belief, this option does NOT make HAProxy
10350 present the client's IP to the server when establishing the connection.
10351
Willy Tarreau844e3c52008-01-11 16:28:18 +010010352 See also: "option transparent"
10353
William Lallemanda73203e2012-03-12 12:48:57 +010010354unique-id-format <string>
10355 Generate a unique ID for each request.
10356 May be used in sections : defaults | frontend | listen | backend
10357 yes | yes | yes | no
10358 Arguments :
10359 <string> is a log-format string.
10360
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010361 This keyword creates a ID for each request using the custom log format. A
10362 unique ID is useful to trace a request passing through many components of
10363 a complex infrastructure. The newly created ID may also be logged using the
10364 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010365
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010366 The format should be composed from elements that are guaranteed to be
10367 unique when combined together. For instance, if multiple haproxy instances
10368 are involved, it might be important to include the node name. It is often
10369 needed to log the incoming connection's source and destination addresses
10370 and ports. Note that since multiple requests may be performed over the same
10371 connection, including a request counter may help differentiate them.
10372 Similarly, a timestamp may protect against a rollover of the counter.
10373 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010374
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010375 It is recommended to use hexadecimal notation for many fields since it
10376 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010377
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010378 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010379
Julien Vehentf21be322014-03-07 08:27:34 -050010380 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010381
10382 will generate:
10383
10384 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10385
10386 See also: "unique-id-header"
10387
10388unique-id-header <name>
10389 Add a unique ID header in the HTTP request.
10390 May be used in sections : defaults | frontend | listen | backend
10391 yes | yes | yes | no
10392 Arguments :
10393 <name> is the name of the header.
10394
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010395 Add a unique-id header in the HTTP request sent to the server, using the
10396 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010397
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010398 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010399
Julien Vehentf21be322014-03-07 08:27:34 -050010400 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010401 unique-id-header X-Unique-ID
10402
10403 will generate:
10404
10405 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10406
10407 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010408
Willy Tarreauf51658d2014-04-23 01:21:56 +020010409use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010410 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010411 May be used in sections : defaults | frontend | listen | backend
10412 no | yes | yes | no
10413 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010414 <backend> is the name of a valid backend or "listen" section, or a
10415 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010416
Willy Tarreauf51658d2014-04-23 01:21:56 +020010417 <condition> is a condition composed of ACLs, as described in section 7. If
10418 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010419
10420 When doing content-switching, connections arrive on a frontend and are then
10421 dispatched to various backends depending on a number of conditions. The
10422 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010423 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010424 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010425 source address validation) or combined with a "tcp-request" rule to wait for
10426 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010427
10428 There may be as many "use_backend" rules as desired. All of these rules are
10429 evaluated in their declaration order, and the first one which matches will
10430 assign the backend.
10431
10432 In the first form, the backend will be used if the condition is met. In the
10433 second form, the backend will be used if the condition is not met. If no
10434 condition is valid, the backend defined with "default_backend" will be used.
10435 If no default backend is defined, either the servers in the same section are
10436 used (in case of a "listen" section) or, in case of a frontend, no server is
10437 used and a 503 service unavailable response is returned.
10438
Willy Tarreau51aecc72009-07-12 09:47:04 +020010439 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010440 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010441 and backend processing will immediately follow, or the backend will wait for
10442 a complete HTTP request to get in. This feature is useful when a frontend
10443 must decode several protocols on a unique port, one of them being HTTP.
10444
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010445 When <backend> is a simple name, it is resolved at configuration time, and an
10446 error is reported if the specified backend does not exist. If <backend> is
10447 a log-format string instead, no check may be done at configuration time, so
10448 the backend name is resolved dynamically at run time. If the resulting
10449 backend name does not correspond to any valid backend, no other rule is
10450 evaluated, and the default_backend directive is applied instead. Note that
10451 when using dynamic backend names, it is highly recommended to use a prefix
10452 that no other backend uses in order to ensure that an unauthorized backend
10453 cannot be forced from the request.
10454
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010455 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010456 used to detect the association between frontends and backends to compute the
10457 backend's "fullconn" setting. This cannot be done for dynamic names.
10458
10459 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10460 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010461
Willy Tarreau036fae02008-01-06 13:24:40 +010010462
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010463use-server <server> if <condition>
10464use-server <server> unless <condition>
10465 Only use a specific server if/unless an ACL-based condition is matched.
10466 May be used in sections : defaults | frontend | listen | backend
10467 no | no | yes | yes
10468 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010469 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010470
10471 <condition> is a condition composed of ACLs, as described in section 7.
10472
10473 By default, connections which arrive to a backend are load-balanced across
10474 the available servers according to the configured algorithm, unless a
10475 persistence mechanism such as a cookie is used and found in the request.
10476
10477 Sometimes it is desirable to forward a particular request to a specific
10478 server without having to declare a dedicated backend for this server. This
10479 can be achieved using the "use-server" rules. These rules are evaluated after
10480 the "redirect" rules and before evaluating cookies, and they have precedence
10481 on them. There may be as many "use-server" rules as desired. All of these
10482 rules are evaluated in their declaration order, and the first one which
10483 matches will assign the server.
10484
10485 If a rule designates a server which is down, and "option persist" is not used
10486 and no force-persist rule was validated, it is ignored and evaluation goes on
10487 with the next rules until one matches.
10488
10489 In the first form, the server will be used if the condition is met. In the
10490 second form, the server will be used if the condition is not met. If no
10491 condition is valid, the processing continues and the server will be assigned
10492 according to other persistence mechanisms.
10493
10494 Note that even if a rule is matched, cookie processing is still performed but
10495 does not assign the server. This allows prefixed cookies to have their prefix
10496 stripped.
10497
10498 The "use-server" statement works both in HTTP and TCP mode. This makes it
10499 suitable for use with content-based inspection. For instance, a server could
10500 be selected in a farm according to the TLS SNI field. And if these servers
10501 have their weight set to zero, they will not be used for other traffic.
10502
10503 Example :
10504 # intercept incoming TLS requests based on the SNI field
10505 use-server www if { req_ssl_sni -i www.example.com }
10506 server www 192.168.0.1:443 weight 0
10507 use-server mail if { req_ssl_sni -i mail.example.com }
10508 server mail 192.168.0.1:587 weight 0
10509 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010510 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010511 # all the rest is forwarded to this server
10512 server default 192.168.0.2:443 check
10513
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010514 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010515
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010516
Davor Ocelice9ed2812017-12-25 17:49:28 +0100105175. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010518--------------------------
10519
10520The "bind", "server" and "default-server" keywords support a number of settings
10521depending on some build options and on the system HAProxy was built on. These
10522settings generally each consist in one word sometimes followed by a value,
10523written on the same line as the "bind" or "server" line. All these options are
10524described in this section.
10525
10526
105275.1. Bind options
10528-----------------
10529
10530The "bind" keyword supports a certain number of settings which are all passed
10531as arguments on the same line. The order in which those arguments appear makes
10532no importance, provided that they appear after the bind address. All of these
10533parameters are optional. Some of them consist in a single words (booleans),
10534while other ones expect a value after them. In this case, the value must be
10535provided immediately after the setting name.
10536
10537The currently supported settings are the following ones.
10538
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010539accept-netscaler-cip <magic number>
10540 Enforces the use of the NetScaler Client IP insertion protocol over any
10541 connection accepted by any of the TCP sockets declared on the same line. The
10542 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10543 the incoming connection to be used everywhere an address is used, with the
10544 only exception of "tcp-request connection" rules which will only see the
10545 real connection address. Logs will reflect the addresses indicated in the
10546 protocol, unless it is violated, in which case the real address will still
10547 be used. This keyword combined with support from external components can be
10548 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010549 mechanism which is not always reliable and not even always usable. See also
10550 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10551 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010552
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010553accept-proxy
10554 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010555 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10556 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010557 3/4 addresses of the incoming connection to be used everywhere an address is
10558 used, with the only exception of "tcp-request connection" rules which will
10559 only see the real connection address. Logs will reflect the addresses
10560 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010561 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010562 components can be used as an efficient and reliable alternative to the
10563 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010564 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10565 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010566
Olivier Houchardc2aae742017-09-22 18:26:28 +020010567allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010568 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010010569 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050010570 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010010571 that are idempotent. You can use the "wait-for-handshake" action for any
10572 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020010573
Willy Tarreauab861d32013-04-02 02:30:41 +020010574alpn <protocols>
10575 This enables the TLS ALPN extension and advertises the specified protocol
10576 list as supported on top of ALPN. The protocol list consists in a comma-
10577 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050010578 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020010579 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010580 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10581 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10582 now obsolete NPN extension. At the time of writing this, most browsers still
10583 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10584 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10585 are expected to be supported, both versions can be advertised, in order of
10586 preference, like below :
10587
10588 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010589
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010590backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010010591 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010592 backlog is used instead, which generally defaults to the maxconn value.
10593
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010594curves <curves>
10595 This setting is only available when support for OpenSSL was built in. It sets
10596 the string describing the list of elliptic curves algorithms ("curve suite")
10597 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10598 string is a colon-delimited list of curve name.
10599 Example: "X25519:P-256" (without quote)
10600 When "curves" is set, "ecdhe" parameter is ignored.
10601
Emeric Brun7fb34422012-09-28 15:26:15 +020010602ecdhe <named curve>
10603 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010604 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10605 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010606
Emeric Brunfd33a262012-10-11 16:28:27 +020010607ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010608 This setting is only available when support for OpenSSL was built in. It
10609 designates a PEM file from which to load CA certificates used to verify
10610 client's certificate.
10611
Emeric Brunb6dc9342012-09-28 17:55:37 +020010612ca-ignore-err [all|<errorID>,...]
10613 This setting is only available when support for OpenSSL was built in.
10614 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10615 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10616 error is ignored.
10617
Christopher Faulet31af49d2015-06-09 17:29:50 +020010618ca-sign-file <cafile>
10619 This setting is only available when support for OpenSSL was built in. It
10620 designates a PEM file containing both the CA certificate and the CA private
10621 key used to create and sign server's certificates. This is a mandatory
10622 setting when the dynamic generation of certificates is enabled. See
10623 'generate-certificates' for details.
10624
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010625ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010626 This setting is only available when support for OpenSSL was built in. It is
10627 the CA private key passphrase. This setting is optional and used only when
10628 the dynamic generation of certificates is enabled. See
10629 'generate-certificates' for details.
10630
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010631ciphers <ciphers>
10632 This setting is only available when support for OpenSSL was built in. It sets
10633 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000010634 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010635 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010636 information and recommendations see e.g.
10637 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10638 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10639 cipher configuration, please check the "ciphersuites" keyword.
10640
10641ciphersuites <ciphersuites>
10642 This setting is only available when support for OpenSSL was built in and
10643 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10644 the list of cipher algorithms ("cipher suite") that are negotiated during the
10645 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010646 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
10647 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010648
Emeric Brunfd33a262012-10-11 16:28:27 +020010649crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010650 This setting is only available when support for OpenSSL was built in. It
10651 designates a PEM file from which to load certificate revocation list used
10652 to verify client's certificate.
10653
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010654crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010655 This setting is only available when support for OpenSSL was built in. It
10656 designates a PEM file containing both the required certificates and any
10657 associated private keys. This file can be built by concatenating multiple
10658 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10659 requires an intermediate certificate, this can also be concatenated into this
10660 file.
10661
10662 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10663 are loaded.
10664
10665 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010666 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010667 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10668 specified multiple times in order to load certificates from multiple files or
10669 directories. The certificates will be presented to clients who provide a
10670 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010671 subjects. Wildcards are supported, where a wildcard character '*' is used
10672 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010673 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010674
10675 If no SNI is provided by the client or if the SSL library does not support
10676 TLS extensions, or if the client provides an SNI hostname which does not
10677 match any certificate, then the first loaded certificate will be presented.
10678 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010679 recommended to load the default one first as a file or to ensure that it will
10680 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010681
Emeric Brune032bfa2012-09-28 13:01:45 +020010682 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010683
Davor Ocelice9ed2812017-12-25 17:49:28 +010010684 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010685 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010686 choose a web server that the CA believes requires an intermediate CA (for
10687 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010688 others, e.g. nginx, result in a wrong bundle that will not work for some
10689 clients).
10690
Emeric Brun4147b2e2014-06-16 18:36:30 +020010691 For each PEM file, haproxy checks for the presence of file at the same path
10692 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10693 Status Request extension (also known as "OCSP stapling") is automatically
10694 enabled. The content of this file is optional. If not empty, it must contain
10695 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10696 must comply with the following rules: it has to indicate a good status,
10697 it has to be a single response for the certificate of the PEM file, and it
10698 has to be valid at the moment of addition. If these rules are not respected
10699 the OCSP Response is ignored and a warning is emitted. In order to identify
10700 which certificate an OCSP Response applies to, the issuer's certificate is
10701 necessary. If the issuer's certificate is not found in the PEM file, it will
10702 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10703 if it exists otherwise it will fail with an error.
10704
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010705 For each PEM file, haproxy also checks for the presence of file at the same
10706 path suffixed by ".sctl". If such file is found, support for Certificate
10707 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10708 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10709 to check basic syntax, but no signatures are verified.
10710
yanbzhu6c25e9e2016-01-05 12:52:02 -050010711 There are cases where it is desirable to support multiple key types, e.g. RSA
10712 and ECDSA in the cipher suites offered to the clients. This allows clients
10713 that support EC certificates to be able to use EC ciphers, while
10714 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010715
10716 In order to provide this functionality, multiple PEM files, each with a
10717 different key type, are required. To associate these PEM files into a
10718 "cert bundle" that is recognized by haproxy, they must be named in the
10719 following way: All PEM files that are to be bundled must have the same base
10720 name, with a suffix indicating the key type. Currently, three suffixes are
10721 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10722 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10723 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10724 suffix matters. To load this bundle into haproxy, specify the base name only:
10725
10726 Example : bind :8443 ssl crt example.pem
10727
yanbzhu6c25e9e2016-01-05 12:52:02 -050010728 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010729 a cert bundle.
10730
Davor Ocelice9ed2812017-12-25 17:49:28 +010010731 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010732 support multiple key types. PEM files are combined based on Common Name
10733 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10734 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10735 entries in the certificates in that bundle, haproxy will not be able to
10736 provide multi-cert support.
10737
10738 Assuming bundle in the example above contained the following:
10739
10740 Filename | CN | SAN
10741 -------------------+-----------------+-------------------
10742 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010743 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010744 example.pem.ecdsa | www.example.com | ecdsa.example.com
10745 -------------------+-----------------+-------------------
10746
10747 Users connecting with an SNI of "www.example.com" will be able
10748 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10749 "rsa.example.com" will only be able to use RSA cipher suites, and users
10750 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010751 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10752 no need to bundle certificates. ECDSA certificate will be preferred if client
10753 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010754
10755 If a directory name is given as the <cert> argument, haproxy will
10756 automatically search and load bundled files in that directory.
10757
10758 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10759 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10760 time, sctl is not supported in multi-certificate bundling.
10761
Emeric Brunb6dc9342012-09-28 17:55:37 +020010762crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010763 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010764 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010765 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010766 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010767
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010768crt-list <file>
10769 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010770 designates a list of PEM file with an optional ssl configuration and a SNI
10771 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010772
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010773 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10774
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010775 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10776 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010777 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010778 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010779
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010780 Wildcards are supported in the SNI filter. Negative filter are also supported,
10781 only useful in combination with a wildcard filter to exclude a particular SNI.
10782 The certificates will be presented to clients who provide a valid TLS Server
10783 Name Indication field matching one of the SNI filters. If no SNI filter is
10784 specified, the CN and alt subjects are used. This directive may be specified
10785 multiple times. See the "crt" option for more information. The default
10786 certificate is still needed to meet OpenSSL expectations. If it is not used,
10787 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010788
yanbzhu6c25e9e2016-01-05 12:52:02 -050010789 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010790 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010791 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10792 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10793 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010794
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010795 crt-list file example:
10796 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010797 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010798 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010799 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010800
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010801defer-accept
10802 Is an optional keyword which is supported only on certain Linux kernels. It
10803 states that a connection will only be accepted once some data arrive on it,
10804 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010010805 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010806 performance by ensuring that most of the request is already available when
10807 the connection is accepted. On the other hand, it will not be able to detect
10808 connections which don't talk. It is important to note that this option is
10809 broken in all kernels up to 2.6.31, as the connection is never accepted until
10810 the client talks. This can cause issues with front firewalls which would see
10811 an established connection while the proxy will only see it in SYN_RECV. This
10812 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
10813
William Lallemandf6975e92017-05-26 17:42:10 +020010814expose-fd listeners
10815 This option is only usable with the stats socket. It gives your stats socket
10816 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020010817 During a reload with the master-worker mode, the process is automatically
10818 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010010819 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020010820
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010821force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010822 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010823 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010824 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010825 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010826
10827force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010828 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010829 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010830 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010831
10832force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010833 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010834 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010835 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010836
10837force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010838 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010839 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010840 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010841
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010842force-tlsv13
10843 This option enforces use of TLSv1.3 only on SSL connections instantiated from
10844 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010845 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010846
Christopher Faulet31af49d2015-06-09 17:29:50 +020010847generate-certificates
10848 This setting is only available when support for OpenSSL was built in. It
10849 enables the dynamic SSL certificates generation. A CA certificate and its
10850 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
10851 a transparent forward proxy, SSL requests generate errors because of a common
10852 name mismatch on the certificate presented to the client. With this option
10853 enabled, HAProxy will try to forge a certificate using the SNI hostname
10854 indicated by the client. This is done only if no certificate matches the SNI
10855 hostname (see 'crt-list'). If an error occurs, the default certificate is
10856 used, else the 'strict-sni' option is set.
10857 It can also be used when HAProxy is configured as a reverse proxy to ease the
10858 deployment of an architecture with many backends.
10859
10860 Creating a SSL certificate is an expensive operation, so a LRU cache is used
10861 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010010862 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020010863 certificate is used many times.
10864
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010865gid <gid>
10866 Sets the group of the UNIX sockets to the designated system gid. It can also
10867 be set by default in the global section's "unix-bind" statement. Note that
10868 some platforms simply ignore this. This setting is equivalent to the "group"
10869 setting except that the group ID is used instead of its name. This setting is
10870 ignored by non UNIX sockets.
10871
10872group <group>
10873 Sets the group of the UNIX sockets to the designated system group. It can
10874 also be set by default in the global section's "unix-bind" statement. Note
10875 that some platforms simply ignore this. This setting is equivalent to the
10876 "gid" setting except that the group name is used instead of its gid. This
10877 setting is ignored by non UNIX sockets.
10878
10879id <id>
10880 Fixes the socket ID. By default, socket IDs are automatically assigned, but
10881 sometimes it is more convenient to fix them to ease monitoring. This value
10882 must be strictly positive and unique within the listener/frontend. This
10883 option can only be used when defining only a single socket.
10884
10885interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010010886 Restricts the socket to a specific interface. When specified, only packets
10887 received from that particular interface are processed by the socket. This is
10888 currently only supported on Linux. The interface must be a primary system
10889 interface, not an aliased interface. It is also possible to bind multiple
10890 frontends to the same address if they are bound to different interfaces. Note
10891 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010010892 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
10893 uses the same interface as inbound traffic, and its associated routing table,
10894 even if there are explicit routes through different interfaces configured.
10895 This can prove useful to address asymmetric routing issues when the same
10896 client IP addresses need to be able to reach frontends hosted on different
10897 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010898
Willy Tarreauabb175f2012-09-24 12:43:26 +020010899level <level>
10900 This setting is used with the stats sockets only to restrict the nature of
10901 the commands that can be issued on the socket. It is ignored by other
10902 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010010903 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020010904 read, and no change is allowed. It would make sense on systems where it
10905 is not easy to restrict access to the socket.
10906 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010907 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020010908 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010010909 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020010910 all counters).
10911
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020010912severity-output <format>
10913 This setting is used with the stats sockets only to configure severity
10914 level output prepended to informational feedback messages. Severity
10915 level of messages can range between 0 and 7, conforming to syslog
10916 rfc5424. Valid and successful socket commands requesting data
10917 (i.e. "show map", "get acl foo" etc.) will never have a severity level
10918 prepended. It is ignored by other sockets. <format> can be one of :
10919 - "none" (default) no severity level is prepended to feedback messages.
10920 - "number" severity level is prepended as a number.
10921 - "string" severity level is prepended as a string following the
10922 rfc5424 convention.
10923
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010924maxconn <maxconn>
10925 Limits the sockets to this number of concurrent connections. Extraneous
10926 connections will remain in the system's backlog until a connection is
10927 released. If unspecified, the limit will be the same as the frontend's
10928 maxconn. Note that in case of port ranges or multiple addresses, the same
10929 value will be applied to each socket. This setting enables different
10930 limitations on expensive sockets, for instance SSL entries which may easily
10931 eat all memory.
10932
10933mode <mode>
10934 Sets the octal mode used to define access permissions on the UNIX socket. It
10935 can also be set by default in the global section's "unix-bind" statement.
10936 Note that some platforms simply ignore this. This setting is ignored by non
10937 UNIX sockets.
10938
10939mss <maxseg>
10940 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
10941 connections. This can be used to force a lower MSS for certain specific
10942 ports, for instance for connections passing through a VPN. Note that this
10943 relies on a kernel feature which is theoretically supported under Linux but
10944 was buggy in all versions prior to 2.6.28. It may or may not work on other
10945 operating systems. It may also not change the advertised value but change the
10946 effective size of outgoing segments. The commonly advertised value for TCPv4
10947 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
10948 positive, it will be used as the advertised MSS. If it is negative, it will
10949 indicate by how much to reduce the incoming connection's advertised MSS for
10950 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
10951
10952name <name>
10953 Sets an optional name for these sockets, which will be reported on the stats
10954 page.
10955
Willy Tarreaud72f0f32015-10-13 14:50:22 +020010956namespace <name>
10957 On Linux, it is possible to specify which network namespace a socket will
10958 belong to. This directive makes it possible to explicitly bind a listener to
10959 a namespace different from the default one. Please refer to your operating
10960 system's documentation to find more details about network namespaces.
10961
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010962nice <nice>
10963 Sets the 'niceness' of connections initiated from the socket. Value must be
10964 in the range -1024..1024 inclusive, and defaults to zero. Positive values
10965 means that such connections are more friendly to others and easily offer
10966 their place in the scheduler. On the opposite, negative values mean that
10967 connections want to run with a higher priority than others. The difference
10968 only happens under high loads when the system is close to saturation.
10969 Negative values are appropriate for low-latency or administration services,
10970 and high values are generally recommended for CPU intensive tasks such as SSL
10971 processing or bulk transfers which are less sensible to latency. For example,
10972 it may make sense to use a positive value for an SMTP socket and a negative
10973 one for an RDP socket.
10974
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010975no-ca-names
10976 This setting is only available when support for OpenSSL was built in. It
10977 prevents from send CA names in server hello message when ca-file is used.
10978
Emeric Brun9b3009b2012-10-05 11:55:06 +020010979no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010980 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010981 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010982 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010983 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010984 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
10985 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010986
Emeric Brun90ad8722012-10-02 14:00:59 +020010987no-tls-tickets
10988 This setting is only available when support for OpenSSL was built in. It
10989 disables the stateless session resumption (RFC 5077 TLS Ticket
10990 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010991 session resumption is more expensive in CPU usage. This option is also
10992 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020010993
Emeric Brun9b3009b2012-10-05 11:55:06 +020010994no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010995 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010996 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010997 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010998 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010999 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11000 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011001
Emeric Brun9b3009b2012-10-05 11:55:06 +020011002no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011003 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011004 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011005 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011006 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011007 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11008 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011009
Emeric Brun9b3009b2012-10-05 11:55:06 +020011010no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011011 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011012 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011013 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011014 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011015 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11016 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011017
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011018no-tlsv13
11019 This setting is only available when support for OpenSSL was built in. It
11020 disables support for TLSv1.3 on any sockets instantiated from the listener
11021 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11022 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011023 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11024 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011025
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011026npn <protocols>
11027 This enables the NPN TLS extension and advertises the specified protocol list
11028 as supported on top of NPN. The protocol list consists in a comma-delimited
11029 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050011030 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011031 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011032 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11033 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11034 version of OpenSSL, NPN might still be used as most clients still support it
11035 at the time of writing this. It is possible to enable both NPN and ALPN
11036 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011037
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011038prefer-client-ciphers
11039 Use the client's preference when selecting the cipher suite, by default
11040 the server's preference is enforced. This option is also available on
11041 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011042 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11043 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11044 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011045
Christopher Fauletc644fa92017-11-23 22:44:11 +010011046process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010011047 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010011048 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011049 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011050 between the two is applied. If in the end the listener is not allowed to run
11051 on any remaining process, a warning is emitted, and the listener will either
11052 run on the first process of the listener if a single process was specified,
11053 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011054 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010011055 connections for this listener, for the the process set. If multiple processes
11056 and threads are configured, a warning is emitted, as it either results from a
11057 configuration error or a misunderstanding of these models. For the unlikely
11058 case where several ranges are needed, this directive may be repeated.
11059 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010011060
11061 all | odd | even | number[-[number]]
11062
11063 Ranges can be partially defined. The higher bound can be omitted. In such
11064 case, it is replaced by the corresponding maximum value. The main purpose of
11065 this directive is to be used with the stats sockets and have one different
11066 socket per process. The second purpose is to have multiple bind lines sharing
11067 the same IP:port but not the same process in a listener, so that the system
11068 can distribute the incoming connections into multiple queues and allow a
11069 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11070 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011071
Christopher Fauleta717b992018-04-10 14:43:00 +020011072proto <name>
11073 Forces the multiplexer's protocol to use for the incoming connections. It
11074 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11075 be usable on the frontend side. The list of available protocols is reported
11076 in haproxy -vv.
11077 Idea behind this optipon is to bypass the selection of the best multiplexer's
11078 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011079 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011080 h2" on the bind line.
11081
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011082ssl
11083 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011084 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011085 certificate is necessary (see "crt" above). All contents in the buffers will
11086 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011087 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11088 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011089
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011090ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11091 This option enforces use of <version> or lower on SSL connections instantiated
11092 from this listener. This option is also available on global statement
11093 "ssl-default-bind-options". See also "ssl-min-ver".
11094
11095ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11096 This option enforces use of <version> or upper on SSL connections instantiated
11097 from this listener. This option is also available on global statement
11098 "ssl-default-bind-options". See also "ssl-max-ver".
11099
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011100strict-sni
11101 This setting is only available when support for OpenSSL was built in. The
11102 SSL/TLS negotiation is allow only if the client provided an SNI which match
11103 a certificate. The default certificate is not used.
11104 See the "crt" option for more information.
11105
Willy Tarreau2af207a2015-02-04 00:45:58 +010011106tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011107 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011108 listening socket. This option is available on Linux since version 2.6.37. It
11109 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011110 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011111 useful on long-lived connections experiencing long idle periods such as
11112 remote terminals or database connection pools, where the client and server
11113 timeouts must remain high to allow a long period of idle, but where it is
11114 important to detect that the client has disappeared in order to release all
11115 resources associated with its connection (and the server's session). The
11116 argument is a delay expressed in milliseconds by default. This only works
11117 for regular TCP connections, and is ignored for other protocols.
11118
Willy Tarreau1c862c52012-10-05 16:21:00 +020011119tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011120 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011121 enables TCP Fast Open on the listening socket, which means that clients which
11122 support this feature will be able to send a request and receive a response
11123 during the 3-way handshake starting from second connection, thus saving one
11124 round-trip after the first connection. This only makes sense with protocols
11125 that use high connection rates and where each round trip matters. This can
11126 possibly cause issues with many firewalls which do not accept data on SYN
11127 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011128 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11129 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11130 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011131
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011132tls-ticket-keys <keyfile>
11133 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010011134 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
11135 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
11136 The first key determines the key length used for next keys: you can't mix
11137 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
11138 build option (default 3) and at least as many keys need to be present in
11139 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
11140 penultimate one for encryption. This enables easy key rotation by just
11141 appending new key to the file and reloading the process. Keys must be
11142 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
11143 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011144 storage such as hard drives (hint: use tmpfs and don't swap those files).
11145 Lifetime hint can be changed using tune.ssl.timeout.
11146
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011147transparent
11148 Is an optional keyword which is supported only on certain Linux kernels. It
11149 indicates that the addresses will be bound even if they do not belong to the
11150 local machine, and that packets targeting any of these addresses will be
11151 intercepted just as if the addresses were locally configured. This normally
11152 requires that IP forwarding is enabled. Caution! do not use this with the
11153 default address '*', as it would redirect any traffic for the specified port.
11154 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11155 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11156 kernel version. Some distribution kernels include backports of the feature,
11157 so check for support with your vendor.
11158
Willy Tarreau77e3af92012-11-24 15:07:23 +010011159v4v6
11160 Is an optional keyword which is supported only on most recent systems
11161 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11162 and IPv6 when it uses the default address. Doing so is sometimes necessary
11163 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011164 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011165
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011166v6only
11167 Is an optional keyword which is supported only on most recent systems
11168 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11169 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011170 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11171 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011172
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011173uid <uid>
11174 Sets the owner of the UNIX sockets to the designated system uid. It can also
11175 be set by default in the global section's "unix-bind" statement. Note that
11176 some platforms simply ignore this. This setting is equivalent to the "user"
11177 setting except that the user numeric ID is used instead of its name. This
11178 setting is ignored by non UNIX sockets.
11179
11180user <user>
11181 Sets the owner of the UNIX sockets to the designated system user. It can also
11182 be set by default in the global section's "unix-bind" statement. Note that
11183 some platforms simply ignore this. This setting is equivalent to the "uid"
11184 setting except that the user name is used instead of its uid. This setting is
11185 ignored by non UNIX sockets.
11186
Emeric Brun1a073b42012-09-28 17:07:34 +020011187verify [none|optional|required]
11188 This setting is only available when support for OpenSSL was built in. If set
11189 to 'none', client certificate is not requested. This is the default. In other
11190 cases, a client certificate is requested. If the client does not provide a
11191 certificate after the request and if 'verify' is set to 'required', then the
11192 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011193 certificate provided by the client is always verified using CAs from
11194 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11195 is aborted, regardless of the 'verify' option, unless the error code exactly
11196 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011197
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200111985.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011199------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011200
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011201The "server" and "default-server" keywords support a certain number of settings
11202which are all passed as arguments on the server line. The order in which those
11203arguments appear does not count, and they are all optional. Some of those
11204settings are single words (booleans) while others expect one or several values
11205after them. In this case, the values must immediately follow the setting name.
11206Except default-server, all those settings must be specified after the server's
11207address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011208
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011209 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011210 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011211
Frédéric Lécailled2376272017-03-21 18:52:12 +010011212Note that all these settings are supported both by "server" and "default-server"
11213keywords, except "id" which is only supported by "server".
11214
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011215The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011216
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011217addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011218 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011219 to send health-checks or to probe the agent-check. On some servers, it may be
11220 desirable to dedicate an IP address to specific component able to perform
11221 complex tests which are more suitable to health-checks than the application.
11222 This parameter is ignored if the "check" parameter is not set. See also the
11223 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011224
Simon Hormand60d6912013-11-25 10:46:36 +090011225agent-check
11226 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011227 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010011228 to the port set by the "agent-port" parameter and reading an ASCII string
11229 terminated by the first '\r' or '\n' met. The string is made of a series of
11230 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011231
Willy Tarreau81f5d942013-12-09 20:51:51 +010011232 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011233 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011234 weight of a server as configured when haproxy starts. Note that a zero
11235 weight is reported on the stats page as "DRAIN" since it has the same
11236 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011237
Davor Ocelice9ed2812017-12-25 17:49:28 +010011238 - The string "maxconn:" followed by an integer (no space between). Values
11239 in this format will set the maxconn of a server. The maximum number of
11240 connections advertised needs to be multiplied by the number of load
11241 balancers and different backends that use this health check to get the
11242 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011243
Willy Tarreau81f5d942013-12-09 20:51:51 +010011244 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011245 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011246
Willy Tarreau81f5d942013-12-09 20:51:51 +010011247 - The word "drain". This will turn the server's administrative state to the
11248 DRAIN mode, thus it will not accept any new connections other than those
11249 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011250
Willy Tarreau81f5d942013-12-09 20:51:51 +010011251 - The word "maint". This will turn the server's administrative state to the
11252 MAINT mode, thus it will not accept any new connections at all, and health
11253 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011254
Willy Tarreau81f5d942013-12-09 20:51:51 +010011255 - The words "down", "failed", or "stopped", optionally followed by a
11256 description string after a sharp ('#'). All of these mark the server's
11257 operating state as DOWN, but since the word itself is reported on the stats
11258 page, the difference allows an administrator to know if the situation was
11259 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011260 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011261 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011262
Willy Tarreau81f5d942013-12-09 20:51:51 +010011263 - The word "up" sets back the server's operating state as UP if health checks
11264 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011265
Willy Tarreau81f5d942013-12-09 20:51:51 +010011266 Parameters which are not advertised by the agent are not changed. For
11267 example, an agent might be designed to monitor CPU usage and only report a
11268 relative weight and never interact with the operating status. Similarly, an
11269 agent could be designed as an end-user interface with 3 radio buttons
11270 allowing an administrator to change only the administrative state. However,
11271 it is important to consider that only the agent may revert its own actions,
11272 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11273 agent must implement the other equivalent actions to bring the service into
11274 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011275
Simon Horman2f1f9552013-11-25 10:46:37 +090011276 Failure to connect to the agent is not considered an error as connectivity
11277 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011278 parameter. Warning though, it is not a good idea to stop an agent after it
11279 reports "down", since only an agent reporting "up" will be able to turn the
11280 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011281 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011282
Willy Tarreau81f5d942013-12-09 20:51:51 +010011283 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011284 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011285
James Brown55f9ff12015-10-21 18:19:05 -070011286agent-send <string>
11287 If this option is specified, haproxy will send the given string (verbatim)
11288 to the agent server upon connection. You could, for example, encode
11289 the backend name into this string, which would enable your agent to send
11290 different responses based on the backend. Make sure to include a '\n' if
11291 you want to terminate your request with a newline.
11292
Simon Hormand60d6912013-11-25 10:46:36 +090011293agent-inter <delay>
11294 The "agent-inter" parameter sets the interval between two agent checks
11295 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11296
11297 Just as with every other time-based parameter, it may be entered in any
11298 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11299 parameter also serves as a timeout for agent checks "timeout check" is
11300 not set. In order to reduce "resonance" effects when multiple servers are
11301 hosted on the same hardware, the agent and health checks of all servers
11302 are started with a small time offset between them. It is also possible to
11303 add some random noise in the agent and health checks interval using the
11304 global "spread-checks" keyword. This makes sense for instance when a lot
11305 of backends use the same servers.
11306
11307 See also the "agent-check" and "agent-port" parameters.
11308
Misiek768d8602017-01-09 09:52:43 +010011309agent-addr <addr>
11310 The "agent-addr" parameter sets address for agent check.
11311
11312 You can offload agent-check to another target, so you can make single place
11313 managing status and weights of servers defined in haproxy in case you can't
11314 make self-aware and self-managing services. You can specify both IP or
11315 hostname, it will be resolved.
11316
Simon Hormand60d6912013-11-25 10:46:36 +090011317agent-port <port>
11318 The "agent-port" parameter sets the TCP port used for agent checks.
11319
11320 See also the "agent-check" and "agent-inter" parameters.
11321
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020011322allow-0rtt
11323 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020011324 Note that early data will be sent only if the client used early data, or
11325 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020011326
Olivier Houchardc7566002018-11-20 23:33:50 +010011327alpn <protocols>
11328 This enables the TLS ALPN extension and advertises the specified protocol
11329 list as supported on top of ALPN. The protocol list consists in a comma-
11330 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050011331 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010011332 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
11333 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
11334 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11335 now obsolete NPN extension.
11336 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
11337 be advertised, in order of preference, like below :
11338
11339 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
11340
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011341backup
11342 When "backup" is present on a server line, the server is only used in load
11343 balancing when all other non-backup servers are unavailable. Requests coming
11344 with a persistence cookie referencing the server will always be served
11345 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011346 the "allbackups" option is set in the backend. See also the "no-backup" and
11347 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011348
Emeric Brunef42d922012-10-11 16:11:36 +020011349ca-file <cafile>
11350 This setting is only available when support for OpenSSL was built in. It
11351 designates a PEM file from which to load CA certificates used to verify
11352 server's certificate.
11353
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011354check
11355 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011356 always considered available. If "check" is set, the server is available when
11357 accepting periodic TCP connections, to ensure that it is really able to serve
11358 requests. The default address and port to send the tests to are those of the
11359 server, and the default source is the same as the one defined in the
11360 backend. It is possible to change the address using the "addr" parameter, the
11361 port using the "port" parameter, the source address using the "source"
11362 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011363 parameters. The request method is define in the backend using the "httpchk",
11364 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011365 refer to those options and parameters for more information. See also
11366 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011367
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011368check-send-proxy
11369 This option forces emission of a PROXY protocol line with outgoing health
11370 checks, regardless of whether the server uses send-proxy or not for the
11371 normal traffic. By default, the PROXY protocol is enabled for health checks
11372 if it is already enabled for normal traffic and if no "port" nor "addr"
11373 directive is present. However, if such a directive is present, the
11374 "check-send-proxy" option needs to be used to force the use of the
11375 protocol. See also the "send-proxy" option for more information.
11376
Olivier Houchard92150142018-12-21 19:47:01 +010011377check-alpn <protocols>
11378 Defines which protocols to advertise with ALPN. The protocol list consists in
11379 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
11380 (without quotes). If it is not set, the server ALPN is used.
11381
Jérôme Magninae9bb762018-12-09 16:08:26 +010011382check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020011383 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010011384 over SSL. It is only possible to use a string to set <sni>. If you want to
11385 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020011386
Willy Tarreau763a95b2012-10-04 23:15:39 +020011387check-ssl
11388 This option forces encryption of all health checks over SSL, regardless of
11389 whether the server uses SSL or not for the normal traffic. This is generally
11390 used when an explicit "port" or "addr" directive is specified and SSL health
11391 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011392 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011393 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11394 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011395 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011396 See the "ssl" option for more information and "no-check-ssl" to disable
11397 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011398
Alexander Liu2a54bb72019-05-22 19:44:48 +080011399check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050011400 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080011401 default, the health checks won't go through socks tunnel even it was enabled
11402 for normal traffic.
11403
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011404ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011405 This setting is only available when support for OpenSSL was built in. This
11406 option sets the string describing the list of cipher algorithms that is
11407 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011408 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
11409 information and recommendations see e.g.
11410 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
11411 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
11412 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011413
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011414ciphersuites <ciphersuites>
11415 This setting is only available when support for OpenSSL was built in and
11416 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11417 describing the list of cipher algorithms that is negotiated during the TLS
11418 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011419 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
11420 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
11421 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011422
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011423cookie <value>
11424 The "cookie" parameter sets the cookie value assigned to the server to
11425 <value>. This value will be checked in incoming requests, and the first
11426 operational server possessing the same value will be selected. In return, in
11427 cookie insertion or rewrite modes, this value will be assigned to the cookie
11428 sent to the client. There is nothing wrong in having several servers sharing
11429 the same cookie value, and it is in fact somewhat common between normal and
11430 backup servers. See also the "cookie" keyword in backend section.
11431
Emeric Brunef42d922012-10-11 16:11:36 +020011432crl-file <crlfile>
11433 This setting is only available when support for OpenSSL was built in. It
11434 designates a PEM file from which to load certificate revocation list used
11435 to verify server's certificate.
11436
Emeric Bruna7aa3092012-10-26 12:58:00 +020011437crt <cert>
11438 This setting is only available when support for OpenSSL was built in.
11439 It designates a PEM file from which to load both a certificate and the
11440 associated private key. This file can be built by concatenating both PEM
11441 files into one. This certificate will be sent if the server send a client
11442 certificate request.
11443
Willy Tarreau96839092010-03-29 10:02:24 +020011444disabled
11445 The "disabled" keyword starts the server in the "disabled" state. That means
11446 that it is marked down in maintenance mode, and no connection other than the
11447 ones allowed by persist mode will reach it. It is very well suited to setup
11448 new servers, because normal traffic will never reach them, while it is still
11449 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011450 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011451
Frédéric Lécailled2376272017-03-21 18:52:12 +010011452enabled
11453 This option may be used as 'server' setting to reset any 'disabled'
11454 setting which would have been inherited from 'default-server' directive as
11455 default value.
11456 It may also be used as 'default-server' setting to reset any previous
11457 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011458
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011459error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011460 If health observing is enabled, the "error-limit" parameter specifies the
11461 number of consecutive errors that triggers event selected by the "on-error"
11462 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011463
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011464 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011465
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011466fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011467 The "fall" parameter states that a server will be considered as dead after
11468 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11469 unspecified. See also the "check", "inter" and "rise" parameters.
11470
Emeric Brun8694b9a2012-10-05 14:39:07 +020011471force-sslv3
11472 This option enforces use of SSLv3 only when SSL is used to communicate with
11473 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011474 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011475 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011476
11477force-tlsv10
11478 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011479 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011480 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011481
11482force-tlsv11
11483 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011484 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011485 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011486
11487force-tlsv12
11488 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011489 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011490 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011491
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011492force-tlsv13
11493 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11494 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011495 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011496
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011497id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011498 Set a persistent ID for the server. This ID must be positive and unique for
11499 the proxy. An unused ID will automatically be assigned if unset. The first
11500 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011501
Willy Tarreau6a031d12016-11-07 19:42:35 +010011502init-addr {last | libc | none | <ip>},[...]*
11503 Indicate in what order the server's address should be resolved upon startup
11504 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011505 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011506 method which succeeds is used. If the end of the list is reached without
11507 finding a working method, an error is thrown. Method "last" suggests to pick
11508 the address which appears in the state file (see "server-state-file"). Method
11509 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11510 depending on the operating system and build options). Method "none"
11511 specifically indicates that the server should start without any valid IP
11512 address in a down state. It can be useful to ignore some DNS issues upon
11513 startup, waiting for the situation to get fixed later. Finally, an IP address
11514 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011515 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011516 server used to catch old sessions and present them with a decent error
11517 message for example. When the "first" load balancing algorithm is used, this
11518 IP address could point to a fake server used to trigger the creation of new
11519 instances on the fly. This option defaults to "last,libc" indicating that the
11520 previous address found in the state file (if any) is used first, otherwise
11521 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011522 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011523
11524 Example:
11525 defaults
11526 # never fail on address resolution
11527 default-server init-addr last,libc,none
11528
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011529inter <delay>
11530fastinter <delay>
11531downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011532 The "inter" parameter sets the interval between two consecutive health checks
11533 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11534 It is also possible to use "fastinter" and "downinter" to optimize delays
11535 between checks depending on the server state :
11536
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011537 Server state | Interval used
11538 ----------------------------------------+----------------------------------
11539 UP 100% (non-transitional) | "inter"
11540 ----------------------------------------+----------------------------------
11541 Transitionally UP (going down "fall"), | "fastinter" if set,
11542 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11543 or yet unchecked. |
11544 ----------------------------------------+----------------------------------
11545 DOWN 100% (non-transitional) | "downinter" if set,
11546 | "inter" otherwise.
11547 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011548
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011549 Just as with every other time-based parameter, they can be entered in any
11550 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11551 serves as a timeout for health checks sent to servers if "timeout check" is
11552 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011553 hosted on the same hardware, the agent and health checks of all servers
11554 are started with a small time offset between them. It is also possible to
11555 add some random noise in the agent and health checks interval using the
11556 global "spread-checks" keyword. This makes sense for instance when a lot
11557 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011558
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011559maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011560 The "maxconn" parameter specifies the maximal number of concurrent
11561 connections that will be sent to this server. If the number of incoming
11562 concurrent requests goes higher than this value, they will be queued, waiting
11563 for a connection to be released. This parameter is very important as it can
11564 save fragile servers from going down under extreme loads. If a "minconn"
11565 parameter is specified, the limit becomes dynamic. The default value is "0"
11566 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11567 the backend's "fullconn" keyword.
11568
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011569maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011570 The "maxqueue" parameter specifies the maximal number of connections which
11571 will wait in the queue for this server. If this limit is reached, next
11572 requests will be redispatched to other servers instead of indefinitely
11573 waiting to be served. This will break persistence but may allow people to
11574 quickly re-log in when the server they try to connect to is dying. The
11575 default value is "0" which means the queue is unlimited. See also the
11576 "maxconn" and "minconn" parameters.
11577
Willy Tarreau9c538e02019-01-23 10:21:49 +010011578max-reuse <count>
11579 The "max-reuse" argument indicates the HTTP connection processors that they
11580 should not reuse a server connection more than this number of times to send
11581 new requests. Permitted values are -1 (the default), which disables this
11582 limit, or any positive value. Value zero will effectively disable keep-alive.
11583 This is only used to work around certain server bugs which cause them to leak
11584 resources over time. The argument is not necessarily respected by the lower
11585 layers as there might be technical limitations making it impossible to
11586 enforce. At least HTTP/2 connections to servers will respect it.
11587
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011588minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011589 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11590 limit following the backend's load. The server will always accept at least
11591 <minconn> connections, never more than <maxconn>, and the limit will be on
11592 the ramp between both values when the backend has less than <fullconn>
11593 concurrent connections. This makes it possible to limit the load on the
11594 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011595 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011596 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011597
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011598namespace <name>
11599 On Linux, it is possible to specify which network namespace a socket will
11600 belong to. This directive makes it possible to explicitly bind a server to
11601 a namespace different from the default one. Please refer to your operating
11602 system's documentation to find more details about network namespaces.
11603
Frédéric Lécailled2376272017-03-21 18:52:12 +010011604no-agent-check
11605 This option may be used as "server" setting to reset any "agent-check"
11606 setting which would have been inherited from "default-server" directive as
11607 default value.
11608 It may also be used as "default-server" setting to reset any previous
11609 "default-server" "agent-check" setting.
11610
11611no-backup
11612 This option may be used as "server" setting to reset any "backup"
11613 setting which would have been inherited from "default-server" directive as
11614 default value.
11615 It may also be used as "default-server" setting to reset any previous
11616 "default-server" "backup" setting.
11617
11618no-check
11619 This option may be used as "server" setting to reset any "check"
11620 setting which would have been inherited from "default-server" directive as
11621 default value.
11622 It may also be used as "default-server" setting to reset any previous
11623 "default-server" "check" setting.
11624
11625no-check-ssl
11626 This option may be used as "server" setting to reset any "check-ssl"
11627 setting which would have been inherited from "default-server" directive as
11628 default value.
11629 It may also be used as "default-server" setting to reset any previous
11630 "default-server" "check-ssl" setting.
11631
Frédéric Lécailled2376272017-03-21 18:52:12 +010011632no-send-proxy
11633 This option may be used as "server" setting to reset any "send-proxy"
11634 setting which would have been inherited from "default-server" directive as
11635 default value.
11636 It may also be used as "default-server" setting to reset any previous
11637 "default-server" "send-proxy" setting.
11638
11639no-send-proxy-v2
11640 This option may be used as "server" setting to reset any "send-proxy-v2"
11641 setting which would have been inherited from "default-server" directive as
11642 default value.
11643 It may also be used as "default-server" setting to reset any previous
11644 "default-server" "send-proxy-v2" setting.
11645
11646no-send-proxy-v2-ssl
11647 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11648 setting which would have been inherited from "default-server" directive as
11649 default value.
11650 It may also be used as "default-server" setting to reset any previous
11651 "default-server" "send-proxy-v2-ssl" setting.
11652
11653no-send-proxy-v2-ssl-cn
11654 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11655 setting which would have been inherited from "default-server" directive as
11656 default value.
11657 It may also be used as "default-server" setting to reset any previous
11658 "default-server" "send-proxy-v2-ssl-cn" setting.
11659
11660no-ssl
11661 This option may be used as "server" setting to reset any "ssl"
11662 setting which would have been inherited from "default-server" directive as
11663 default value.
11664 It may also be used as "default-server" setting to reset any previous
11665 "default-server" "ssl" setting.
11666
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011667no-ssl-reuse
11668 This option disables SSL session reuse when SSL is used to communicate with
11669 the server. It will force the server to perform a full handshake for every
11670 new connection. It's probably only useful for benchmarking, troubleshooting,
11671 and for paranoid users.
11672
Emeric Brun9b3009b2012-10-05 11:55:06 +020011673no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011674 This option disables support for SSLv3 when SSL is used to communicate with
11675 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011676 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011677
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011678 Supported in default-server: No
11679
Emeric Brunf9c5c472012-10-11 15:28:34 +020011680no-tls-tickets
11681 This setting is only available when support for OpenSSL was built in. It
11682 disables the stateless session resumption (RFC 5077 TLS Ticket
11683 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011684 session resumption is more expensive in CPU usage for servers. This option
11685 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011686 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011687
Emeric Brun9b3009b2012-10-05 11:55:06 +020011688no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011689 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011690 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11691 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011692 often makes sense to disable it when communicating with local servers. This
11693 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011694 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011695
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011696 Supported in default-server: No
11697
Emeric Brun9b3009b2012-10-05 11:55:06 +020011698no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011699 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011700 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11701 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011702 often makes sense to disable it when communicating with local servers. This
11703 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011704 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011705
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011706 Supported in default-server: No
11707
Emeric Brun9b3009b2012-10-05 11:55:06 +020011708no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011709 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011710 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11711 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011712 often makes sense to disable it when communicating with local servers. This
11713 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011714 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011715
11716 Supported in default-server: No
11717
11718no-tlsv13
11719 This option disables support for TLSv1.3 when SSL is used to communicate with
11720 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11721 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11722 often makes sense to disable it when communicating with local servers. This
11723 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011724 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011725
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011726 Supported in default-server: No
11727
Frédéric Lécailled2376272017-03-21 18:52:12 +010011728no-verifyhost
11729 This option may be used as "server" setting to reset any "verifyhost"
11730 setting which would have been inherited from "default-server" directive as
11731 default value.
11732 It may also be used as "default-server" setting to reset any previous
11733 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011734
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020011735no-tfo
11736 This option may be used as "server" setting to reset any "tfo"
11737 setting which would have been inherited from "default-server" directive as
11738 default value.
11739 It may also be used as "default-server" setting to reset any previous
11740 "default-server" "tfo" setting.
11741
Simon Hormanfa461682011-06-25 09:39:49 +090011742non-stick
11743 Never add connections allocated to this sever to a stick-table.
11744 This may be used in conjunction with backup to ensure that
11745 stick-table persistence is disabled for backup servers.
11746
Olivier Houchardc7566002018-11-20 23:33:50 +010011747npn <protocols>
11748 This enables the NPN TLS extension and advertises the specified protocol list
11749 as supported on top of NPN. The protocol list consists in a comma-delimited
11750 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050011751 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010011752 enabled (check with haproxy -vv). Note that the NPN extension has been
11753 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11754 only available starting with OpenSSL 1.0.2.
11755
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011756observe <mode>
11757 This option enables health adjusting based on observing communication with
11758 the server. By default this functionality is disabled and enabling it also
11759 requires to enable health checks. There are two supported modes: "layer4" and
11760 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11761 significant. In layer7, which is only allowed for http proxies, responses
11762 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011763 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011764
11765 See also the "check", "on-error" and "error-limit".
11766
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011767on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011768 Select what should happen when enough consecutive errors are detected.
11769 Currently, four modes are available:
11770 - fastinter: force fastinter
11771 - fail-check: simulate a failed check, also forces fastinter (default)
11772 - sudden-death: simulate a pre-fatal failed health check, one more failed
11773 check will mark a server down, forces fastinter
11774 - mark-down: mark the server immediately down and force fastinter
11775
11776 See also the "check", "observe" and "error-limit".
11777
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011778on-marked-down <action>
11779 Modify what occurs when a server is marked down.
11780 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011781 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11782 all connections to the server are immediately terminated when the server
11783 goes down. It might be used if the health check detects more complex cases
11784 than a simple connection status, and long timeouts would cause the service
11785 to remain unresponsive for too long a time. For instance, a health check
11786 might detect that a database is stuck and that there's no chance to reuse
11787 existing connections anymore. Connections killed this way are logged with
11788 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011789
11790 Actions are disabled by default
11791
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011792on-marked-up <action>
11793 Modify what occurs when a server is marked up.
11794 Currently one action is available:
11795 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11796 done only if the server is not in backup state and if it is not disabled
11797 (it must have an effective weight > 0). This can be used sometimes to force
11798 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011799 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11800 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011801 with extreme care. Sessions killed because a server comes up are logged
11802 with an 'U' termination code (for "Up").
11803
11804 Actions are disabled by default
11805
Olivier Houchard006e3102018-12-10 18:30:32 +010011806pool-max-conn <max>
11807 Set the maximum number of idling connections for a server. -1 means unlimited
11808 connections, 0 means no idle connections. The default is -1. When idle
11809 connections are enabled, orphaned idle connections which do not belong to any
11810 client session anymore are moved to a dedicated pool so that they remain
11811 usable by future clients. This only applies to connections that can be shared
11812 according to the same principles as those applying to "http-reuse".
11813
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010011814pool-purge-delay <delay>
11815 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010011816 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020011817 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010011818
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011819port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011820 Using the "port" parameter, it becomes possible to use a different port to
11821 send health-checks. On some servers, it may be desirable to dedicate a port
11822 to a specific component able to perform complex tests which are more suitable
11823 to health-checks than the application. It is common to run a simple script in
11824 inetd for instance. This parameter is ignored if the "check" parameter is not
11825 set. See also the "addr" parameter.
11826
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020011827proto <name>
11828
11829 Forces the multiplexer's protocol to use for the outgoing connections to this
11830 server. It must be compatible with the mode of the backend (TCP or HTTP). It
11831 must also be usable on the backend side. The list of available protocols is
11832 reported in haproxy -vv.
11833 Idea behind this optipon is to bypass the selection of the best multiplexer's
11834 protocol for all connections established to this server.
11835
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011836redir <prefix>
11837 The "redir" parameter enables the redirection mode for all GET and HEAD
11838 requests addressing this server. This means that instead of having HAProxy
11839 forward the request to the server, it will send an "HTTP 302" response with
11840 the "Location" header composed of this prefix immediately followed by the
11841 requested URI beginning at the leading '/' of the path component. That means
11842 that no trailing slash should be used after <prefix>. All invalid requests
11843 will be rejected, and all non-GET or HEAD requests will be normally served by
11844 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011845 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011846 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011847 users to a remote location in case of local disaster. Main use consists in
11848 increasing bandwidth for static servers by having the clients directly
11849 connect to them. Note: never use a relative location here, it would cause a
11850 loop between the client and HAProxy!
11851
11852 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
11853
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011854rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011855 The "rise" parameter states that a server will be considered as operational
11856 after <count> consecutive successful health checks. This value defaults to 2
11857 if unspecified. See also the "check", "inter" and "fall" parameters.
11858
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020011859resolve-opts <option>,<option>,...
11860 Comma separated list of options to apply to DNS resolution linked to this
11861 server.
11862
11863 Available options:
11864
11865 * allow-dup-ip
11866 By default, HAProxy prevents IP address duplication in a backend when DNS
11867 resolution at runtime is in operation.
11868 That said, for some cases, it makes sense that two servers (in the same
11869 backend, being resolved by the same FQDN) have the same IP address.
11870 For such case, simply enable this option.
11871 This is the opposite of prevent-dup-ip.
11872
11873 * prevent-dup-ip
11874 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
11875 an IP address already set to a server in the same backend and sharing the
11876 same fqdn.
11877 This is the opposite of allow-dup-ip.
11878
11879 Example:
11880 backend b_myapp
11881 default-server init-addr none resolvers dns
11882 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
11883 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
11884
11885 With the option allow-dup-ip set:
11886 * if the nameserver returns a single IP address, then both servers will use
11887 it
11888 * If the nameserver returns 2 IP addresses, then each server will pick up a
11889 different address
11890
11891 Default value: not set
11892
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011893resolve-prefer <family>
11894 When DNS resolution is enabled for a server and multiple IP addresses from
11895 different families are returned, HAProxy will prefer using an IP address
11896 from the family mentioned in the "resolve-prefer" parameter.
11897 Available families: "ipv4" and "ipv6"
11898
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020011899 Default value: ipv6
11900
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011901 Example:
11902
11903 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011904
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011905resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050011906 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011907 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010011908 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010011909 different datacenters. The latency between datacenter is not negligible, so
11910 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011911 configured network, another address is selected.
11912
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011913 Example:
11914
11915 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011916
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011917resolvers <id>
11918 Points to an existing "resolvers" section to resolve current server's
11919 hostname.
11920
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011921 Example:
11922
11923 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011924
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011925 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011926
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011927send-proxy
11928 The "send-proxy" parameter enforces use of the PROXY protocol over any
11929 connection established to this server. The PROXY protocol informs the other
11930 end about the layer 3/4 addresses of the incoming connection, so that it can
11931 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011932 upper layer protocol. For connections accepted by an "accept-proxy" or
11933 "accept-netscaler-cip" listener, the advertised address will be used. Only
11934 TCPv4 and TCPv6 address families are supported. Other families such as
11935 Unix sockets, will report an UNKNOWN family. Servers using this option can
11936 fully be chained to another instance of haproxy listening with an
11937 "accept-proxy" setting. This setting must not be used if the server isn't
11938 aware of the protocol. When health checks are sent to the server, the PROXY
11939 protocol is automatically used when this option is set, unless there is an
11940 explicit "port" or "addr" directive, in which case an explicit
11941 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011942 See also the "no-send-proxy" option of this section and "accept-proxy" and
11943 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011944
David Safb76832014-05-08 23:42:08 -040011945send-proxy-v2
11946 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
11947 over any connection established to this server. The PROXY protocol informs
11948 the other end about the layer 3/4 addresses of the incoming connection, so
11949 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020011950 whatever the upper layer protocol. It also send ALPN information if an alpn
11951 have been negotiated. This setting must not be used if the server isn't aware
11952 of this version of the protocol. See also the "no-send-proxy-v2" option of
11953 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011954
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010011955proxy-v2-options <option>[,<option>]*
11956 The "proxy-v2-options" parameter add option to send in PROXY protocol version
11957 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010011958 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
11959 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010011960 certificate, "cert-key": key algorithm of the used certificate), "authority":
11961 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010011962 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010011963
David Safb76832014-05-08 23:42:08 -040011964send-proxy-v2-ssl
11965 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11966 2 over any connection established to this server. The PROXY protocol informs
11967 the other end about the layer 3/4 addresses of the incoming connection, so
11968 that it can know the client's address or the public address it accessed to,
11969 whatever the upper layer protocol. In addition, the SSL information extension
11970 of the PROXY protocol is added to the PROXY protocol header. This setting
11971 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011972 See also the "no-send-proxy-v2-ssl" option of this section and the
11973 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011974
11975send-proxy-v2-ssl-cn
11976 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11977 2 over any connection established to this server. The PROXY protocol informs
11978 the other end about the layer 3/4 addresses of the incoming connection, so
11979 that it can know the client's address or the public address it accessed to,
11980 whatever the upper layer protocol. In addition, the SSL information extension
11981 of the PROXY protocol, along along with the Common Name from the subject of
11982 the client certificate (if any), is added to the PROXY protocol header. This
11983 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011984 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
11985 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011986
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011987slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011988 The "slowstart" parameter for a server accepts a value in milliseconds which
11989 indicates after how long a server which has just come back up will run at
11990 full speed. Just as with every other time-based parameter, it can be entered
11991 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
11992 linearly from 0 to 100% during this time. The limitation applies to two
11993 parameters :
11994
11995 - maxconn: the number of connections accepted by the server will grow from 1
11996 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
11997
11998 - weight: when the backend uses a dynamic weighted algorithm, the weight
11999 grows linearly from 1 to 100%. In this case, the weight is updated at every
12000 health-check. For this reason, it is important that the "inter" parameter
12001 is smaller than the "slowstart", in order to maximize the number of steps.
12002
12003 The slowstart never applies when haproxy starts, otherwise it would cause
12004 trouble to running servers. It only applies when a server has been previously
12005 seen as failed.
12006
Willy Tarreau732eac42015-07-09 11:40:25 +020012007sni <expression>
12008 The "sni" parameter evaluates the sample fetch expression, converts it to a
12009 string and uses the result as the host name sent in the SNI TLS extension to
12010 the server. A typical use case is to send the SNI received from the client in
12011 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012012 expression, though alternatives such as req.hdr(host) can also make sense. If
12013 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012014 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012015 "verify" directive for more details. If you want to set a SNI for health
12016 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012017
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012018source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012019source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012020source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012021 The "source" parameter sets the source address which will be used when
12022 connecting to the server. It follows the exact same parameters and principle
12023 as the backend "source" keyword, except that it only applies to the server
12024 referencing it. Please consult the "source" keyword for details.
12025
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012026 Additionally, the "source" statement on a server line allows one to specify a
12027 source port range by indicating the lower and higher bounds delimited by a
12028 dash ('-'). Some operating systems might require a valid IP address when a
12029 source port range is specified. It is permitted to have the same IP/range for
12030 several servers. Doing so makes it possible to bypass the maximum of 64k
12031 total concurrent connections. The limit will then reach 64k connections per
12032 server.
12033
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012034 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12035 specifying the source address without port(s).
12036
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012037ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012038 This option enables SSL ciphering on outgoing connections to the server. It
12039 is critical to verify server certificates using "verify" when using SSL to
12040 connect to servers, otherwise the communication is prone to trivial man in
12041 the-middle attacks rendering SSL useless. When this option is used, health
12042 checks are automatically sent in SSL too unless there is a "port" or an
12043 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012044 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12045 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012046
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012047ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12048 This option enforces use of <version> or lower when SSL is used to communicate
12049 with the server. This option is also available on global statement
12050 "ssl-default-server-options". See also "ssl-min-ver".
12051
12052ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12053 This option enforces use of <version> or upper when SSL is used to communicate
12054 with the server. This option is also available on global statement
12055 "ssl-default-server-options". See also "ssl-max-ver".
12056
Frédéric Lécailled2376272017-03-21 18:52:12 +010012057ssl-reuse
12058 This option may be used as "server" setting to reset any "no-ssl-reuse"
12059 setting which would have been inherited from "default-server" directive as
12060 default value.
12061 It may also be used as "default-server" setting to reset any previous
12062 "default-server" "no-ssl-reuse" setting.
12063
12064stick
12065 This option may be used as "server" setting to reset any "non-stick"
12066 setting which would have been inherited from "default-server" directive as
12067 default value.
12068 It may also be used as "default-server" setting to reset any previous
12069 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012070
Alexander Liu2a54bb72019-05-22 19:44:48 +080012071socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050012072 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080012073 server. Using this option won't force the health check to go via socks4 by
12074 default. You will have to use the keyword "check-via-socks4" to enable it.
12075
Willy Tarreau163d4622015-10-13 16:16:41 +020012076tcp-ut <delay>
12077 Sets the TCP User Timeout for all outgoing connections to this server. This
12078 option is available on Linux since version 2.6.37. It allows haproxy to
12079 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012080 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012081 long-lived connections experiencing long idle periods such as remote
12082 terminals or database connection pools, where the client and server timeouts
12083 must remain high to allow a long period of idle, but where it is important to
12084 detect that the server has disappeared in order to release all resources
12085 associated with its connection (and the client's session). One typical use
12086 case is also to force dead server connections to die when health checks are
12087 too slow or during a soft reload since health checks are then disabled. The
12088 argument is a delay expressed in milliseconds by default. This only works for
12089 regular TCP connections, and is ignored for other protocols.
12090
Willy Tarreau034c88c2017-01-23 23:36:45 +010012091tfo
12092 This option enables using TCP fast open when connecting to servers, on
12093 systems that support it (currently only the Linux kernel >= 4.11).
12094 See the "tfo" bind option for more information about TCP fast open.
12095 Please note that when using tfo, you should also use the "conn-failure",
12096 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020012097 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010012098
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012099track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012100 This option enables ability to set the current state of the server by tracking
12101 another one. It is possible to track a server which itself tracks another
12102 server, provided that at the end of the chain, a server has health checks
12103 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012104 used, it has to be enabled on both proxies.
12105
Frédéric Lécailled2376272017-03-21 18:52:12 +010012106tls-tickets
12107 This option may be used as "server" setting to reset any "no-tls-tickets"
12108 setting which would have been inherited from "default-server" directive as
12109 default value.
12110 It may also be used as "default-server" setting to reset any previous
12111 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012112
Emeric Brunef42d922012-10-11 16:11:36 +020012113verify [none|required]
12114 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012115 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012116 certificate provided by the server is verified using CAs from 'ca-file' and
12117 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012118 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012119 the name passed using the "sni" directive, or if not provided, the static
12120 host name passed using the "verifyhost" directive. When no name is found, the
12121 certificate's names are ignored. For this reason, without SNI it's important
12122 to use "verifyhost". On verification failure the handshake is aborted. It is
12123 critically important to verify server certificates when using SSL to connect
12124 to servers, otherwise the communication is prone to trivial man-in-the-middle
12125 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12126 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012127
Evan Broderbe554312013-06-27 00:05:25 -070012128verifyhost <hostname>
12129 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012130 only takes effect if 'verify required' is also specified. This directive sets
12131 a default static hostname to check the server's certificate against when no
12132 SNI was used to connect to the server. If SNI is not used, this is the only
12133 way to enable hostname verification. This static hostname, when set, will
12134 also be used for health checks (which cannot provide an SNI value). If none
12135 of the hostnames in the certificate match the specified hostname, the
12136 handshake is aborted. The hostnames in the server-provided certificate may
12137 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012138
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012139weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012140 The "weight" parameter is used to adjust the server's weight relative to
12141 other servers. All servers will receive a load proportional to their weight
12142 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012143 load. The default weight is 1, and the maximal value is 256. A value of 0
12144 means the server will not participate in load-balancing but will still accept
12145 persistent connections. If this parameter is used to distribute the load
12146 according to server's capacity, it is recommended to start with values which
12147 can both grow and shrink, for instance between 10 and 100 to leave enough
12148 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012149
12150
Cyril Bonté46175dd2015-07-02 22:45:32 +0200121515.3. Server IP address resolution using DNS
12152-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012153
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012154HAProxy allows using a host name on the server line to retrieve its IP address
12155using name servers. By default, HAProxy resolves the name when parsing the
12156configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012157This is not sufficient in some cases, such as in Amazon where a server's IP
12158can change after a reboot or an ELB Virtual IP can change based on current
12159workload.
12160This chapter describes how HAProxy can be configured to process server's name
12161resolution at run time.
12162Whether run time server name resolution has been enable or not, HAProxy will
12163carry on doing the first resolution when parsing the configuration.
12164
12165
Cyril Bonté46175dd2015-07-02 22:45:32 +0200121665.3.1. Global overview
12167----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012168
12169As we've seen in introduction, name resolution in HAProxy occurs at two
12170different steps of the process life:
12171
12172 1. when starting up, HAProxy parses the server line definition and matches a
12173 host name. It uses libc functions to get the host name resolved. This
12174 resolution relies on /etc/resolv.conf file.
12175
Christopher Faulet67957bd2017-09-27 11:00:59 +020012176 2. at run time, HAProxy performs periodically name resolutions for servers
12177 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012178
12179A few other events can trigger a name resolution at run time:
12180 - when a server's health check ends up in a connection timeout: this may be
12181 because the server has a new IP address. So we need to trigger a name
12182 resolution to know this new IP.
12183
Christopher Faulet67957bd2017-09-27 11:00:59 +020012184When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012185HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012186SRV label is specified, then the corresponding SRV records will be retrieved
12187from the DNS server, and the provided hostnames will be used. The SRV label
12188will be checked periodically, and if any server are added or removed, haproxy
12189will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012190
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012191A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050012192 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012193 first valid response.
12194
12195 - a resolution is considered as invalid (NX, timeout, refused), when all the
12196 servers return an error.
12197
12198
Cyril Bonté46175dd2015-07-02 22:45:32 +0200121995.3.2. The resolvers section
12200----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012201
12202This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012203HAProxy. There can be as many as resolvers section as needed. Each section can
12204contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012205
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012206When multiple name servers are configured in a resolvers section, then HAProxy
12207uses the first valid response. In case of invalid responses, only the last one
12208is treated. Purpose is to give the chance to a slow server to deliver a valid
12209answer after a fast faulty or outdated server.
12210
12211When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012212used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012213
Christopher Faulet67957bd2017-09-27 11:00:59 +020012214 1. HAProxy retries the same DNS query with a new query type. The A queries are
12215 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12216 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012217
Christopher Faulet67957bd2017-09-27 11:00:59 +020012218 2. When the fallback on the query type was done (or not applicable), HAProxy
12219 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012220
Christopher Faulet67957bd2017-09-27 11:00:59 +020012221 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12222 response is received after that, it stops the DNS resolution and reports
12223 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012224
Christopher Faulet67957bd2017-09-27 11:00:59 +020012225For example, with 2 name servers configured in a resolvers section, the
12226following scenarios are possible:
12227
12228 - First response is valid and is applied directly, second response is
12229 ignored
12230
12231 - First response is invalid and second one is valid, then second response is
12232 applied
12233
12234 - First response is a NX domain and second one a truncated response, then
12235 HAProxy retries the query with a new type
12236
12237 - First response is a NX domain and second one is a timeout, then HAProxy
12238 retries the query with a new type
12239
12240 - Query timed out for both name servers, then HAProxy retries it with the
12241 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012242
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012243As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12244a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012245<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012246
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012247
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012248resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012249 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012250
12251A resolvers section accept the following parameters:
12252
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012253accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012254 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012255 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012256 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12257 by RFC 6891)
12258
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012259 Note: the maximum allowed value is 8192.
12260
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012261nameserver <id> <ip>:<port>
12262 DNS server description:
12263 <id> : label of the server, should be unique
12264 <ip> : IP address of the server
12265 <port> : port where the DNS service actually runs
12266
Ben Draut44e609b2018-05-29 15:40:08 -060012267parse-resolv-conf
12268 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12269 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12270 placed in the resolvers section in place of this directive.
12271
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012272hold <status> <period>
12273 Defines <period> during which the last name resolution should be kept based
12274 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012275 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012276 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012277 <period> : interval between two successive name resolution when the last
12278 answer was in <status>. It follows the HAProxy time format.
12279 <period> is in milliseconds by default.
12280
Baptiste Assmann686408b2017-08-18 10:15:42 +020012281 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012282
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012283resolve_retries <nb>
12284 Defines the number <nb> of queries to send to resolve a server name before
12285 giving up.
12286 Default value: 3
12287
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012288 A retry occurs on name server timeout or when the full sequence of DNS query
12289 type failover is over and we need to start up from the default ANY query
12290 type.
12291
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012292timeout <event> <time>
12293 Defines timeouts related to name resolution
12294 <event> : the event on which the <time> timeout period applies to.
12295 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012296 - resolve : default time to trigger name resolutions when no
12297 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012298 Default value: 1s
12299 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012300 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012301 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012302 <time> : time related to the event. It follows the HAProxy time format.
12303 <time> is expressed in milliseconds.
12304
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012305 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012306
12307 resolvers mydns
12308 nameserver dns1 10.0.0.1:53
12309 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012310 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012311 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012312 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012313 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012314 hold other 30s
12315 hold refused 30s
12316 hold nx 30s
12317 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012318 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012319 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012320
12321
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200123226. HTTP header manipulation
12323---------------------------
12324
12325In HTTP mode, it is possible to rewrite, add or delete some of the request and
12326response headers based on regular expressions. It is also possible to block a
12327request or a response if a particular header matches a regular expression,
12328which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010012329against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012330
Willy Tarreau70dffda2014-01-30 03:07:23 +010012331If HAProxy encounters an "Informational Response" (status code 1xx), it is able
12332to process all rsp* rules which can allow, deny, rewrite or delete a header,
12333but it will refuse to add a header to any such messages as this is not
12334HTTP-compliant. The reason for still processing headers in such responses is to
12335stop and/or fix any possible information leak which may happen, for instance
12336because another downstream equipment would unconditionally add a header, or if
12337a server name appears there. When such messages are seen, normal processing
12338still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020012339
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012340This section covers common usage of the following keywords, described in detail
12341in section 4.2 :
12342
12343 - reqadd <string>
12344 - reqallow <search>
12345 - reqiallow <search>
12346 - reqdel <search>
12347 - reqidel <search>
12348 - reqdeny <search>
12349 - reqideny <search>
12350 - reqpass <search>
12351 - reqipass <search>
12352 - reqrep <search> <replace>
12353 - reqirep <search> <replace>
12354 - reqtarpit <search>
12355 - reqitarpit <search>
12356 - rspadd <string>
12357 - rspdel <search>
12358 - rspidel <search>
12359 - rspdeny <search>
12360 - rspideny <search>
12361 - rsprep <search> <replace>
12362 - rspirep <search> <replace>
12363
12364With all these keywords, the same conventions are used. The <search> parameter
12365is a POSIX extended regular expression (regex) which supports grouping through
12366parenthesis (without the backslash). Spaces and other delimiters must be
12367prefixed with a backslash ('\') to avoid confusion with a field delimiter.
12368Other characters may be prefixed with a backslash to change their meaning :
12369
12370 \t for a tab
12371 \r for a carriage return (CR)
12372 \n for a new line (LF)
12373 \ to mark a space and differentiate it from a delimiter
12374 \# to mark a sharp and differentiate it from a comment
12375 \\ to use a backslash in a regex
12376 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
12377 \xXX to write the ASCII hex code XX as in the C language
12378
12379The <replace> parameter contains the string to be used to replace the largest
12380portion of text matching the regex. It can make use of the special characters
12381above, and can reference a substring which is delimited by parenthesis in the
12382regex, by writing a backslash ('\') immediately followed by one digit from 0 to
123839 indicating the group position (0 designating the entire line). This practice
12384is very common to users of the "sed" program.
12385
12386The <string> parameter represents the string which will systematically be added
12387after the last header line. It can also use special character sequences above.
12388
12389Notes related to these keywords :
12390---------------------------------
12391 - these keywords are not always convenient to allow/deny based on header
12392 contents. It is strongly recommended to use ACLs with the "block" keyword
12393 instead, resulting in far more flexible and manageable rules.
12394
12395 - lines are always considered as a whole. It is not possible to reference
12396 a header name only or a value only. This is important because of the way
12397 headers are written (notably the number of spaces after the colon).
12398
12399 - the first line is always considered as a header, which makes it possible to
12400 rewrite or filter HTTP requests URIs or response codes, but in turn makes
12401 it harder to distinguish between headers and request line. The regex prefix
12402 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
12403 ^[^ \t:]*: matches any header name followed by a colon.
12404
12405 - for performances reasons, the number of characters added to a request or to
12406 a response is limited at build time to values between 1 and 4 kB. This
12407 should normally be far more than enough for most usages. If it is too short
12408 on occasional usages, it is possible to gain some space by removing some
12409 useless headers before adding new ones.
12410
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012411 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012412 without the 'i' letter except that they ignore case when matching patterns.
12413
12414 - when a request passes through a frontend then a backend, all req* rules
12415 from the frontend will be evaluated, then all req* rules from the backend
12416 will be evaluated. The reverse path is applied to responses.
12417
12418 - req* statements are applied after "block" statements, so that "block" is
12419 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010012420 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012421
12422
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200124236. Cache
12424---------
12425
12426HAProxy provides a cache, which was designed to perform cache on small objects
12427(favicon, css...). This is a minimalist low-maintenance cache which runs in
12428RAM.
12429
12430The cache is based on a memory which is shared between processes and threads,
12431this memory is split in blocks of 1k.
12432
12433If an object is not used anymore, it can be deleted to store a new object
12434independently of its expiration date. The oldest objects are deleted first
12435when we try to allocate a new one.
12436
12437The cache uses a hash of the host header and the URI as the key.
12438
12439It's possible to view the status of a cache using the Unix socket command
12440"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
12441for more details.
12442
12443When an object is delivered from the cache, the server name in the log is
12444replaced by "<CACHE>".
12445
12446
124476.1. Limitation
12448----------------
12449
12450The cache won't store and won't deliver objects in these cases:
12451
12452- If the response is not a 200
12453- If the response contains a Vary header
12454- If the Content-Length + the headers size is greater than "max-object-size"
12455- If the response is not cacheable
12456
12457- If the request is not a GET
12458- If the HTTP version of the request is smaller than 1.1
12459- If the request contains an Authorization header
12460
12461
124626.2. Setup
12463-----------
12464
12465To setup a cache, you must define a cache section and use it in a proxy with
12466the corresponding http-request and response actions.
12467
12468
124696.2.1. Cache section
12470---------------------
12471
12472cache <name>
12473 Declare a cache section, allocate a shared cache memory named <name>, the
12474 size of cache is mandatory.
12475
12476total-max-size <megabytes>
12477 Define the size in RAM of the cache in megabytes. This size is split in
12478 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
12479
12480max-object-size <bytes>
12481 Define the maximum size of the objects to be cached. Must not be greater than
12482 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
12483 All objects with sizes larger than "max-object-size" will not be cached.
12484
12485max-age <seconds>
12486 Define the maximum expiration duration. The expiration is set has the lowest
12487 value between the s-maxage or max-age (in this order) directive in the
12488 Cache-Control response header and this value. The default value is 60
12489 seconds, which means that you can't cache an object more than 60 seconds by
12490 default.
12491
12492
124936.2.2. Proxy section
12494---------------------
12495
12496http-request cache-use <name> [ { if | unless } <condition> ]
12497 Try to deliver a cached object from the cache <name>. This directive is also
12498 mandatory to store the cache as it calculates the cache hash. If you want to
12499 use a condition for both storage and delivering that's a good idea to put it
12500 after this one.
12501
12502http-response cache-store <name> [ { if | unless } <condition> ]
12503 Store an http-response within the cache. The storage of the response headers
12504 is done at this step, which means you can use others http-response actions
12505 to modify headers before or after the storage of the response. This action
12506 is responsible for the setup of the cache storage filter.
12507
12508
12509Example:
12510
12511 backend bck1
12512 mode http
12513
12514 http-request cache-use foobar
12515 http-response cache-store foobar
12516 server srv1 127.0.0.1:80
12517
12518 cache foobar
12519 total-max-size 4
12520 max-age 240
12521
12522
Willy Tarreau74ca5042013-06-11 23:12:07 +0200125237. Using ACLs and fetching samples
12524----------------------------------
12525
Davor Ocelice9ed2812017-12-25 17:49:28 +010012526HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012527client or server information, from tables, environmental information etc...
12528The action of extracting such data is called fetching a sample. Once retrieved,
12529these samples may be used for various purposes such as a key to a stick-table,
12530but most common usages consist in matching them against predefined constant
12531data called patterns.
12532
12533
125347.1. ACL basics
12535---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012536
12537The use of Access Control Lists (ACL) provides a flexible solution to perform
12538content switching and generally to take decisions based on content extracted
12539from the request, the response or any environmental status. The principle is
12540simple :
12541
Willy Tarreau74ca5042013-06-11 23:12:07 +020012542 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012543 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012544 - apply one or multiple pattern matching methods on this sample
12545 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012546
Willy Tarreau74ca5042013-06-11 23:12:07 +020012547The actions generally consist in blocking a request, selecting a backend, or
12548adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012549
12550In order to define a test, the "acl" keyword is used. The syntax is :
12551
Willy Tarreau74ca5042013-06-11 23:12:07 +020012552 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012553
12554This creates a new ACL <aclname> or completes an existing one with new tests.
12555Those tests apply to the portion of request/response specified in <criterion>
12556and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012557an operator which may be specified before the set of values. Optionally some
12558conversion operators may be applied to the sample, and they will be specified
12559as a comma-delimited list of keywords just after the first keyword. The values
12560are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012561
12562ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12563'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12564which means that "my_acl" and "My_Acl" are two different ACLs.
12565
12566There is no enforced limit to the number of ACLs. The unused ones do not affect
12567performance, they just consume a small amount of memory.
12568
Willy Tarreau74ca5042013-06-11 23:12:07 +020012569The criterion generally is the name of a sample fetch method, or one of its ACL
12570specific declinations. The default test method is implied by the output type of
12571this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012572methods of a same sample fetch method. The sample fetch methods are the only
12573ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012574
12575Sample fetch methods return data which can be of the following types :
12576 - boolean
12577 - integer (signed or unsigned)
12578 - IPv4 or IPv6 address
12579 - string
12580 - data block
12581
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012582Converters transform any of these data into any of these. For example, some
12583converters might convert a string to a lower-case string while other ones
12584would turn a string to an IPv4 address, or apply a netmask to an IP address.
12585The resulting sample is of the type of the last converter applied to the list,
12586which defaults to the type of the sample fetch method.
12587
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012588Each sample or converter returns data of a specific type, specified with its
12589keyword in this documentation. When an ACL is declared using a standard sample
12590fetch method, certain types automatically involved a default matching method
12591which are summarized in the table below :
12592
12593 +---------------------+-----------------+
12594 | Sample or converter | Default |
12595 | output type | matching method |
12596 +---------------------+-----------------+
12597 | boolean | bool |
12598 +---------------------+-----------------+
12599 | integer | int |
12600 +---------------------+-----------------+
12601 | ip | ip |
12602 +---------------------+-----------------+
12603 | string | str |
12604 +---------------------+-----------------+
12605 | binary | none, use "-m" |
12606 +---------------------+-----------------+
12607
12608Note that in order to match a binary samples, it is mandatory to specify a
12609matching method, see below.
12610
Willy Tarreau74ca5042013-06-11 23:12:07 +020012611The ACL engine can match these types against patterns of the following types :
12612 - boolean
12613 - integer or integer range
12614 - IP address / network
12615 - string (exact, substring, suffix, prefix, subdir, domain)
12616 - regular expression
12617 - hex block
12618
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012619The following ACL flags are currently supported :
12620
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012621 -i : ignore case during matching of all subsequent patterns.
12622 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012623 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012624 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012625 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012626 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012627 -- : force end of flags. Useful when a string looks like one of the flags.
12628
Willy Tarreau74ca5042013-06-11 23:12:07 +020012629The "-f" flag is followed by the name of a file from which all lines will be
12630read as individual values. It is even possible to pass multiple "-f" arguments
12631if the patterns are to be loaded from multiple files. Empty lines as well as
12632lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12633will be stripped. If it is absolutely necessary to insert a valid pattern
12634beginning with a sharp, just prefix it with a space so that it is not taken for
12635a comment. Depending on the data type and match method, haproxy may load the
12636lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12637exact string matching. In this case, duplicates will automatically be removed.
12638
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012639The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12640parsed as two column file. The first column contains the patterns used by the
12641ACL, and the second column contain the samples. The sample can be used later by
12642a map. This can be useful in some rare cases where an ACL would just be used to
12643check for the existence of a pattern in a map before a mapping is applied.
12644
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012645The "-u" flag forces the unique id of the ACL. This unique id is used with the
12646socket interface to identify ACL and dynamically change its values. Note that a
12647file is always identified by its name even if an id is set.
12648
Willy Tarreau74ca5042013-06-11 23:12:07 +020012649Also, note that the "-i" flag applies to subsequent entries and not to entries
12650loaded from files preceding it. For instance :
12651
12652 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12653
12654In this example, each line of "exact-ua.lst" will be exactly matched against
12655the "user-agent" header of the request. Then each line of "generic-ua" will be
12656case-insensitively matched. Then the word "test" will be insensitively matched
12657as well.
12658
12659The "-m" flag is used to select a specific pattern matching method on the input
12660sample. All ACL-specific criteria imply a pattern matching method and generally
12661do not need this flag. However, this flag is useful with generic sample fetch
12662methods to describe how they're going to be matched against the patterns. This
12663is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012664obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012665followed by a pattern matching method name, this method is used instead of the
12666default one for the criterion. This makes it possible to match contents in ways
12667that were not initially planned, or with sample fetch methods which return a
12668string. The matching method also affects the way the patterns are parsed.
12669
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012670The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12671By default, if the parser cannot parse ip address it considers that the parsed
12672string is maybe a domain name and try dns resolution. The flag "-n" disable this
12673resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12674server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050012675waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012676flag "-n" disable this behavior. Note also that during the runtime, this
12677function is disabled for the dynamic acl modifications.
12678
Willy Tarreau74ca5042013-06-11 23:12:07 +020012679There are some restrictions however. Not all methods can be used with all
12680sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12681be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012682
12683 - "found" : only check if the requested sample could be found in the stream,
12684 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012685 to pass any pattern to avoid confusion. This matching method is
12686 particularly useful to detect presence of certain contents such
12687 as headers, cookies, etc... even if they are empty and without
12688 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012689
12690 - "bool" : check the value as a boolean. It can only be applied to fetches
12691 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012692 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012693
12694 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012695 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012696
12697 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012698 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012699
Davor Ocelice9ed2812017-12-25 17:49:28 +010012700 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012701 binary sequence. This may be used with binary or string samples.
12702
12703 - "len" : match the sample's length as an integer. This may be used with
12704 binary or string samples.
12705
Willy Tarreau74ca5042013-06-11 23:12:07 +020012706 - "str" : exact match : match the contents against a string. This may be
12707 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012708
Willy Tarreau74ca5042013-06-11 23:12:07 +020012709 - "sub" : substring match : check that the contents contain at least one of
12710 the provided string patterns. This may be used with binary or
12711 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012712
Willy Tarreau74ca5042013-06-11 23:12:07 +020012713 - "reg" : regex match : match the contents against a list of regular
12714 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012715
Willy Tarreau74ca5042013-06-11 23:12:07 +020012716 - "beg" : prefix match : check that the contents begin like the provided
12717 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012718
Willy Tarreau74ca5042013-06-11 23:12:07 +020012719 - "end" : suffix match : check that the contents end like the provided
12720 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012721
Willy Tarreau74ca5042013-06-11 23:12:07 +020012722 - "dir" : subdir match : check that a slash-delimited portion of the
12723 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012724 This may be used with binary or string samples.
12725
Willy Tarreau74ca5042013-06-11 23:12:07 +020012726 - "dom" : domain match : check that a dot-delimited portion of the contents
12727 exactly match one of the provided string patterns. This may be
12728 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012729
12730For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12731request, it is possible to do :
12732
12733 acl jsess_present cook(JSESSIONID) -m found
12734
12735In order to apply a regular expression on the 500 first bytes of data in the
12736buffer, one would use the following acl :
12737
12738 acl script_tag payload(0,500) -m reg -i <script>
12739
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012740On systems where the regex library is much slower when using "-i", it is
12741possible to convert the sample to lowercase before matching, like this :
12742
12743 acl script_tag payload(0,500),lower -m reg <script>
12744
Willy Tarreau74ca5042013-06-11 23:12:07 +020012745All ACL-specific criteria imply a default matching method. Most often, these
12746criteria are composed by concatenating the name of the original sample fetch
12747method and the matching method. For example, "hdr_beg" applies the "beg" match
12748to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12749criteria rely on a sample fetch method, it is always possible instead to use
12750the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012751
Willy Tarreau74ca5042013-06-11 23:12:07 +020012752If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012753the matching method is simply applied to the underlying sample fetch method.
12754For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012755
Willy Tarreau74ca5042013-06-11 23:12:07 +020012756 acl short_form hdr_beg(host) www.
12757 acl alternate1 hdr_beg(host) -m beg www.
12758 acl alternate2 hdr_dom(host) -m beg www.
12759 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012760
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012761
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012762The table below summarizes the compatibility matrix between sample or converter
12763types and the pattern types to fetch against. It indicates for each compatible
12764combination the name of the matching method to be used, surrounded with angle
12765brackets ">" and "<" when the method is the default one and will work by
12766default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012767
Willy Tarreau74ca5042013-06-11 23:12:07 +020012768 +-------------------------------------------------+
12769 | Input sample type |
12770 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012771 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012772 +----------------------+---------+---------+---------+---------+---------+
12773 | none (presence only) | found | found | found | found | found |
12774 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012775 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012776 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012777 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012778 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012779 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012780 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012781 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012782 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012783 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012784 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012785 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012786 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012787 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012788 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012789 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012790 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012791 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012792 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012793 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012794 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012795 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012796 +----------------------+---------+---------+---------+---------+---------+
12797 | hex block | | | | bin | bin |
12798 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012799
12800
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128017.1.1. Matching booleans
12802------------------------
12803
12804In order to match a boolean, no value is needed and all values are ignored.
12805Boolean matching is used by default for all fetch methods of type "boolean".
12806When boolean matching is used, the fetched value is returned as-is, which means
12807that a boolean "true" will always match and a boolean "false" will never match.
12808
12809Boolean matching may also be enforced using "-m bool" on fetch methods which
12810return an integer value. Then, integer value 0 is converted to the boolean
12811"false" and all other values are converted to "true".
12812
Willy Tarreau6a06a402007-07-15 20:15:28 +020012813
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128147.1.2. Matching integers
12815------------------------
12816
12817Integer matching applies by default to integer fetch methods. It can also be
12818enforced on boolean fetches using "-m int". In this case, "false" is converted
12819to the integer 0, and "true" is converted to the integer 1.
12820
12821Integer matching also supports integer ranges and operators. Note that integer
12822matching only applies to positive values. A range is a value expressed with a
12823lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012824
12825For instance, "1024:65535" is a valid range to represent a range of
12826unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12827representation of privileged ports, and ":1023" would also work.
12828
Willy Tarreau62644772008-07-16 18:36:06 +020012829As a special case, some ACL functions support decimal numbers which are in fact
12830two integers separated by a dot. This is used with some version checks for
12831instance. All integer properties apply to those decimal numbers, including
12832ranges and operators.
12833
Willy Tarreau6a06a402007-07-15 20:15:28 +020012834For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012835operators with ranges does not make much sense and is strongly discouraged.
12836Similarly, it does not make much sense to perform order comparisons with a set
12837of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012838
Willy Tarreau0ba27502007-12-24 16:55:16 +010012839Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012840
12841 eq : true if the tested value equals at least one value
12842 ge : true if the tested value is greater than or equal to at least one value
12843 gt : true if the tested value is greater than at least one value
12844 le : true if the tested value is less than or equal to at least one value
12845 lt : true if the tested value is less than at least one value
12846
Willy Tarreau0ba27502007-12-24 16:55:16 +010012847For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012848
12849 acl negative-length hdr_val(content-length) lt 0
12850
Willy Tarreau62644772008-07-16 18:36:06 +020012851This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12852
12853 acl sslv3 req_ssl_ver 3:3.1
12854
Willy Tarreau6a06a402007-07-15 20:15:28 +020012855
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128567.1.3. Matching strings
12857-----------------------
12858
12859String matching applies to string or binary fetch methods, and exists in 6
12860different forms :
12861
12862 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012863 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012864
12865 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012866 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012867
12868 - prefix match (-m beg) : the patterns are compared with the beginning of
12869 the extracted string, and the ACL matches if any of them matches.
12870
12871 - suffix match (-m end) : the patterns are compared with the end of the
12872 extracted string, and the ACL matches if any of them matches.
12873
Baptiste Assmann33db6002016-03-06 23:32:10 +010012874 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012875 string, delimited with slashes ("/"), and the ACL matches if any of them
12876 matches.
12877
12878 - domain match (-m dom) : the patterns are looked up inside the extracted
12879 string, delimited with dots ("."), and the ACL matches if any of them
12880 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012881
12882String matching applies to verbatim strings as they are passed, with the
12883exception of the backslash ("\") which makes it possible to escape some
12884characters such as the space. If the "-i" flag is passed before the first
12885string, then the matching will be performed ignoring the case. In order
12886to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012887before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012888
12889
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128907.1.4. Matching regular expressions (regexes)
12891---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012892
12893Just like with string matching, regex matching applies to verbatim strings as
12894they are passed, with the exception of the backslash ("\") which makes it
12895possible to escape some characters such as the space. If the "-i" flag is
12896passed before the first regex, then the matching will be performed ignoring
12897the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012898the "--" flag before the first string. Same principle applies of course to
12899match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012900
12901
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129027.1.5. Matching arbitrary data blocks
12903-------------------------------------
12904
12905It is possible to match some extracted samples against a binary block which may
12906not safely be represented as a string. For this, the patterns must be passed as
12907a series of hexadecimal digits in an even number, when the match method is set
12908to binary. Each sequence of two digits will represent a byte. The hexadecimal
12909digits may be used upper or lower case.
12910
12911Example :
12912 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12913 acl hello payload(0,6) -m bin 48656c6c6f0a
12914
12915
129167.1.6. Matching IPv4 and IPv6 addresses
12917---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012918
12919IPv4 addresses values can be specified either as plain addresses or with a
12920netmask appended, in which case the IPv4 address matches whenever it is
12921within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012922host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012923difficult to read and debug configurations. If hostnames are used, you should
12924at least ensure that they are present in /etc/hosts so that the configuration
12925does not depend on any random DNS match at the moment the configuration is
12926parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012927
Daniel Schnellereba56342016-04-13 00:26:52 +020012928The dotted IPv4 address notation is supported in both regular as well as the
12929abbreviated form with all-0-octets omitted:
12930
12931 +------------------+------------------+------------------+
12932 | Example 1 | Example 2 | Example 3 |
12933 +------------------+------------------+------------------+
12934 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12935 | 192.168.1 | 10.12 | 127.1 |
12936 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
12937 | 192.168.1/22 | 10.12/8 | 127.1/8 |
12938 +------------------+------------------+------------------+
12939
12940Notice that this is different from RFC 4632 CIDR address notation in which
12941192.168.42/24 would be equivalent to 192.168.42.0/24.
12942
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012943IPv6 may be entered in their usual form, with or without a netmask appended.
12944Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
12945trouble with randomly resolved IP addresses, host names are never allowed in
12946IPv6 patterns.
12947
12948HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
12949following situations :
12950 - tested address is IPv4, pattern address is IPv4, the match applies
12951 in IPv4 using the supplied mask if any.
12952 - tested address is IPv6, pattern address is IPv6, the match applies
12953 in IPv6 using the supplied mask if any.
12954 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
12955 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
12956 ::IPV4 or ::ffff:IPV4, otherwise it fails.
12957 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
12958 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
12959 applied in IPv6 using the supplied IPv6 mask.
12960
Willy Tarreau74ca5042013-06-11 23:12:07 +020012961
129627.2. Using ACLs to form conditions
12963----------------------------------
12964
12965Some actions are only performed upon a valid condition. A condition is a
12966combination of ACLs with operators. 3 operators are supported :
12967
12968 - AND (implicit)
12969 - OR (explicit with the "or" keyword or the "||" operator)
12970 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020012971
Willy Tarreau74ca5042013-06-11 23:12:07 +020012972A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020012973
Willy Tarreau74ca5042013-06-11 23:12:07 +020012974 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020012975
Willy Tarreau74ca5042013-06-11 23:12:07 +020012976Such conditions are generally used after an "if" or "unless" statement,
12977indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020012978
Willy Tarreau74ca5042013-06-11 23:12:07 +020012979For instance, to block HTTP requests to the "*" URL with methods other than
12980"OPTIONS", as well as POST requests without content-length, and GET or HEAD
12981requests with a content-length greater than 0, and finally every request which
12982is not either GET/HEAD/POST/OPTIONS !
12983
12984 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012985 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
12986 http-request deny if METH_GET HTTP_CONTENT
12987 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020012988
12989To select a different backend for requests to static contents on the "www" site
12990and to every request on the "img", "video", "download" and "ftp" hosts :
12991
12992 acl url_static path_beg /static /images /img /css
12993 acl url_static path_end .gif .png .jpg .css .js
12994 acl host_www hdr_beg(host) -i www
12995 acl host_static hdr_beg(host) -i img. video. download. ftp.
12996
Davor Ocelice9ed2812017-12-25 17:49:28 +010012997 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020012998 # of host "www". Use backend "www" for the rest.
12999 use_backend static if host_static or host_www url_static
13000 use_backend www if host_www
13001
13002It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13003expressions that are built on the fly without needing to be declared. They must
13004be enclosed between braces, with a space before and after each brace (because
13005the braces must be seen as independent words). Example :
13006
13007 The following rule :
13008
13009 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013010 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013011
13012 Can also be written that way :
13013
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013014 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013015
13016It is generally not recommended to use this construct because it's a lot easier
13017to leave errors in the configuration when written that way. However, for very
13018simple rules matching only one source IP address for instance, it can make more
13019sense to use them than to declare ACLs with random names. Another example of
13020good use is the following :
13021
13022 With named ACLs :
13023
13024 acl site_dead nbsrv(dynamic) lt 2
13025 acl site_dead nbsrv(static) lt 2
13026 monitor fail if site_dead
13027
13028 With anonymous ACLs :
13029
13030 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13031
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013032See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13033keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013034
13035
130367.3. Fetching samples
13037---------------------
13038
13039Historically, sample fetch methods were only used to retrieve data to match
13040against patterns using ACLs. With the arrival of stick-tables, a new class of
13041sample fetch methods was created, most often sharing the same syntax as their
13042ACL counterpart. These sample fetch methods are also known as "fetches". As
13043of now, ACLs and fetches have converged. All ACL fetch methods have been made
13044available as fetch methods, and ACLs may use any sample fetch method as well.
13045
13046This section details all available sample fetch methods and their output type.
13047Some sample fetch methods have deprecated aliases that are used to maintain
13048compatibility with existing configurations. They are then explicitly marked as
13049deprecated and should not be used in new setups.
13050
13051The ACL derivatives are also indicated when available, with their respective
13052matching methods. These ones all have a well defined default pattern matching
13053method, so it is never necessary (though allowed) to pass the "-m" option to
13054indicate how the sample will be matched using ACLs.
13055
13056As indicated in the sample type versus matching compatibility matrix above,
13057when using a generic sample fetch method in an ACL, the "-m" option is
13058mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13059the same keyword exists as an ACL keyword and as a standard fetch method, the
13060ACL engine will automatically pick the ACL-only one by default.
13061
13062Some of these keywords support one or multiple mandatory arguments, and one or
13063multiple optional arguments. These arguments are strongly typed and are checked
13064when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013065incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13066are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013067is optional, it will be indicated below between square brackets ('[ ]'). When
13068all arguments are optional, the parenthesis may be omitted.
13069
13070Thus, the syntax of a standard sample fetch method is one of the following :
13071 - name
13072 - name(arg1)
13073 - name(arg1,arg2)
13074
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013075
130767.3.1. Converters
13077-----------------
13078
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013079Sample fetch methods may be combined with transformations to be applied on top
13080of the fetched sample (also called "converters"). These combinations form what
13081is called "sample expressions" and the result is a "sample". Initially this
13082was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013083has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013084unique-id-format, add-header, ...).
13085
13086These transformations are enumerated as a series of specific keywords after the
13087sample fetch method. These keywords may equally be appended immediately after
13088the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013089support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013090
Willy Tarreau97707872015-01-27 15:12:13 +010013091A certain category of converters are bitwise and arithmetic operators which
13092support performing basic operations on integers. Some bitwise operations are
13093supported (and, or, xor, cpl) and some arithmetic operations are supported
13094(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13095bool) which make it possible to report a match without having to write an ACL.
13096
Willy Tarreau74ca5042013-06-11 23:12:07 +020013097The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013098
Ben Shillitof25e8e52016-12-02 14:25:37 +00001309951d.single(<prop>[,<prop>*])
13100 Returns values for the properties requested as a string, where values are
13101 separated by the delimiter specified with "51degrees-property-separator".
13102 The device is identified using the User-Agent header passed to the
13103 converter. The function can be passed up to five property names, and if a
13104 property name can't be found, the value "NoData" is returned.
13105
13106 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013107 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13108 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013109 # User-Agent passed to the converter.
13110 frontend http-in
13111 bind *:8081
13112 default_backend servers
13113 http-request set-header X-51D-DeviceTypeMobileTablet \
13114 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13115
Willy Tarreau97707872015-01-27 15:12:13 +010013116add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013117 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013118 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013119 name. The name of the variable starts with an indication about its scope. The
13120 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013121 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013122 "sess" : the variable is shared with the whole session
13123 "txn" : the variable is shared with the transaction (request and response)
13124 "req" : the variable is shared only during request processing
13125 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013126 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013127 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013128
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010013129aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
13130 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
13131 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
13132 need to be base64 encoded and the returned result is in raw byte format.
13133 If the <aead_tag> validation fails, the converter doesn't return any data.
13134 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
13135 converter requires at least OpenSSL 1.0.1.
13136
13137 Example:
13138 http-response set-header X-Decrypted-Text %[var(txn.enc),\
13139 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
13140
Willy Tarreau97707872015-01-27 15:12:13 +010013141and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013142 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013143 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013144 numeric value or a variable name. The name of the variable starts with an
13145 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013146 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013147 "sess" : the variable is shared with the whole session
13148 "txn" : the variable is shared with the transaction (request and response)
13149 "req" : the variable is shared only during request processing
13150 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013151 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013152 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013153
Holger Just1bfc24b2017-05-06 00:56:53 +020013154b64dec
13155 Converts (decodes) a base64 encoded input string to its binary
13156 representation. It performs the inverse operation of base64().
13157
Emeric Brun53d1a982014-04-30 18:21:37 +020013158base64
13159 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013160 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013161 an SSL ID can be copied in a header).
13162
Willy Tarreau97707872015-01-27 15:12:13 +010013163bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013164 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013165 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013166 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013167 presence of a flag).
13168
Emeric Brun54c4ac82014-11-03 15:32:43 +010013169bytes(<offset>[,<length>])
13170 Extracts some bytes from an input binary sample. The result is a binary
13171 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013172 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013173
Willy Tarreau280f42b2018-02-19 15:34:12 +010013174concat([<start>],[<var>],[<end>])
13175 Concatenates up to 3 fields after the current sample which is then turned to
13176 a string. The first one, <start>, is a constant string, that will be appended
13177 immediately after the existing sample. It may be omitted if not used. The
13178 second one, <var>, is a variable name. The variable will be looked up, its
13179 contents converted to a string, and it will be appended immediately after the
13180 <first> part. If the variable is not found, nothing is appended. It may be
13181 omitted as well. The third field, <end> is a constant string that will be
13182 appended after the variable. It may also be omitted. Together, these elements
13183 allow to concatenate variables with delimiters to an existing set of
13184 variables. This can be used to build new variables made of a succession of
John Roeslerfb2fce12019-07-10 15:45:51 -050013185 other variables, such as colon-delimited values. Note that due to the config
Willy Tarreau280f42b2018-02-19 15:34:12 +010013186 parser, it is not possible to use a comma nor a closing parenthesis as
John Roeslerfb2fce12019-07-10 15:45:51 -050013187 delimiters.
Willy Tarreau280f42b2018-02-19 15:34:12 +010013188
13189 Example:
13190 tcp-request session set-var(sess.src) src
13191 tcp-request session set-var(sess.dn) ssl_c_s_dn
13192 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13193 http-request set-header x-hap-sig %[var(txn.sig)]
13194
Willy Tarreau97707872015-01-27 15:12:13 +010013195cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013196 Takes the input value of type signed integer, applies a ones-complement
13197 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013198
Willy Tarreau80599772015-01-20 19:35:24 +010013199crc32([<avalanche>])
13200 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13201 hash function. Optionally, it is possible to apply a full avalanche hash
13202 function to the output if the optional <avalanche> argument equals 1. This
13203 converter uses the same functions as used by the various hash-based load
13204 balancing algorithms, so it will provide exactly the same results. It is
13205 provided for compatibility with other software which want a CRC32 to be
13206 computed on some input keys, so it follows the most common implementation as
13207 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13208 but may provide a better or at least less predictable distribution. It must
13209 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013210 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13211
13212crc32c([<avalanche>])
13213 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13214 hash function. Optionally, it is possible to apply a full avalanche hash
13215 function to the output if the optional <avalanche> argument equals 1. This
13216 converter uses the same functions as described in RFC4960, Appendix B [8].
13217 It is provided for compatibility with other software which want a CRC32C to be
13218 computed on some input keys. It is slower than the other algorithms and it must
13219 not be used for security purposes as a 32-bit hash is trivial to break. See
13220 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013221
David Carlier29b3ca32015-09-25 14:09:21 +010013222da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013223 Asks the DeviceAtlas converter to identify the User Agent string passed on
13224 input, and to emit a string made of the concatenation of the properties
13225 enumerated in argument, delimited by the separator defined by the global
13226 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013227 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013228 configuration language.
13229
13230 Example:
13231 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013232 bind *:8881
13233 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013234 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 +020013235
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013236debug
13237 This converter is used as debug tool. It dumps on screen the content and the
13238 type of the input sample. The sample is returned as is on its output. This
13239 converter only exists when haproxy was built with debugging enabled.
13240
Willy Tarreau97707872015-01-27 15:12:13 +010013241div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013242 Divides the input value of type signed integer by <value>, and returns the
13243 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013244 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013245 variable name. The name of the variable starts with an indication about its
13246 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013247 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013248 "sess" : the variable is shared with the whole session
13249 "txn" : the variable is shared with the transaction (request and response)
13250 "req" : the variable is shared only during request processing
13251 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013252 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013253 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013254
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013255djb2([<avalanche>])
13256 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13257 hash function. Optionally, it is possible to apply a full avalanche hash
13258 function to the output if the optional <avalanche> argument equals 1. This
13259 converter uses the same functions as used by the various hash-based load
13260 balancing algorithms, so it will provide exactly the same results. It is
13261 mostly intended for debugging, but can be used as a stick-table entry to
13262 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013263 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13264 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013265
Willy Tarreau97707872015-01-27 15:12:13 +010013266even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013267 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013268 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13269
Marcin Deranek9631a282018-04-16 14:30:46 +020013270field(<index>,<delimiters>[,<count>])
13271 Extracts the substring at the given index counting from the beginning
13272 (positive index) or from the end (negative index) considering given delimiters
13273 from an input string. Indexes start at 1 or -1 and delimiters are a string
13274 formatted list of chars. Optionally you can specify <count> of fields to
13275 extract (default: 1). Value of 0 indicates extraction of all remaining
13276 fields.
13277
13278 Example :
13279 str(f1_f2_f3__f5),field(5,_) # f5
13280 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13281 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13282 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13283 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013284
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013285hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013286 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013287 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013288 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 +020013289 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013290
Dragan Dosen3f957b22017-10-24 09:27:34 +020013291hex2i
13292 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050013293 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020013294
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013295http_date([<offset>])
13296 Converts an integer supposed to contain a date since epoch to a string
13297 representing this date in a format suitable for use in HTTP header fields. If
13298 an offset value is specified, then it is a number of seconds that is added to
13299 the date before the conversion is operated. This is particularly useful to
13300 emit Date header fields, Expires values in responses when combined with a
13301 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013302
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013303in_table(<table>)
13304 Uses the string representation of the input sample to perform a look up in
13305 the specified table. If the key is not found in the table, a boolean false
13306 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013307 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013308 or not a source IP address or an Authorization header was already seen).
13309
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013310ipmask(<mask4>, [<mask6>])
13311 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013312 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013313 table entries and as such use the same server. The mask4 can be passed in
13314 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13315 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13316 If no mask6 is given IPv6 addresses will fail to convert for backwards
13317 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013318
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013319json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013320 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013321 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013322 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013323 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13324 of errors:
13325 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13326 bytes, ...)
13327 - invalid range (the decoded value is within a UTF-8 prohibited range),
13328 - code overlong (the value is encoded with more bytes than necessary).
13329
13330 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13331 character is greater than 0xffff because the JSON string escape specification
13332 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13333 in 4 variants designated by a combination of two suffix letters : "p" for
13334 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13335 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013336 - "ascii" : never fails;
13337 - "utf8" : fails on any detected errors;
13338 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013339 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013340 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013341 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13342 characters corresponding to the other errors.
13343
13344 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013345 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013346
13347 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013348 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013349 capture request header user-agent len 150
13350 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013351
13352 Input request from client 127.0.0.1:
13353 GET / HTTP/1.0
13354 User-Agent: Very "Ugly" UA 1/2
13355
13356 Output log:
13357 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13358
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013359language(<value>[,<default>])
13360 Returns the value with the highest q-factor from a list as extracted from the
13361 "accept-language" header using "req.fhdr". Values with no q-factor have a
13362 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13363 belong to the list of semi-colon delimited <values> will be considered. The
13364 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13365 given list and a default value is provided, it is returned. Note that language
13366 names may have a variant after a dash ('-'). If this variant is present in the
13367 list, it will be matched, but if it is not, only the base language is checked.
13368 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013369 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013370 ordering of the values in the request counts, as the first value among
13371 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013372
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013373 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013374
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013375 # this configuration switches to the backend matching a
13376 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013377
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013378 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13379 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13380 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13381 use_backend spanish if es
13382 use_backend french if fr
13383 use_backend english if en
13384 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013385
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013386length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013387 Get the length of the string. This can only be placed after a string
13388 sample fetch function or after a transformation keyword returning a string
13389 type. The result is of type integer.
13390
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013391lower
13392 Convert a string sample to lower case. This can only be placed after a string
13393 sample fetch function or after a transformation keyword returning a string
13394 type. The result is of type string.
13395
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013396ltime(<format>[,<offset>])
13397 Converts an integer supposed to contain a date since epoch to a string
13398 representing this date in local time using a format defined by the <format>
13399 string using strftime(3). The purpose is to allow any date format to be used
13400 in logs. An optional <offset> in seconds may be applied to the input date
13401 (positive or negative). See the strftime() man page for the format supported
13402 by your operating system. See also the utime converter.
13403
13404 Example :
13405
13406 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013407 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013408 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13409
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013410map(<map_file>[,<default_value>])
13411map_<match_type>(<map_file>[,<default_value>])
13412map_<match_type>_<output_type>(<map_file>[,<default_value>])
13413 Search the input value from <map_file> using the <match_type> matching method,
13414 and return the associated value converted to the type <output_type>. If the
13415 input value cannot be found in the <map_file>, the converter returns the
13416 <default_value>. If the <default_value> is not set, the converter fails and
13417 acts as if no input value could be fetched. If the <match_type> is not set, it
13418 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13419 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13420 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013421
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013422 It is important to avoid overlapping between the keys : IP addresses and
13423 strings are stored in trees, so the first of the finest match will be used.
13424 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013425
Tim Düsterhus4896c442016-11-29 02:15:19 +010013426 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013427 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013428
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013429 input type | match method | output type str | output type int | output type ip
13430 -----------+--------------+-----------------+-----------------+---------------
13431 str | str | map_str | map_str_int | map_str_ip
13432 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013433 str | beg | map_beg | map_beg_int | map_end_ip
13434 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013435 str | sub | map_sub | map_sub_int | map_sub_ip
13436 -----------+--------------+-----------------+-----------------+---------------
13437 str | dir | map_dir | map_dir_int | map_dir_ip
13438 -----------+--------------+-----------------+-----------------+---------------
13439 str | dom | map_dom | map_dom_int | map_dom_ip
13440 -----------+--------------+-----------------+-----------------+---------------
13441 str | end | map_end | map_end_int | map_end_ip
13442 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013443 str | reg | map_reg | map_reg_int | map_reg_ip
13444 -----------+--------------+-----------------+-----------------+---------------
13445 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013446 -----------+--------------+-----------------+-----------------+---------------
13447 int | int | map_int | map_int_int | map_int_ip
13448 -----------+--------------+-----------------+-----------------+---------------
13449 ip | ip | map_ip | map_ip_int | map_ip_ip
13450 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013451
Thierry Fournier8feaa662016-02-10 22:55:20 +010013452 The special map called "map_regm" expect matching zone in the regular
13453 expression and modify the output replacing back reference (like "\1") by
13454 the corresponding match text.
13455
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013456 The file contains one key + value per line. Lines which start with '#' are
13457 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13458 is then the first "word" (series of non-space/tabs characters), and the value
13459 is what follows this series of space/tab till the end of the line excluding
13460 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013461
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013462 Example :
13463
13464 # this is a comment and is ignored
13465 2.22.246.0/23 United Kingdom \n
13466 <-><-----------><--><------------><---->
13467 | | | | `- trailing spaces ignored
13468 | | | `---------- value
13469 | | `-------------------- middle spaces ignored
13470 | `---------------------------- key
13471 `------------------------------------ leading spaces ignored
13472
Willy Tarreau97707872015-01-27 15:12:13 +010013473mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013474 Divides the input value of type signed integer by <value>, and returns the
13475 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013476 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013477 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013478 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013479 "sess" : the variable is shared with the whole session
13480 "txn" : the variable is shared with the transaction (request and response)
13481 "req" : the variable is shared only during request processing
13482 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013483 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013484 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013485
13486mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013487 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013488 the product as an signed integer. In case of overflow, the largest possible
13489 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013490 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013491 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013492 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013493 "sess" : the variable is shared with the whole session
13494 "txn" : the variable is shared with the transaction (request and response)
13495 "req" : the variable is shared only during request processing
13496 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013497 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013498 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013499
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013500nbsrv
13501 Takes an input value of type string, interprets it as a backend name and
13502 returns the number of usable servers in that backend. Can be used in places
13503 where we want to look up a backend from a dynamic name, like a result of a
13504 map lookup.
13505
Willy Tarreau97707872015-01-27 15:12:13 +010013506neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013507 Takes the input value of type signed integer, computes the opposite value,
13508 and returns the remainder as an signed integer. 0 is identity. This operator
13509 is provided for reversed subtracts : in order to subtract the input from a
13510 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013511
13512not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013513 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013514 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013515 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013516 absence of a flag).
13517
13518odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013519 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013520 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13521
13522or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013523 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013524 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013525 numeric value or a variable name. The name of the variable starts with an
13526 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013527 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013528 "sess" : the variable is shared with the whole session
13529 "txn" : the variable is shared with the transaction (request and response)
13530 "req" : the variable is shared only during request processing
13531 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013532 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013533 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013534
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013535protobuf(<field_number>,[<field_type>])
13536 This extracts the protocol buffers message field in raw mode of an input binary
13537 sample representation of a protocol buffer message with <field_number> as field
13538 number (dotted notation) if <field_type> is not present, or as an integer sample
13539 if this field is present (see also "ungrpc" below).
13540 The list of the authorized types is the following one: "int32", "int64", "uint32",
13541 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13542 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13543 "float" for the wire type 5. Note that "string" is considered as a length-delimited
13544 type, so it does not require any <field_type> argument to be extracted.
13545 More information may be found here about the protocol buffers message field types:
13546 https://developers.google.com/protocol-buffers/docs/encoding
13547
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013548regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013549 Applies a regex-based substitution to the input string. It does the same
13550 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13551 default it will replace in the input string the first occurrence of the
13552 largest part matching the regular expression <regex> with the substitution
13553 string <subst>. It is possible to replace all occurrences instead by adding
13554 the flag "g" in the third argument <flags>. It is also possible to make the
13555 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13556 string, it is made up from the concatenation of all desired flags. Thus if
13557 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13558 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013559 configuration parser, some characters such as closing parenthesis, closing
13560 square brackets or comma are not possible to use in the arguments. The first
13561 use of this converter is to replace certain characters or sequence of
13562 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013563
13564 Example :
13565
13566 # de-duplicate "/" in header "x-path".
13567 # input: x-path: /////a///b/c/xzxyz/
13568 # output: x-path: /a/b/c/xzxyz/
13569 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13570
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013571capture-req(<id>)
13572 Capture the string entry in the request slot <id> and returns the entry as
13573 is. If the slot doesn't exist, the capture fails silently.
13574
13575 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013576 "http-response capture", "capture.req.hdr" and
13577 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013578
13579capture-res(<id>)
13580 Capture the string entry in the response slot <id> and returns the entry as
13581 is. If the slot doesn't exist, the capture fails silently.
13582
13583 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013584 "http-response capture", "capture.req.hdr" and
13585 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013586
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013587sdbm([<avalanche>])
13588 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13589 hash function. Optionally, it is possible to apply a full avalanche hash
13590 function to the output if the optional <avalanche> argument equals 1. This
13591 converter uses the same functions as used by the various hash-based load
13592 balancing algorithms, so it will provide exactly the same results. It is
13593 mostly intended for debugging, but can be used as a stick-table entry to
13594 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013595 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13596 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013597
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013598set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013599 Sets a variable with the input content and returns the content on the output
13600 as-is. The variable keeps the value and the associated input type. The name of
13601 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013602 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013603 "sess" : the variable is shared with the whole session
13604 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013605 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013606 "req" : the variable is shared only during request processing,
13607 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013608 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013609 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013610
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013611sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020013612 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013613 sample with length of 20 bytes.
13614
Tim Duesterhusd4376302019-06-17 12:41:44 +020013615sha2([<bits>])
13616 Converts a binary input sample to a digest in the SHA-2 family. The result
13617 is a binary sample with length of <bits>/8 bytes.
13618
13619 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
13620 SHA-<bits>. The default value is 256.
13621
13622 Please note that this converter is only available when haproxy has been
13623 compiled with USE_OPENSSL.
13624
Tim Duesterhusca097c12018-04-27 21:18:45 +020013625strcmp(<var>)
13626 Compares the contents of <var> with the input value of type string. Returns
13627 the result as a signed integer compatible with strcmp(3): 0 if both strings
13628 are identical. A value less than 0 if the left string is lexicographically
13629 smaller than the right string or if the left string is shorter. A value greater
13630 than 0 otherwise (right string greater than left string or the right string is
13631 shorter).
13632
13633 Example :
13634
13635 http-request set-var(txn.host) hdr(host)
13636 # Check whether the client is attempting domain fronting.
13637 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13638
13639
Willy Tarreau97707872015-01-27 15:12:13 +010013640sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013641 Subtracts <value> from the input value of type signed integer, and returns
13642 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013643 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013644 or a variable name. The name of the variable starts with an indication about
13645 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013646 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013647 "sess" : the variable is shared with the whole session
13648 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013649 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013650 "req" : the variable is shared only during request processing,
13651 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013652 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013653 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013654
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013655table_bytes_in_rate(<table>)
13656 Uses the string representation of the input sample to perform a look up in
13657 the specified table. If the key is not found in the table, integer value zero
13658 is returned. Otherwise the converter returns the average client-to-server
13659 bytes rate associated with the input sample in the designated table, measured
13660 in amount of bytes over the period configured in the table. See also the
13661 sc_bytes_in_rate sample fetch keyword.
13662
13663
13664table_bytes_out_rate(<table>)
13665 Uses the string representation of the input sample to perform a look up in
13666 the specified table. If the key is not found in the table, integer value zero
13667 is returned. Otherwise the converter returns the average server-to-client
13668 bytes rate associated with the input sample in the designated table, measured
13669 in amount of bytes over the period configured in the table. See also the
13670 sc_bytes_out_rate sample fetch keyword.
13671
13672table_conn_cnt(<table>)
13673 Uses the string representation of the input sample to perform a look up in
13674 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013675 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013676 connections associated with the input sample in the designated table. See
13677 also the sc_conn_cnt sample fetch keyword.
13678
13679table_conn_cur(<table>)
13680 Uses the string representation of the input sample to perform a look up in
13681 the specified table. If the key is not found in the table, integer value zero
13682 is returned. Otherwise the converter returns the current amount of concurrent
13683 tracked connections associated with the input sample in the designated table.
13684 See also the sc_conn_cur sample fetch keyword.
13685
13686table_conn_rate(<table>)
13687 Uses the string representation of the input sample to perform a look up in
13688 the specified table. If the key is not found in the table, integer value zero
13689 is returned. Otherwise the converter returns the average incoming connection
13690 rate associated with the input sample in the designated table. See also the
13691 sc_conn_rate sample fetch keyword.
13692
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013693table_gpt0(<table>)
13694 Uses the string representation of the input sample to perform a look up in
13695 the specified table. If the key is not found in the table, boolean value zero
13696 is returned. Otherwise the converter returns the current value of the first
13697 general purpose tag associated with the input sample in the designated table.
13698 See also the sc_get_gpt0 sample fetch keyword.
13699
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013700table_gpc0(<table>)
13701 Uses the string representation of the input sample to perform a look up in
13702 the specified table. If the key is not found in the table, integer value zero
13703 is returned. Otherwise the converter returns the current value of the first
13704 general purpose counter associated with the input sample in the designated
13705 table. See also the sc_get_gpc0 sample fetch keyword.
13706
13707table_gpc0_rate(<table>)
13708 Uses the string representation of the input sample to perform a look up in
13709 the specified table. If the key is not found in the table, integer value zero
13710 is returned. Otherwise the converter returns the frequency which the gpc0
13711 counter was incremented over the configured period in the table, associated
13712 with the input sample in the designated table. See also the sc_get_gpc0_rate
13713 sample fetch keyword.
13714
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013715table_gpc1(<table>)
13716 Uses the string representation of the input sample to perform a look up in
13717 the specified table. If the key is not found in the table, integer value zero
13718 is returned. Otherwise the converter returns the current value of the second
13719 general purpose counter associated with the input sample in the designated
13720 table. See also the sc_get_gpc1 sample fetch keyword.
13721
13722table_gpc1_rate(<table>)
13723 Uses the string representation of the input sample to perform a look up in
13724 the specified table. If the key is not found in the table, integer value zero
13725 is returned. Otherwise the converter returns the frequency which the gpc1
13726 counter was incremented over the configured period in the table, associated
13727 with the input sample in the designated table. See also the sc_get_gpc1_rate
13728 sample fetch keyword.
13729
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013730table_http_err_cnt(<table>)
13731 Uses the string representation of the input sample to perform a look up in
13732 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013733 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013734 errors associated with the input sample in the designated table. See also the
13735 sc_http_err_cnt sample fetch keyword.
13736
13737table_http_err_rate(<table>)
13738 Uses the string representation of the input sample to perform a look up in
13739 the specified table. If the key is not found in the table, integer value zero
13740 is returned. Otherwise the average rate of HTTP errors associated with the
13741 input sample in the designated table, measured in amount of errors over the
13742 period configured in the table. See also the sc_http_err_rate sample fetch
13743 keyword.
13744
13745table_http_req_cnt(<table>)
13746 Uses the string representation of the input sample to perform a look up in
13747 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013748 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013749 requests associated with the input sample in the designated table. See also
13750 the sc_http_req_cnt sample fetch keyword.
13751
13752table_http_req_rate(<table>)
13753 Uses the string representation of the input sample to perform a look up in
13754 the specified table. If the key is not found in the table, integer value zero
13755 is returned. Otherwise the average rate of HTTP requests associated with the
13756 input sample in the designated table, measured in amount of requests over the
13757 period configured in the table. See also the sc_http_req_rate sample fetch
13758 keyword.
13759
13760table_kbytes_in(<table>)
13761 Uses the string representation of the input sample to perform a look up in
13762 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013763 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013764 to-server data associated with the input sample in the designated table,
13765 measured in kilobytes. The test is currently performed on 32-bit integers,
13766 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13767 keyword.
13768
13769table_kbytes_out(<table>)
13770 Uses the string representation of the input sample to perform a look up in
13771 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013772 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013773 to-client data associated with the input sample in the designated table,
13774 measured in kilobytes. The test is currently performed on 32-bit integers,
13775 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13776 keyword.
13777
13778table_server_id(<table>)
13779 Uses the string representation of the input sample to perform a look up in
13780 the specified table. If the key is not found in the table, integer value zero
13781 is returned. Otherwise the converter returns the server ID associated with
13782 the input sample in the designated table. A server ID is associated to a
13783 sample by a "stick" rule when a connection to a server succeeds. A server ID
13784 zero means that no server is associated with this key.
13785
13786table_sess_cnt(<table>)
13787 Uses the string representation of the input sample to perform a look up in
13788 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013789 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013790 sessions associated with the input sample in the designated table. Note that
13791 a session here refers to an incoming connection being accepted by the
13792 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13793 keyword.
13794
13795table_sess_rate(<table>)
13796 Uses the string representation of the input sample to perform a look up in
13797 the specified table. If the key is not found in the table, integer value zero
13798 is returned. Otherwise the converter returns the average incoming session
13799 rate associated with the input sample in the designated table. Note that a
13800 session here refers to an incoming connection being accepted by the
13801 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13802 keyword.
13803
13804table_trackers(<table>)
13805 Uses the string representation of the input sample to perform a look up in
13806 the specified table. If the key is not found in the table, integer value zero
13807 is returned. Otherwise the converter returns the current amount of concurrent
13808 connections tracking the same key as the input sample in the designated
13809 table. It differs from table_conn_cur in that it does not rely on any stored
13810 information but on the table's reference count (the "use" value which is
13811 returned by "show table" on the CLI). This may sometimes be more suited for
13812 layer7 tracking. It can be used to tell a server how many concurrent
13813 connections there are from a given address for example. See also the
13814 sc_trackers sample fetch keyword.
13815
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013816upper
13817 Convert a string sample to upper case. This can only be placed after a string
13818 sample fetch function or after a transformation keyword returning a string
13819 type. The result is of type string.
13820
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013821url_dec
13822 Takes an url-encoded string provided as input and returns the decoded
13823 version as output. The input and the output are of type string.
13824
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013825ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013826 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013827 sample representation of a gRPC message with <field_number> as field number
13828 (dotted notation) if <field_type> is not present, or as an integer sample if this
13829 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013830 The list of the authorized types is the following one: "int32", "int64", "uint32",
13831 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13832 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13833 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013834 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013835 More information may be found here about the protocol buffers message field types:
13836 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013837
13838 Example:
13839 // with such a protocol buffer .proto file content adapted from
13840 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
13841
13842 message Point {
13843 int32 latitude = 1;
13844 int32 longitude = 2;
13845 }
13846
13847 message PPoint {
13848 Point point = 59;
13849 }
13850
13851 message Rectangle {
13852 // One corner of the rectangle.
13853 PPoint lo = 48;
13854 // The other corner of the rectangle.
13855 PPoint hi = 49;
13856 }
13857
13858 let's say a body request is made of a "Rectangle" object value (two PPoint
13859 protocol buffers messages), the four protocol buffers fields could be
13860 extracted with these "ungrpc" directives:
13861
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013862 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
13863 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050013864 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013865 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
13866
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013867 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013868
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013869 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013870
John Roeslerfb2fce12019-07-10 15:45:51 -050013871 As a gRPC message is always made of a gRPC header followed by protocol buffers
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013872 messages, in the previous example the "latitude" of "lo" first PPoint
13873 could be extracted with these equivalent directives:
13874
13875 req.body,ungrpc(48.59),protobuf(1,int32)
13876 req.body,ungrpc(48),protobuf(59.1,int32)
13877 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
13878
13879 Note that the first convert must be "ungrpc", the remaining ones must be
13880 "protobuf" and only the last one may have or not a second argument to
13881 interpret the previous binary sample.
13882
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013883
Christopher Faulet85d79c92016-11-09 16:54:56 +010013884unset-var(<var name>)
13885 Unsets a variable if the input content is defined. The name of the variable
13886 starts with an indication about its scope. The scopes allowed are:
13887 "proc" : the variable is shared with the whole process
13888 "sess" : the variable is shared with the whole session
13889 "txn" : the variable is shared with the transaction (request and
13890 response),
13891 "req" : the variable is shared only during request processing,
13892 "res" : the variable is shared only during response processing.
13893 This prefix is followed by a name. The separator is a '.'. The name may only
13894 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13895
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013896utime(<format>[,<offset>])
13897 Converts an integer supposed to contain a date since epoch to a string
13898 representing this date in UTC time using a format defined by the <format>
13899 string using strftime(3). The purpose is to allow any date format to be used
13900 in logs. An optional <offset> in seconds may be applied to the input date
13901 (positive or negative). See the strftime() man page for the format supported
13902 by your operating system. See also the ltime converter.
13903
13904 Example :
13905
13906 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013907 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013908 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13909
Marcin Deranek9631a282018-04-16 14:30:46 +020013910word(<index>,<delimiters>[,<count>])
13911 Extracts the nth word counting from the beginning (positive index) or from
13912 the end (negative index) considering given delimiters from an input string.
13913 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
13914 Optionally you can specify <count> of words to extract (default: 1).
13915 Value of 0 indicates extraction of all remaining words.
13916
13917 Example :
13918 str(f1_f2_f3__f5),word(4,_) # f5
13919 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
13920 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
13921 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
13922 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013923
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013924wt6([<avalanche>])
13925 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13926 hash function. Optionally, it is possible to apply a full avalanche hash
13927 function to the output if the optional <avalanche> argument equals 1. This
13928 converter uses the same functions as used by the various hash-based load
13929 balancing algorithms, so it will provide exactly the same results. It is
13930 mostly intended for debugging, but can be used as a stick-table entry to
13931 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013932 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
13933 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013934
Willy Tarreau97707872015-01-27 15:12:13 +010013935xor(<value>)
13936 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013937 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013938 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013939 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013940 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013941 "sess" : the variable is shared with the whole session
13942 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013943 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013944 "req" : the variable is shared only during request processing,
13945 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013946 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013947 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013948
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013949xxh32([<seed>])
13950 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13951 variant of the XXHash hash function. This hash supports a seed which defaults
13952 to zero but a different value maybe passed as the <seed> argument. This hash
13953 is known to be very good and very fast so it can be used to hash URLs and/or
13954 URL parameters for use as stick-table keys to collect statistics with a low
13955 collision rate, though care must be taken as the algorithm is not considered
13956 as cryptographically secure.
13957
13958xxh64([<seed>])
13959 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13960 variant of the XXHash hash function. This hash supports a seed which defaults
13961 to zero but a different value maybe passed as the <seed> argument. This hash
13962 is known to be very good and very fast so it can be used to hash URLs and/or
13963 URL parameters for use as stick-table keys to collect statistics with a low
13964 collision rate, though care must be taken as the algorithm is not considered
13965 as cryptographically secure.
13966
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013967
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200139687.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020013969--------------------------------------------
13970
13971A first set of sample fetch methods applies to internal information which does
13972not even relate to any client information. These ones are sometimes used with
13973"monitor-fail" directives to report an internal status to external watchers.
13974The sample fetch methods described in this section are usable anywhere.
13975
13976always_false : boolean
13977 Always returns the boolean "false" value. It may be used with ACLs as a
13978 temporary replacement for another one when adjusting configurations.
13979
13980always_true : boolean
13981 Always returns the boolean "true" value. It may be used with ACLs as a
13982 temporary replacement for another one when adjusting configurations.
13983
13984avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013985 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020013986 divided by the number of active servers. The current backend is used if no
13987 backend is specified. This is very similar to "queue" except that the size of
13988 the farm is considered, in order to give a more accurate measurement of the
13989 time it may take for a new connection to be processed. The main usage is with
13990 ACL to return a sorry page to new users when it becomes certain they will get
13991 a degraded service, or to pass to the backend servers in a header so that
13992 they decide to work in degraded mode or to disable some functions to speed up
13993 the processing a bit. Note that in the event there would not be any active
13994 server anymore, twice the number of queued connections would be considered as
13995 the measured value. This is a fair estimate, as we expect one server to get
13996 back soon anyway, but we still prefer to send new traffic to another backend
13997 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
13998 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010013999
Willy Tarreau74ca5042013-06-11 23:12:07 +020014000be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020014001 Applies to the number of currently established connections on the backend,
14002 possibly including the connection being evaluated. If no backend name is
14003 specified, the current one is used. But it is also possible to check another
14004 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014005 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
14006
14007be_conn_free([<backend>]) : integer
14008 Returns an integer value corresponding to the number of available connections
14009 across available servers in the backend. Queue slots are not included. Backup
14010 servers are also not included, unless all other servers are down. If no
14011 backend name is specified, the current one is used. But it is also possible
14012 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014013 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
14014 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014015
14016 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
14017 (meaning unlimited), then this fetch clearly does not make sense, in which
14018 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014019
Willy Tarreau74ca5042013-06-11 23:12:07 +020014020be_sess_rate([<backend>]) : integer
14021 Returns an integer value corresponding to the sessions creation rate on the
14022 backend, in number of new sessions per second. This is used with ACLs to
14023 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010014024 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014025 online dictionary). It can also be useful to add this element to logs using a
14026 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014027
14028 Example :
14029 # Redirect to an error page if the dictionary is requested too often
14030 backend dynamic
14031 mode http
14032 acl being_scanned be_sess_rate gt 100
14033 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010014034
Davor Ocelice9ed2812017-12-25 17:49:28 +010014035bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014036 Returns a binary chain. The input is the hexadecimal representation
14037 of the string.
14038
14039bool(<bool>) : bool
14040 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14041 'false' and '0' are the same. 'true' and '1' are the same.
14042
Willy Tarreau74ca5042013-06-11 23:12:07 +020014043connslots([<backend>]) : integer
14044 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014045 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014046 connections on all servers and the maximum queue size. This is probably only
14047 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014048
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014049 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014050 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014051 usage; see "use_backend" keyword) can be redirected to a different backend.
14052
Willy Tarreau55165fe2009-05-10 12:02:55 +020014053 'connslots' = number of available server connection slots, + number of
14054 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014055
Willy Tarreaua36af912009-10-10 12:02:45 +020014056 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014057 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014058 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014059 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014060 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014061 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014062 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014063
Willy Tarreau55165fe2009-05-10 12:02:55 +020014064 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14065 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014066 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014067 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014068
Willy Tarreau70fe9442018-11-22 16:07:39 +010014069cpu_calls : integer
14070 Returns the number of calls to the task processing the stream or current
14071 request since it was allocated. This number is reset for each new request on
14072 the same connections in case of HTTP keep-alive. This value should usually be
14073 low and stable (around 2 calls for a typically simple request) but may become
14074 high if some processing (compression, caching or analysis) is performed. This
14075 is purely for performance monitoring purposes.
14076
14077cpu_ns_avg : integer
14078 Returns the average number of nanoseconds spent in each call to the task
14079 processing the stream or current request. This number is reset for each new
14080 request on the same connections in case of HTTP keep-alive. This value
14081 indicates the overall cost of processing the request or the connection for
14082 each call. There is no good nor bad value but the time spent in a call
14083 automatically causes latency for other processing (see lat_ns_avg below),
14084 and may affect other connection's apparent response time. Certain operations
14085 like compression, complex regex matching or heavy Lua operations may directly
14086 affect this value, and having it in the logs will make it easier to spot the
14087 faulty processing that needs to be fixed to recover decent performance.
14088 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14089
14090cpu_ns_tot : integer
14091 Returns the total number of nanoseconds spent in each call to the task
14092 processing the stream or current request. This number is reset for each new
14093 request on the same connections in case of HTTP keep-alive. This value
14094 indicates the overall cost of processing the request or the connection for
14095 each call. There is no good nor bad value but the time spent in a call
14096 automatically causes latency for other processing (see lat_ns_avg below),
14097 induces CPU costs on the machine, and may affect other connection's apparent
14098 response time. Certain operations like compression, complex regex matching or
14099 heavy Lua operations may directly affect this value, and having it in the
14100 logs will make it easier to spot the faulty processing that needs to be fixed
14101 to recover decent performance. The value may be artificially high due to a
14102 high cpu_calls count, for example when processing many HTTP chunks, and for
14103 this reason it is often preferred to log cpu_ns_avg instead.
14104
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014105date([<offset>]) : integer
14106 Returns the current date as the epoch (number of seconds since 01/01/1970).
14107 If an offset value is specified, then it is a number of seconds that is added
14108 to the current date before returning the value. This is particularly useful
14109 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014110 It is useful combined with the http_date converter.
14111
14112 Example :
14113
14114 # set an expires header to now+1 hour in every response
14115 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014116
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014117date_us : integer
14118 Return the microseconds part of the date (the "second" part is returned by
14119 date sample). This sample is coherent with the date sample as it is comes
14120 from the same timeval structure.
14121
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014122distcc_body(<token>[,<occ>]) : binary
14123 Parses a distcc message and returns the body associated to occurrence #<occ>
14124 of the token <token>. Occurrences start at 1, and when unspecified, any may
14125 match though in practice only the first one is checked for now. This can be
14126 used to extract file names or arguments in files built using distcc through
14127 haproxy. Please refer to distcc's protocol documentation for the complete
14128 list of supported tokens.
14129
14130distcc_param(<token>[,<occ>]) : integer
14131 Parses a distcc message and returns the parameter associated to occurrence
14132 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14133 any may match though in practice only the first one is checked for now. This
14134 can be used to extract certain information such as the protocol version, the
14135 file size or the argument in files built using distcc through haproxy.
14136 Another use case consists in waiting for the start of the preprocessed file
14137 contents before connecting to the server to avoid keeping idle connections.
14138 Please refer to distcc's protocol documentation for the complete list of
14139 supported tokens.
14140
14141 Example :
14142 # wait up to 20s for the pre-processed file to be uploaded
14143 tcp-request inspect-delay 20s
14144 tcp-request content accept if { distcc_param(DOTI) -m found }
14145 # send large files to the big farm
14146 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14147
Willy Tarreau595ec542013-06-12 21:34:28 +020014148env(<name>) : string
14149 Returns a string containing the value of environment variable <name>. As a
14150 reminder, environment variables are per-process and are sampled when the
14151 process starts. This can be useful to pass some information to a next hop
14152 server, or with ACLs to take specific action when the process is started a
14153 certain way.
14154
14155 Examples :
14156 # Pass the Via header to next hop with the local hostname in it
14157 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14158
14159 # reject cookie-less requests when the STOP environment variable is set
14160 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14161
Willy Tarreau74ca5042013-06-11 23:12:07 +020014162fe_conn([<frontend>]) : integer
14163 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014164 possibly including the connection being evaluated. If no frontend name is
14165 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014166 frontend. It can be used to return a sorry page before hard-blocking, or to
14167 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014168 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014169 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14170 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014171
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014172fe_req_rate([<frontend>]) : integer
14173 Returns an integer value corresponding to the number of HTTP requests per
14174 second sent to a frontend. This number can differ from "fe_sess_rate" in
14175 situations where client-side keep-alive is enabled.
14176
Willy Tarreau74ca5042013-06-11 23:12:07 +020014177fe_sess_rate([<frontend>]) : integer
14178 Returns an integer value corresponding to the sessions creation rate on the
14179 frontend, in number of new sessions per second. This is used with ACLs to
14180 limit the incoming session rate to an acceptable range in order to prevent
14181 abuse of service at the earliest moment, for example when combined with other
14182 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14183 down below the limit. It can also be useful to add this element to logs using
14184 a log-format directive. See also the "rate-limit sessions" directive for use
14185 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014186
14187 Example :
14188 # This frontend limits incoming mails to 10/s with a max of 100
14189 # concurrent connections. We accept any connection below 10/s, and
14190 # force excess clients to wait for 100 ms. Since clients are limited to
14191 # 100 max, there cannot be more than 10 incoming mails per second.
14192 frontend mail
14193 bind :25
14194 mode tcp
14195 maxconn 100
14196 acl too_fast fe_sess_rate ge 10
14197 tcp-request inspect-delay 100ms
14198 tcp-request content accept if ! too_fast
14199 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014200
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014201hostname : string
14202 Returns the system hostname.
14203
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014204int(<integer>) : signed integer
14205 Returns a signed integer.
14206
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014207ipv4(<ipv4>) : ipv4
14208 Returns an ipv4.
14209
14210ipv6(<ipv6>) : ipv6
14211 Returns an ipv6.
14212
Willy Tarreau70fe9442018-11-22 16:07:39 +010014213lat_ns_avg : integer
14214 Returns the average number of nanoseconds spent between the moment the task
14215 handling the stream is woken up and the moment it is effectively called. This
14216 number is reset for each new request on the same connections in case of HTTP
14217 keep-alive. This value indicates the overall latency inflicted to the current
14218 request by all other requests being processed in parallel, and is a direct
14219 indicator of perceived performance due to noisy neighbours. In order to keep
14220 the value low, it is possible to reduce the scheduler's run queue depth using
14221 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14222 once using "tune.maxpollevents", to decrease the stream's nice value using
14223 the "nice" option on the "bind" lines or in the frontend, or to look for
14224 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14225 whose processing needs to be adjusted or fixed. Compression of large buffers
14226 could be a culprit, like heavy regex or long lists of regex.
14227 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14228
14229lat_ns_tot : integer
14230 Returns the total number of nanoseconds spent between the moment the task
14231 handling the stream is woken up and the moment it is effectively called. This
14232 number is reset for each new request on the same connections in case of HTTP
14233 keep-alive. This value indicates the overall latency inflicted to the current
14234 request by all other requests being processed in parallel, and is a direct
14235 indicator of perceived performance due to noisy neighbours. In order to keep
14236 the value low, it is possible to reduce the scheduler's run queue depth using
14237 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14238 once using "tune.maxpollevents", to decrease the stream's nice value using
14239 the "nice" option on the "bind" lines or in the frontend, or to look for
14240 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14241 whose processing needs to be adjusted or fixed. Compression of large buffers
14242 could be a culprit, like heavy regex or long lists of regex. Note: while it
14243 may intuitively seem that the total latency adds to a transfer time, it is
14244 almost never true because while a task waits for the CPU, network buffers
14245 continue to fill up and the next call will process more at once. The value
14246 may be artificially high due to a high cpu_calls count, for example when
14247 processing many HTTP chunks, and for this reason it is often preferred to log
14248 lat_ns_avg instead, which is a more relevant performance indicator.
14249
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014250meth(<method>) : method
14251 Returns a method.
14252
Willy Tarreau0f30d262014-11-24 16:02:05 +010014253nbproc : integer
14254 Returns an integer value corresponding to the number of processes that were
14255 started (it equals the global "nbproc" setting). This is useful for logging
14256 and debugging purposes.
14257
Willy Tarreau74ca5042013-06-11 23:12:07 +020014258nbsrv([<backend>]) : integer
14259 Returns an integer value corresponding to the number of usable servers of
14260 either the current backend or the named backend. This is mostly used with
14261 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014262 switch to an alternate backend when the number of servers is too low to
14263 to handle some load. It is useful to report a failure when combined with
14264 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014265
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014266prio_class : integer
14267 Returns the priority class of the current session for http mode or connection
14268 for tcp mode. The value will be that set by the last call to "http-request
14269 set-priority-class" or "tcp-request content set-priority-class".
14270
14271prio_offset : integer
14272 Returns the priority offset of the current session for http mode or
14273 connection for tcp mode. The value will be that set by the last call to
14274 "http-request set-priority-offset" or "tcp-request content
14275 set-priority-offset".
14276
Willy Tarreau0f30d262014-11-24 16:02:05 +010014277proc : integer
14278 Returns an integer value corresponding to the position of the process calling
14279 the function, between 1 and global.nbproc. This is useful for logging and
14280 debugging purposes.
14281
Willy Tarreau74ca5042013-06-11 23:12:07 +020014282queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014283 Returns the total number of queued connections of the designated backend,
14284 including all the connections in server queues. If no backend name is
14285 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014286 one. This is useful with ACLs or to pass statistics to backend servers. This
14287 can be used to take actions when queuing goes above a known level, generally
14288 indicating a surge of traffic or a massive slowdown on the servers. One
14289 possible action could be to reject new users but still accept old ones. See
14290 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14291
Willy Tarreau84310e22014-02-14 11:59:04 +010014292rand([<range>]) : integer
14293 Returns a random integer value within a range of <range> possible values,
14294 starting at zero. If the range is not specified, it defaults to 2^32, which
14295 gives numbers between 0 and 4294967295. It can be useful to pass some values
14296 needed to take some routing decisions for example, or just for debugging
14297 purposes. This random must not be used for security purposes.
14298
Willy Tarreau74ca5042013-06-11 23:12:07 +020014299srv_conn([<backend>/]<server>) : integer
14300 Returns an integer value corresponding to the number of currently established
14301 connections on the designated server, possibly including the connection being
14302 evaluated. If <backend> is omitted, then the server is looked up in the
14303 current backend. It can be used to use a specific farm when one server is
14304 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014305 connections with it. See also the "fe_conn", "be_conn", "queue", and
14306 "srv_conn_free" fetch methods.
14307
14308srv_conn_free([<backend>/]<server>) : integer
14309 Returns an integer value corresponding to the number of available connections
14310 on the designated server, possibly including the connection being evaluated.
14311 The value does not include queue slots. If <backend> is omitted, then the
14312 server is looked up in the current backend. It can be used to use a specific
14313 farm when one server is full, or to inform the server about our view of the
14314 number of active connections with it. See also the "be_conn_free" and
14315 "srv_conn" fetch methods.
14316
14317 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14318 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014319
14320srv_is_up([<backend>/]<server>) : boolean
14321 Returns true when the designated server is UP, and false when it is either
14322 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14323 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014324 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014325 availability). Another possible use which is more of a hack consists in
14326 using dummy servers as boolean variables that can be enabled or disabled from
14327 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14328
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014329srv_queue([<backend>/]<server>) : integer
14330 Returns an integer value corresponding to the number of connections currently
14331 pending in the designated server's queue. If <backend> is omitted, then the
14332 server is looked up in the current backend. It can sometimes be used together
14333 with the "use-server" directive to force to use a known faster server when it
14334 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14335 fetch methods.
14336
Willy Tarreau74ca5042013-06-11 23:12:07 +020014337srv_sess_rate([<backend>/]<server>) : integer
14338 Returns an integer corresponding to the sessions creation rate on the
14339 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014340 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014341 used with ACLs but can make sense with logs too. This is used to switch to an
14342 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014343 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014344 overloading servers).
14345
14346 Example :
14347 # Redirect to a separate back
14348 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14349 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14350 use_backend be2 if srv1_full or srv2_full
14351
Willy Tarreau0f30d262014-11-24 16:02:05 +010014352stopping : boolean
14353 Returns TRUE if the process calling the function is currently stopping. This
14354 can be useful for logging, or for relaxing certain checks or helping close
14355 certain connections upon graceful shutdown.
14356
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014357str(<string>) : string
14358 Returns a string.
14359
Willy Tarreau74ca5042013-06-11 23:12:07 +020014360table_avl([<table>]) : integer
14361 Returns the total number of available entries in the current proxy's
14362 stick-table or in the designated stick-table. See also table_cnt.
14363
14364table_cnt([<table>]) : integer
14365 Returns the total number of entries currently in use in the current proxy's
14366 stick-table or in the designated stick-table. See also src_conn_cnt and
14367 table_avl for other entry counting methods.
14368
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014369thread : integer
14370 Returns an integer value corresponding to the position of the thread calling
14371 the function, between 0 and (global.nbthread-1). This is useful for logging
14372 and debugging purposes.
14373
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014374var(<var-name>) : undefined
14375 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014376 sample fetch fails. The name of the variable starts with an indication
14377 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014378 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014379 "sess" : the variable is shared with the whole session
14380 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014381 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014382 "req" : the variable is shared only during request processing,
14383 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014384 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014385 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014386
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200143877.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014388----------------------------------
14389
14390The layer 4 usually describes just the transport layer which in haproxy is
14391closest to the connection, where no content is yet made available. The fetch
14392methods described here are usable as low as the "tcp-request connection" rule
14393sets unless they require some future information. Those generally include
14394TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014395the incoming connection. For retrieving a value from a sticky counters, the
14396counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014397"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14398used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14399can be specified as the first integer argument when using the "sc_" prefix.
14400Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14401table may be specified with the "sc*" form, in which case the currently
14402tracked key will be looked up into this alternate table instead of the table
14403currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014404
Jérôme Magnin35e53a62019-01-16 14:38:37 +010014405bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010014406 Returns the backend connection's HTTP major version encoding, which may be 1
14407 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14408 encoding and not the version present in the request header.
14409
Willy Tarreau74ca5042013-06-11 23:12:07 +020014410be_id : integer
14411 Returns an integer containing the current backend's id. It can be used in
14412 frontends with responses to check which backend processed the request.
14413
Marcin Deranekd2471c22016-12-12 14:08:05 +010014414be_name : string
14415 Returns a string containing the current backend's name. It can be used in
14416 frontends with responses to check which backend processed the request.
14417
Willy Tarreau74ca5042013-06-11 23:12:07 +020014418dst : ip
14419 This is the destination IPv4 address of the connection on the client side,
14420 which is the address the client connected to. It can be useful when running
14421 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14422 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014423 RFC 4291. When the incoming connection passed through address translation or
14424 redirection involving connection tracking, the original destination address
14425 before the redirection will be reported. On Linux systems, the source and
14426 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
14427 is set, because a late response may reopen a timed out connection and switch
14428 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014429
14430dst_conn : integer
14431 Returns an integer value corresponding to the number of currently established
14432 connections on the same socket including the one being evaluated. It is
14433 normally used with ACLs but can as well be used to pass the information to
14434 servers in an HTTP header or in logs. It can be used to either return a sorry
14435 page before hard-blocking, or to use a specific backend to drain new requests
14436 when the socket is considered saturated. This offers the ability to assign
14437 different limits to different listening ports or addresses. See also the
14438 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014439
Willy Tarreau16e01562016-08-09 16:46:18 +020014440dst_is_local : boolean
14441 Returns true if the destination address of the incoming connection is local
14442 to the system, or false if the address doesn't exist on the system, meaning
14443 that it was intercepted in transparent mode. It can be useful to apply
14444 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014445 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014446 be delivered only on this address, or SSH access could be locally redirected.
14447 Please note that the check involves a few system calls, so it's better to do
14448 it only once per connection.
14449
Willy Tarreau74ca5042013-06-11 23:12:07 +020014450dst_port : integer
14451 Returns an integer value corresponding to the destination TCP port of the
14452 connection on the client side, which is the port the client connected to.
14453 This might be used when running in transparent mode, when assigning dynamic
14454 ports to some clients for a whole application session, to stick all users to
14455 a same server, or to pass the destination port information to a server using
14456 an HTTP header.
14457
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014458fc_http_major : integer
14459 Reports the front connection's HTTP major version encoding, which may be 1
14460 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14461 encoding and not on the version present in the request header.
14462
Emeric Brun4f603012017-01-05 15:11:44 +010014463fc_rcvd_proxy : boolean
14464 Returns true if the client initiated the connection with a PROXY protocol
14465 header.
14466
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014467fc_rtt(<unit>) : integer
14468 Returns the Round Trip Time (RTT) measured by the kernel for the client
14469 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14470 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14471 connection is not established, if the connection is not TCP or if the
14472 operating system does not support TCP_INFO, for example Linux kernels before
14473 2.4, the sample fetch fails.
14474
14475fc_rttvar(<unit>) : integer
14476 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14477 client connection. <unit> is facultative, by default the unit is milliseconds.
14478 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14479 server connection is not established, if the connection is not TCP or if the
14480 operating system does not support TCP_INFO, for example Linux kernels before
14481 2.4, the sample fetch fails.
14482
Joe Williams30fcd392016-08-10 07:06:44 -070014483fc_unacked(<unit>) : integer
14484 Returns the unacked counter measured by the kernel for the client connection.
14485 If the server connection is not established, if the connection is not TCP or
14486 if the operating system does not support TCP_INFO, for example Linux kernels
14487 before 2.4, the sample fetch fails.
14488
14489fc_sacked(<unit>) : integer
14490 Returns the sacked counter measured by the kernel for the client connection.
14491 If the server connection is not established, if the connection is not TCP or
14492 if the operating system does not support TCP_INFO, for example Linux kernels
14493 before 2.4, the sample fetch fails.
14494
14495fc_retrans(<unit>) : integer
14496 Returns the retransmits counter measured by the kernel for the client
14497 connection. If the server connection is not established, if the connection is
14498 not TCP or if the operating system does not support TCP_INFO, for example
14499 Linux kernels before 2.4, the sample fetch fails.
14500
14501fc_fackets(<unit>) : integer
14502 Returns the fack counter measured by the kernel for the client
14503 connection. If the server connection is not established, if the connection is
14504 not TCP or if the operating system does not support TCP_INFO, for example
14505 Linux kernels before 2.4, the sample fetch fails.
14506
14507fc_lost(<unit>) : integer
14508 Returns the lost counter measured by the kernel for the client
14509 connection. If the server connection is not established, if the connection is
14510 not TCP or if the operating system does not support TCP_INFO, for example
14511 Linux kernels before 2.4, the sample fetch fails.
14512
14513fc_reordering(<unit>) : integer
14514 Returns the reordering counter measured by the kernel for the client
14515 connection. If the server connection is not established, if the connection is
14516 not TCP or if the operating system does not support TCP_INFO, for example
14517 Linux kernels before 2.4, the sample fetch fails.
14518
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014519fe_defbe : string
14520 Returns a string containing the frontend's default backend name. It can be
14521 used in frontends to check which backend will handle requests by default.
14522
Willy Tarreau74ca5042013-06-11 23:12:07 +020014523fe_id : integer
14524 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014525 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014526 coming via a same frontend to the same server.
14527
Marcin Deranekd2471c22016-12-12 14:08:05 +010014528fe_name : string
14529 Returns a string containing the current frontend's name. It can be used in
14530 backends to check from which frontend it was called, or to stick all users
14531 coming via a same frontend to the same server.
14532
Cyril Bonté62ba8702014-04-22 23:52:25 +020014533sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014534sc0_bytes_in_rate([<table>]) : integer
14535sc1_bytes_in_rate([<table>]) : integer
14536sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014537 Returns the average client-to-server bytes rate from the currently tracked
14538 counters, measured in amount of bytes over the period configured in the
14539 table. See also src_bytes_in_rate.
14540
Cyril Bonté62ba8702014-04-22 23:52:25 +020014541sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014542sc0_bytes_out_rate([<table>]) : integer
14543sc1_bytes_out_rate([<table>]) : integer
14544sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014545 Returns the average server-to-client bytes rate from the currently tracked
14546 counters, measured in amount of bytes over the period configured in the
14547 table. See also src_bytes_out_rate.
14548
Cyril Bonté62ba8702014-04-22 23:52:25 +020014549sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014550sc0_clr_gpc0([<table>]) : integer
14551sc1_clr_gpc0([<table>]) : integer
14552sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014553 Clears the first General Purpose Counter associated to the currently tracked
14554 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014555 stored value is zero, so first invocation will always return zero. This is
14556 typically used as a second ACL in an expression in order to mark a connection
14557 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014558
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014559 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014560 # block if 5 consecutive requests continue to come faster than 10 sess
14561 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014562 acl abuse sc0_http_req_rate gt 10
14563 acl kill sc0_inc_gpc0 gt 5
14564 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014565 tcp-request connection accept if !abuse save
14566 tcp-request connection reject if abuse kill
14567
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014568sc_clr_gpc1(<ctr>[,<table>]) : integer
14569sc0_clr_gpc1([<table>]) : integer
14570sc1_clr_gpc1([<table>]) : integer
14571sc2_clr_gpc1([<table>]) : integer
14572 Clears the second General Purpose Counter associated to the currently tracked
14573 counters, and returns its previous value. Before the first invocation, the
14574 stored value is zero, so first invocation will always return zero. This is
14575 typically used as a second ACL in an expression in order to mark a connection
14576 when a first ACL was verified.
14577
Cyril Bonté62ba8702014-04-22 23:52:25 +020014578sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014579sc0_conn_cnt([<table>]) : integer
14580sc1_conn_cnt([<table>]) : integer
14581sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014582 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014583 counters. See also src_conn_cnt.
14584
Cyril Bonté62ba8702014-04-22 23:52:25 +020014585sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014586sc0_conn_cur([<table>]) : integer
14587sc1_conn_cur([<table>]) : integer
14588sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014589 Returns the current amount of concurrent connections tracking the same
14590 tracked counters. This number is automatically incremented when tracking
14591 begins and decremented when tracking stops. See also src_conn_cur.
14592
Cyril Bonté62ba8702014-04-22 23:52:25 +020014593sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014594sc0_conn_rate([<table>]) : integer
14595sc1_conn_rate([<table>]) : integer
14596sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014597 Returns the average connection rate from the currently tracked counters,
14598 measured in amount of connections over the period configured in the table.
14599 See also src_conn_rate.
14600
Cyril Bonté62ba8702014-04-22 23:52:25 +020014601sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014602sc0_get_gpc0([<table>]) : integer
14603sc1_get_gpc0([<table>]) : integer
14604sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014605 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014606 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014607
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014608sc_get_gpc1(<ctr>[,<table>]) : integer
14609sc0_get_gpc1([<table>]) : integer
14610sc1_get_gpc1([<table>]) : integer
14611sc2_get_gpc1([<table>]) : integer
14612 Returns the value of the second General Purpose Counter associated to the
14613 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14614
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014615sc_get_gpt0(<ctr>[,<table>]) : integer
14616sc0_get_gpt0([<table>]) : integer
14617sc1_get_gpt0([<table>]) : integer
14618sc2_get_gpt0([<table>]) : integer
14619 Returns the value of the first General Purpose Tag associated to the
14620 currently tracked counters. See also src_get_gpt0.
14621
Cyril Bonté62ba8702014-04-22 23:52:25 +020014622sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014623sc0_gpc0_rate([<table>]) : integer
14624sc1_gpc0_rate([<table>]) : integer
14625sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014626 Returns the average increment rate of the first General Purpose Counter
14627 associated to the currently tracked counters. It reports the frequency
14628 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014629 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14630 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14631 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014632
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014633sc_gpc1_rate(<ctr>[,<table>]) : integer
14634sc0_gpc1_rate([<table>]) : integer
14635sc1_gpc1_rate([<table>]) : integer
14636sc2_gpc1_rate([<table>]) : integer
14637 Returns the average increment rate of the second General Purpose Counter
14638 associated to the currently tracked counters. It reports the frequency
14639 which the gpc1 counter was incremented over the configured period. See also
14640 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14641 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14642 be returned, as "gpc1" only holds the event count.
14643
Cyril Bonté62ba8702014-04-22 23:52:25 +020014644sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014645sc0_http_err_cnt([<table>]) : integer
14646sc1_http_err_cnt([<table>]) : integer
14647sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014648 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014649 counters. This includes the both request errors and 4xx error responses.
14650 See also src_http_err_cnt.
14651
Cyril Bonté62ba8702014-04-22 23:52:25 +020014652sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014653sc0_http_err_rate([<table>]) : integer
14654sc1_http_err_rate([<table>]) : integer
14655sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014656 Returns the average rate of HTTP errors from the currently tracked counters,
14657 measured in amount of errors over the period configured in the table. This
14658 includes the both request errors and 4xx error responses. See also
14659 src_http_err_rate.
14660
Cyril Bonté62ba8702014-04-22 23:52:25 +020014661sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014662sc0_http_req_cnt([<table>]) : integer
14663sc1_http_req_cnt([<table>]) : integer
14664sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014665 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014666 counters. This includes every started request, valid or not. See also
14667 src_http_req_cnt.
14668
Cyril Bonté62ba8702014-04-22 23:52:25 +020014669sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014670sc0_http_req_rate([<table>]) : integer
14671sc1_http_req_rate([<table>]) : integer
14672sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014673 Returns the average rate of HTTP requests from the currently tracked
14674 counters, measured in amount of requests over the period configured in
14675 the table. This includes every started request, valid or not. See also
14676 src_http_req_rate.
14677
Cyril Bonté62ba8702014-04-22 23:52:25 +020014678sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014679sc0_inc_gpc0([<table>]) : integer
14680sc1_inc_gpc0([<table>]) : integer
14681sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014682 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014683 tracked counters, and returns its new value. Before the first invocation,
14684 the stored value is zero, so first invocation will increase it to 1 and will
14685 return 1. This is typically used as a second ACL in an expression in order
14686 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014687
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014688 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014689 acl abuse sc0_http_req_rate gt 10
14690 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014691 tcp-request connection reject if abuse kill
14692
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014693sc_inc_gpc1(<ctr>[,<table>]) : integer
14694sc0_inc_gpc1([<table>]) : integer
14695sc1_inc_gpc1([<table>]) : integer
14696sc2_inc_gpc1([<table>]) : integer
14697 Increments the second General Purpose Counter associated to the currently
14698 tracked counters, and returns its new value. Before the first invocation,
14699 the stored value is zero, so first invocation will increase it to 1 and will
14700 return 1. This is typically used as a second ACL in an expression in order
14701 to mark a connection when a first ACL was verified.
14702
Cyril Bonté62ba8702014-04-22 23:52:25 +020014703sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014704sc0_kbytes_in([<table>]) : integer
14705sc1_kbytes_in([<table>]) : integer
14706sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014707 Returns the total amount of client-to-server data from the currently tracked
14708 counters, measured in kilobytes. The test is currently performed on 32-bit
14709 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014710
Cyril Bonté62ba8702014-04-22 23:52:25 +020014711sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014712sc0_kbytes_out([<table>]) : integer
14713sc1_kbytes_out([<table>]) : integer
14714sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014715 Returns the total amount of server-to-client data from the currently tracked
14716 counters, measured in kilobytes. The test is currently performed on 32-bit
14717 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014718
Cyril Bonté62ba8702014-04-22 23:52:25 +020014719sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014720sc0_sess_cnt([<table>]) : integer
14721sc1_sess_cnt([<table>]) : integer
14722sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014723 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014724 into sessions, which means that they were accepted by a "tcp-request
14725 connection" rule, from the currently tracked counters. A backend may count
14726 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014727 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014728 with the client. See also src_sess_cnt.
14729
Cyril Bonté62ba8702014-04-22 23:52:25 +020014730sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014731sc0_sess_rate([<table>]) : integer
14732sc1_sess_rate([<table>]) : integer
14733sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014734 Returns the average session rate from the currently tracked counters,
14735 measured in amount of sessions over the period configured in the table. A
14736 session is a connection that got past the early "tcp-request connection"
14737 rules. A backend may count more sessions than connections because each
14738 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014739 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014740
Cyril Bonté62ba8702014-04-22 23:52:25 +020014741sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014742sc0_tracked([<table>]) : boolean
14743sc1_tracked([<table>]) : boolean
14744sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014745 Returns true if the designated session counter is currently being tracked by
14746 the current session. This can be useful when deciding whether or not we want
14747 to set some values in a header passed to the server.
14748
Cyril Bonté62ba8702014-04-22 23:52:25 +020014749sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014750sc0_trackers([<table>]) : integer
14751sc1_trackers([<table>]) : integer
14752sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014753 Returns the current amount of concurrent connections tracking the same
14754 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014755 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014756 that it does not rely on any stored information but on the table's reference
14757 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014758 may sometimes be more suited for layer7 tracking. It can be used to tell a
14759 server how many concurrent connections there are from a given address for
14760 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014761
Willy Tarreau74ca5042013-06-11 23:12:07 +020014762so_id : integer
14763 Returns an integer containing the current listening socket's id. It is useful
14764 in frontends involving many "bind" lines, or to stick all users coming via a
14765 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014766
Willy Tarreau74ca5042013-06-11 23:12:07 +020014767src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014768 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014769 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14770 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14771 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014772 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14773 directive is used, it can be the address of a client behind another
14774 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014775 "tcp-request connection" which sees the real address. When the incoming
14776 connection passed through address translation or redirection involving
14777 connection tracking, the original destination address before the redirection
14778 will be reported. On Linux systems, the source and destination may seldom
14779 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
14780 response may reopen a timed out connection and switch what is believed to be
14781 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014782
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014783 Example:
14784 # add an HTTP header in requests with the originating address' country
14785 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14786
Willy Tarreau74ca5042013-06-11 23:12:07 +020014787src_bytes_in_rate([<table>]) : integer
14788 Returns the average bytes rate from the incoming connection's source address
14789 in the current proxy's stick-table or in the designated stick-table, measured
14790 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014791 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014792
Willy Tarreau74ca5042013-06-11 23:12:07 +020014793src_bytes_out_rate([<table>]) : integer
14794 Returns the average bytes rate to the incoming connection's source address in
14795 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014796 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014797 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014798
Willy Tarreau74ca5042013-06-11 23:12:07 +020014799src_clr_gpc0([<table>]) : integer
14800 Clears the first General Purpose Counter associated to the incoming
14801 connection's source address in the current proxy's stick-table or in the
14802 designated stick-table, and returns its previous value. If the address is not
14803 found, an entry is created and 0 is returned. This is typically used as a
14804 second ACL in an expression in order to mark a connection when a first ACL
14805 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014806
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014807 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014808 # block if 5 consecutive requests continue to come faster than 10 sess
14809 # per second, and reset the counter as soon as the traffic slows down.
14810 acl abuse src_http_req_rate gt 10
14811 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014812 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014813 tcp-request connection accept if !abuse save
14814 tcp-request connection reject if abuse kill
14815
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014816src_clr_gpc1([<table>]) : integer
14817 Clears the second General Purpose Counter associated to the incoming
14818 connection's source address in the current proxy's stick-table or in the
14819 designated stick-table, and returns its previous value. If the address is not
14820 found, an entry is created and 0 is returned. This is typically used as a
14821 second ACL in an expression in order to mark a connection when a first ACL
14822 was verified.
14823
Willy Tarreau74ca5042013-06-11 23:12:07 +020014824src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014825 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014826 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014827 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014828 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014829
Willy Tarreau74ca5042013-06-11 23:12:07 +020014830src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014831 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014832 current incoming connection's source address in the current proxy's
14833 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014834 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014835
Willy Tarreau74ca5042013-06-11 23:12:07 +020014836src_conn_rate([<table>]) : integer
14837 Returns the average connection rate from the incoming connection's source
14838 address in the current proxy's stick-table or in the designated stick-table,
14839 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014840 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014841
Willy Tarreau74ca5042013-06-11 23:12:07 +020014842src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014843 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014844 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014845 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014846 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014847
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014848src_get_gpc1([<table>]) : integer
14849 Returns the value of the second General Purpose Counter associated to the
14850 incoming connection's source address in the current proxy's stick-table or in
14851 the designated stick-table. If the address is not found, zero is returned.
14852 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14853
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014854src_get_gpt0([<table>]) : integer
14855 Returns the value of the first General Purpose Tag associated to the
14856 incoming connection's source address in the current proxy's stick-table or in
14857 the designated stick-table. If the address is not found, zero is returned.
14858 See also sc/sc0/sc1/sc2_get_gpt0.
14859
Willy Tarreau74ca5042013-06-11 23:12:07 +020014860src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014861 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014862 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014863 stick-table or in the designated stick-table. It reports the frequency
14864 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014865 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14866 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14867 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014868
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014869src_gpc1_rate([<table>]) : integer
14870 Returns the average increment rate of the second General Purpose Counter
14871 associated to the incoming connection's source address in the current proxy's
14872 stick-table or in the designated stick-table. It reports the frequency
14873 which the gpc1 counter was incremented over the configured period. See also
14874 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14875 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14876 be returned, as "gpc1" only holds the event count.
14877
Willy Tarreau74ca5042013-06-11 23:12:07 +020014878src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014879 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014880 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014881 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014882 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014883 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014884
Willy Tarreau74ca5042013-06-11 23:12:07 +020014885src_http_err_rate([<table>]) : integer
14886 Returns the average rate of HTTP errors from the incoming connection's source
14887 address in the current proxy's stick-table or in the designated stick-table,
14888 measured in amount of errors over the period configured in the table. This
14889 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014890 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014891
Willy Tarreau74ca5042013-06-11 23:12:07 +020014892src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014893 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014894 source address in the current proxy's stick-table or in the designated stick-
14895 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014896 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014897
Willy Tarreau74ca5042013-06-11 23:12:07 +020014898src_http_req_rate([<table>]) : integer
14899 Returns the average rate of HTTP requests from the incoming connection's
14900 source address in the current proxy's stick-table or in the designated stick-
14901 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014902 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014903 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014904
Willy Tarreau74ca5042013-06-11 23:12:07 +020014905src_inc_gpc0([<table>]) : integer
14906 Increments the first General Purpose Counter associated to the incoming
14907 connection's source address in the current proxy's stick-table or in the
14908 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014909 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014910 This is typically used as a second ACL in an expression in order to mark a
14911 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014912
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014913 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014914 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014915 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014916 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014917
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014918src_inc_gpc1([<table>]) : integer
14919 Increments the second General Purpose Counter associated to the incoming
14920 connection's source address in the current proxy's stick-table or in the
14921 designated stick-table, and returns its new value. If the address is not
14922 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14923 This is typically used as a second ACL in an expression in order to mark a
14924 connection when a first ACL was verified.
14925
Willy Tarreau16e01562016-08-09 16:46:18 +020014926src_is_local : boolean
14927 Returns true if the source address of the incoming connection is local to the
14928 system, or false if the address doesn't exist on the system, meaning that it
14929 comes from a remote machine. Note that UNIX addresses are considered local.
14930 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014931 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014932 note that the check involves a few system calls, so it's better to do it only
14933 once per connection.
14934
Willy Tarreau74ca5042013-06-11 23:12:07 +020014935src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014936 Returns the total amount of data received from the incoming connection's
14937 source address in the current proxy's stick-table or in the designated
14938 stick-table, measured in kilobytes. If the address is not found, zero is
14939 returned. The test is currently performed on 32-bit integers, which limits
14940 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014941
Willy Tarreau74ca5042013-06-11 23:12:07 +020014942src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014943 Returns the total amount of data sent to the incoming connection's source
14944 address in the current proxy's stick-table or in the designated stick-table,
14945 measured in kilobytes. If the address is not found, zero is returned. The
14946 test is currently performed on 32-bit integers, which limits values to 4
14947 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014948
Willy Tarreau74ca5042013-06-11 23:12:07 +020014949src_port : integer
14950 Returns an integer value corresponding to the TCP source port of the
14951 connection on the client side, which is the port the client connected from.
14952 Usage of this function is very limited as modern protocols do not care much
14953 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014954
Willy Tarreau74ca5042013-06-11 23:12:07 +020014955src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014956 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020014957 connection's source IPv4 address in the current proxy's stick-table or in the
14958 designated stick-table, that were transformed into sessions, which means that
14959 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014960 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014961
Willy Tarreau74ca5042013-06-11 23:12:07 +020014962src_sess_rate([<table>]) : integer
14963 Returns the average session rate from the incoming connection's source
14964 address in the current proxy's stick-table or in the designated stick-table,
14965 measured in amount of sessions over the period configured in the table. A
14966 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014967 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014968
Willy Tarreau74ca5042013-06-11 23:12:07 +020014969src_updt_conn_cnt([<table>]) : integer
14970 Creates or updates the entry associated to the incoming connection's source
14971 address in the current proxy's stick-table or in the designated stick-table.
14972 This table must be configured to store the "conn_cnt" data type, otherwise
14973 the match will be ignored. The current count is incremented by one, and the
14974 expiration timer refreshed. The updated count is returned, so this match
14975 can't return zero. This was used to reject service abusers based on their
14976 source address. Note: it is recommended to use the more complete "track-sc*"
14977 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014978
14979 Example :
14980 # This frontend limits incoming SSH connections to 3 per 10 second for
14981 # each source address, and rejects excess connections until a 10 second
14982 # silence is observed. At most 20 addresses are tracked.
14983 listen ssh
14984 bind :22
14985 mode tcp
14986 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020014987 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020014988 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014989 server local 127.0.0.1:22
14990
Willy Tarreau74ca5042013-06-11 23:12:07 +020014991srv_id : integer
14992 Returns an integer containing the server's id when processing the response.
14993 While it's almost only used with ACLs, it may be used for logging or
14994 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020014995
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200149967.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020014997----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020014998
Willy Tarreau74ca5042013-06-11 23:12:07 +020014999The layer 5 usually describes just the session layer which in haproxy is
15000closest to the session once all the connection handshakes are finished, but
15001when no content is yet made available. The fetch methods described here are
15002usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015003future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020015004
Ben Shillitof25e8e52016-12-02 14:25:37 +00001500551d.all(<prop>[,<prop>*]) : string
15006 Returns values for the properties requested as a string, where values are
15007 separated by the delimiter specified with "51degrees-property-separator".
15008 The device is identified using all the important HTTP headers from the
15009 request. The function can be passed up to five property names, and if a
15010 property name can't be found, the value "NoData" is returned.
15011
15012 Example :
15013 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
15014 # containing the three properties requested using all relevant headers from
15015 # the request.
15016 frontend http-in
15017 bind *:8081
15018 default_backend servers
15019 http-request set-header X-51D-DeviceTypeMobileTablet \
15020 %[51d.all(DeviceType,IsMobile,IsTablet)]
15021
Emeric Brun645ae792014-04-30 14:21:06 +020015022ssl_bc : boolean
15023 Returns true when the back connection was made via an SSL/TLS transport
15024 layer and is locally deciphered. This means the outgoing connection was made
15025 other a server with the "ssl" option.
15026
15027ssl_bc_alg_keysize : integer
15028 Returns the symmetric cipher key size supported in bits when the outgoing
15029 connection was made over an SSL/TLS transport layer.
15030
Olivier Houchard6b77f492018-11-22 18:18:29 +010015031ssl_bc_alpn : string
15032 This extracts the Application Layer Protocol Negotiation field from an
15033 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020015034 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010015035 server. The SSL library must have been built with support for TLS
15036 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15037 not advertised unless the "alpn" keyword on the "server" line specifies a
15038 protocol list. Also, nothing forces the server to pick a protocol from this
15039 list, any other one may be requested. The TLS ALPN extension is meant to
15040 replace the TLS NPN extension. See also "ssl_bc_npn".
15041
Emeric Brun645ae792014-04-30 14:21:06 +020015042ssl_bc_cipher : string
15043 Returns the name of the used cipher when the outgoing connection was made
15044 over an SSL/TLS transport layer.
15045
Patrick Hemmer65674662019-06-04 08:13:03 -040015046ssl_bc_client_random : binary
15047 Returns the client random of the back connection when the incoming connection
15048 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15049 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15050
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015051ssl_bc_is_resumed : boolean
15052 Returns true when the back connection was made over an SSL/TLS transport
15053 layer and the newly created SSL session was resumed using a cached
15054 session or a TLS ticket.
15055
Olivier Houchard6b77f492018-11-22 18:18:29 +010015056ssl_bc_npn : string
15057 This extracts the Next Protocol Negotiation field from an outgoing connection
15058 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020015059 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010015060 built with support for TLS extensions enabled (check haproxy -vv). Note that
15061 the TLS NPN extension is not advertised unless the "npn" keyword on the
15062 "server" line specifies a protocol list. Also, nothing forces the server to
15063 pick a protocol from this list, any other one may be used. Please note that
15064 the TLS NPN extension was replaced with ALPN.
15065
Emeric Brun645ae792014-04-30 14:21:06 +020015066ssl_bc_protocol : string
15067 Returns the name of the used protocol when the outgoing connection was made
15068 over an SSL/TLS transport layer.
15069
Emeric Brunb73a9b02014-04-30 18:49:19 +020015070ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015071 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015072 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15073 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015074
Patrick Hemmer65674662019-06-04 08:13:03 -040015075ssl_bc_server_random : binary
15076 Returns the server random of the back connection when the incoming connection
15077 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15078 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15079
Emeric Brun645ae792014-04-30 14:21:06 +020015080ssl_bc_session_id : binary
15081 Returns the SSL ID of the back connection when the outgoing connection was
15082 made over an SSL/TLS transport layer. It is useful to log if we want to know
15083 if session was reused or not.
15084
Patrick Hemmere0275472018-04-28 19:15:51 -040015085ssl_bc_session_key : binary
15086 Returns the SSL session master key of the back connection when the outgoing
15087 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15088 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15089 BoringSSL.
15090
Emeric Brun645ae792014-04-30 14:21:06 +020015091ssl_bc_use_keysize : integer
15092 Returns the symmetric cipher key size used in bits when the outgoing
15093 connection was made over an SSL/TLS transport layer.
15094
Willy Tarreau74ca5042013-06-11 23:12:07 +020015095ssl_c_ca_err : integer
15096 When the incoming connection was made over an SSL/TLS transport layer,
15097 returns the ID of the first error detected during verification of the client
15098 certificate at depth > 0, or 0 if no error was encountered during this
15099 verification process. Please refer to your SSL library's documentation to
15100 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015101
Willy Tarreau74ca5042013-06-11 23:12:07 +020015102ssl_c_ca_err_depth : integer
15103 When the incoming connection was made over an SSL/TLS transport layer,
15104 returns the depth in the CA chain of the first error detected during the
15105 verification of the client certificate. If no error is encountered, 0 is
15106 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015107
Emeric Brun43e79582014-10-29 19:03:26 +010015108ssl_c_der : binary
15109 Returns the DER formatted certificate presented by the client when the
15110 incoming connection was made over an SSL/TLS transport layer. When used for
15111 an ACL, the value(s) to match against can be passed in hexadecimal form.
15112
Willy Tarreau74ca5042013-06-11 23:12:07 +020015113ssl_c_err : integer
15114 When the incoming connection was made over an SSL/TLS transport layer,
15115 returns the ID of the first error detected during verification at depth 0, or
15116 0 if no error was encountered during this verification process. Please refer
15117 to your SSL library's documentation to find the exhaustive list of error
15118 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015119
Willy Tarreau74ca5042013-06-11 23:12:07 +020015120ssl_c_i_dn([<entry>[,<occ>]]) : string
15121 When the incoming connection was made over an SSL/TLS transport layer,
15122 returns the full distinguished name of the issuer of the certificate
15123 presented by the client when no <entry> is specified, or the value of the
15124 first given entry found from the beginning of the DN. If a positive/negative
15125 occurrence number is specified as the optional second argument, it returns
15126 the value of the nth given entry value from the beginning/end of the DN.
15127 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15128 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020015129
Willy Tarreau74ca5042013-06-11 23:12:07 +020015130ssl_c_key_alg : string
15131 Returns the name of the algorithm used to generate the key of the certificate
15132 presented by the client when the incoming connection was made over an SSL/TLS
15133 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015134
Willy Tarreau74ca5042013-06-11 23:12:07 +020015135ssl_c_notafter : string
15136 Returns the end date presented by the client as a formatted string
15137 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15138 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015139
Willy Tarreau74ca5042013-06-11 23:12:07 +020015140ssl_c_notbefore : string
15141 Returns the start date presented by the client as a formatted string
15142 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15143 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015144
Willy Tarreau74ca5042013-06-11 23:12:07 +020015145ssl_c_s_dn([<entry>[,<occ>]]) : string
15146 When the incoming connection was made over an SSL/TLS transport layer,
15147 returns the full distinguished name of the subject of the certificate
15148 presented by the client when no <entry> is specified, or the value of the
15149 first given entry found from the beginning of the DN. If a positive/negative
15150 occurrence number is specified as the optional second argument, it returns
15151 the value of the nth given entry value from the beginning/end of the DN.
15152 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15153 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015154
Willy Tarreau74ca5042013-06-11 23:12:07 +020015155ssl_c_serial : binary
15156 Returns the serial of the certificate presented by the client when the
15157 incoming connection was made over an SSL/TLS transport layer. When used for
15158 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015159
Willy Tarreau74ca5042013-06-11 23:12:07 +020015160ssl_c_sha1 : binary
15161 Returns the SHA-1 fingerprint of the certificate presented by the client when
15162 the incoming connection was made over an SSL/TLS transport layer. This can be
15163 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015164 Note that the output is binary, so if you want to pass that signature to the
15165 server, you need to encode it in hex or base64, such as in the example below:
15166
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015167 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015168 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015169
Willy Tarreau74ca5042013-06-11 23:12:07 +020015170ssl_c_sig_alg : string
15171 Returns the name of the algorithm used to sign the certificate presented by
15172 the client when the incoming connection was made over an SSL/TLS transport
15173 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015174
Willy Tarreau74ca5042013-06-11 23:12:07 +020015175ssl_c_used : boolean
15176 Returns true if current SSL session uses a client certificate even if current
15177 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015178
Willy Tarreau74ca5042013-06-11 23:12:07 +020015179ssl_c_verify : integer
15180 Returns the verify result error ID when the incoming connection was made over
15181 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15182 refer to your SSL library's documentation for an exhaustive list of error
15183 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015184
Willy Tarreau74ca5042013-06-11 23:12:07 +020015185ssl_c_version : integer
15186 Returns the version of the certificate presented by the client when the
15187 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015188
Emeric Brun43e79582014-10-29 19:03:26 +010015189ssl_f_der : binary
15190 Returns the DER formatted certificate presented by the frontend when the
15191 incoming connection was made over an SSL/TLS transport layer. When used for
15192 an ACL, the value(s) to match against can be passed in hexadecimal form.
15193
Willy Tarreau74ca5042013-06-11 23:12:07 +020015194ssl_f_i_dn([<entry>[,<occ>]]) : string
15195 When the incoming connection was made over an SSL/TLS transport layer,
15196 returns the full distinguished name of the issuer of the certificate
15197 presented by the frontend when no <entry> is specified, or the value of the
15198 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015199 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015200 the value of the nth given entry value from the beginning/end of the DN.
15201 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15202 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020015203
Willy Tarreau74ca5042013-06-11 23:12:07 +020015204ssl_f_key_alg : string
15205 Returns the name of the algorithm used to generate the key of the certificate
15206 presented by the frontend when the incoming connection was made over an
15207 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015208
Willy Tarreau74ca5042013-06-11 23:12:07 +020015209ssl_f_notafter : string
15210 Returns the end date presented by the frontend as a formatted string
15211 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15212 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015213
Willy Tarreau74ca5042013-06-11 23:12:07 +020015214ssl_f_notbefore : string
15215 Returns the start date presented by the frontend as a formatted string
15216 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15217 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015218
Willy Tarreau74ca5042013-06-11 23:12:07 +020015219ssl_f_s_dn([<entry>[,<occ>]]) : string
15220 When the incoming connection was made over an SSL/TLS transport layer,
15221 returns the full distinguished name of the subject of the certificate
15222 presented by the frontend when no <entry> is specified, or the value of the
15223 first given entry found from the beginning of the DN. If a positive/negative
15224 occurrence number is specified as the optional second argument, it returns
15225 the value of the nth given entry value from the beginning/end of the DN.
15226 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15227 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020015228
Willy Tarreau74ca5042013-06-11 23:12:07 +020015229ssl_f_serial : binary
15230 Returns the serial of the certificate presented by the frontend when the
15231 incoming connection was made over an SSL/TLS transport layer. When used for
15232 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015233
Emeric Brun55f4fa82014-04-30 17:11:25 +020015234ssl_f_sha1 : binary
15235 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15236 when the incoming connection was made over an SSL/TLS transport layer. This
15237 can be used to know which certificate was chosen using SNI.
15238
Willy Tarreau74ca5042013-06-11 23:12:07 +020015239ssl_f_sig_alg : string
15240 Returns the name of the algorithm used to sign the certificate presented by
15241 the frontend when the incoming connection was made over an SSL/TLS transport
15242 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015243
Willy Tarreau74ca5042013-06-11 23:12:07 +020015244ssl_f_version : integer
15245 Returns the version of the certificate presented by the frontend when the
15246 incoming connection was made over an SSL/TLS transport layer.
15247
15248ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015249 Returns true when the front connection was made via an SSL/TLS transport
15250 layer and is locally deciphered. This means it has matched a socket declared
15251 with a "bind" line having the "ssl" option.
15252
Willy Tarreau74ca5042013-06-11 23:12:07 +020015253 Example :
15254 # This passes "X-Proto: https" to servers when client connects over SSL
15255 listen http-https
15256 bind :80
15257 bind :443 ssl crt /etc/haproxy.pem
15258 http-request add-header X-Proto https if { ssl_fc }
15259
15260ssl_fc_alg_keysize : integer
15261 Returns the symmetric cipher key size supported in bits when the incoming
15262 connection was made over an SSL/TLS transport layer.
15263
15264ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015265 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015266 incoming connection made via a TLS transport layer and locally deciphered by
15267 haproxy. The result is a string containing the protocol name advertised by
15268 the client. The SSL library must have been built with support for TLS
15269 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15270 not advertised unless the "alpn" keyword on the "bind" line specifies a
15271 protocol list. Also, nothing forces the client to pick a protocol from this
15272 list, any other one may be requested. The TLS ALPN extension is meant to
15273 replace the TLS NPN extension. See also "ssl_fc_npn".
15274
Willy Tarreau74ca5042013-06-11 23:12:07 +020015275ssl_fc_cipher : string
15276 Returns the name of the used cipher when the incoming connection was made
15277 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015278
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015279ssl_fc_cipherlist_bin : binary
15280 Returns the binary form of the client hello cipher list. The maximum returned
15281 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015282 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015283
15284ssl_fc_cipherlist_hex : string
15285 Returns the binary form of the client hello cipher list encoded as
15286 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015287 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015288
15289ssl_fc_cipherlist_str : string
15290 Returns the decoded text form of the client hello cipher list. The maximum
15291 number of ciphers returned is according with the value of
15292 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015293 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015294 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015295
15296ssl_fc_cipherlist_xxh : integer
15297 Returns a xxh64 of the cipher list. This hash can be return only is the value
15298 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015299 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015300
Patrick Hemmer65674662019-06-04 08:13:03 -040015301ssl_fc_client_random : binary
15302 Returns the client random of the front connection when the incoming connection
15303 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15304 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15305
Willy Tarreau74ca5042013-06-11 23:12:07 +020015306ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015307 Returns true if a client certificate is present in an incoming connection over
15308 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015309 Note: on SSL session resumption with Session ID or TLS ticket, client
15310 certificate is not present in the current connection but may be retrieved
15311 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15312 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015313
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015314ssl_fc_has_early : boolean
15315 Returns true if early data were sent, and the handshake didn't happen yet. As
15316 it has security implications, it is useful to be able to refuse those, or
15317 wait until the handshake happened.
15318
Willy Tarreau74ca5042013-06-11 23:12:07 +020015319ssl_fc_has_sni : boolean
15320 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015321 in an incoming connection was made over an SSL/TLS transport layer. Returns
15322 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050015323 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015324 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015325
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015326ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015327 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015328 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15329 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015330
Willy Tarreau74ca5042013-06-11 23:12:07 +020015331ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015332 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015333 made via a TLS transport layer and locally deciphered by haproxy. The result
15334 is a string containing the protocol name advertised by the client. The SSL
15335 library must have been built with support for TLS extensions enabled (check
15336 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15337 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15338 forces the client to pick a protocol from this list, any other one may be
15339 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015340
Willy Tarreau74ca5042013-06-11 23:12:07 +020015341ssl_fc_protocol : string
15342 Returns the name of the used protocol when the incoming connection was made
15343 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015344
Emeric Brunb73a9b02014-04-30 18:49:19 +020015345ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015346 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015347 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15348 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040015349
Patrick Hemmer65674662019-06-04 08:13:03 -040015350ssl_fc_server_random : binary
15351 Returns the server random of the front connection when the incoming connection
15352 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15353 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15354
Willy Tarreau74ca5042013-06-11 23:12:07 +020015355ssl_fc_session_id : binary
15356 Returns the SSL ID of the front connection when the incoming connection was
15357 made over an SSL/TLS transport layer. It is useful to stick a given client to
15358 a server. It is important to note that some browsers refresh their session ID
15359 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020015360
Patrick Hemmere0275472018-04-28 19:15:51 -040015361ssl_fc_session_key : binary
15362 Returns the SSL session master key of the front connection when the incoming
15363 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15364 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15365 BoringSSL.
15366
15367
Willy Tarreau74ca5042013-06-11 23:12:07 +020015368ssl_fc_sni : string
15369 This extracts the Server Name Indication TLS extension (SNI) field from an
15370 incoming connection made via an SSL/TLS transport layer and locally
15371 deciphered by haproxy. The result (when present) typically is a string
15372 matching the HTTPS host name (253 chars or less). The SSL library must have
15373 been built with support for TLS extensions enabled (check haproxy -vv).
15374
15375 This fetch is different from "req_ssl_sni" above in that it applies to the
15376 connection being deciphered by haproxy and not to SSL contents being blindly
15377 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050015378 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020015379 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020015380
Willy Tarreau74ca5042013-06-11 23:12:07 +020015381 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020015382 ssl_fc_sni_end : suffix match
15383 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020015384
Willy Tarreau74ca5042013-06-11 23:12:07 +020015385ssl_fc_use_keysize : integer
15386 Returns the symmetric cipher key size used in bits when the incoming
15387 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015388
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015389
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200153907.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015391------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015392
Willy Tarreau74ca5042013-06-11 23:12:07 +020015393Fetching samples from buffer contents is a bit different from the previous
15394sample fetches above because the sampled data are ephemeral. These data can
15395only be used when they're available and will be lost when they're forwarded.
15396For this reason, samples fetched from buffer contents during a request cannot
15397be used in a response for example. Even while the data are being fetched, they
15398can change. Sometimes it is necessary to set some delays or combine multiple
15399sample fetch methods to ensure that the expected data are complete and usable,
15400for example through TCP request content inspection. Please see the "tcp-request
15401content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015402
Willy Tarreau74ca5042013-06-11 23:12:07 +020015403payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015404 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015405 "stick on", "stick match"), and for "res.payload" when used in the context of
15406 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015407
Willy Tarreau74ca5042013-06-11 23:12:07 +020015408payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15409 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015410 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015411 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015412
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015413req.hdrs : string
15414 Returns the current request headers as string including the last empty line
15415 separating headers from the request body. The last empty line can be used to
15416 detect a truncated header block. This sample fetch is useful for some SPOE
15417 headers analyzers and for advanced logging.
15418
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015419req.hdrs_bin : binary
15420 Returns the current request headers contained in preparsed binary form. This
15421 is useful for offloading some processing with SPOE. Each string is described
15422 by a length followed by the number of bytes indicated in the length. The
15423 length is represented using the variable integer encoding detailed in the
15424 SPOE documentation. The end of the list is marked by a couple of empty header
15425 names and values (length of 0 for both).
15426
15427 *(<str:header-name><str:header-value>)<empty string><empty string>
15428
15429 int: refer to the SPOE documentation for the encoding
15430 str: <int:length><bytes>
15431
Willy Tarreau74ca5042013-06-11 23:12:07 +020015432req.len : integer
15433req_len : integer (deprecated)
15434 Returns an integer value corresponding to the number of bytes present in the
15435 request buffer. This is mostly used in ACL. It is important to understand
15436 that this test does not return false as long as the buffer is changing. This
15437 means that a check with equality to zero will almost always immediately match
15438 at the beginning of the session, while a test for more data will wait for
15439 that data to come in and return false only when haproxy is certain that no
15440 more data will come in. This test was designed to be used with TCP request
15441 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015442
Willy Tarreau74ca5042013-06-11 23:12:07 +020015443req.payload(<offset>,<length>) : binary
15444 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015445 in the request buffer. As a special case, if the <length> argument is zero,
15446 the the whole buffer from <offset> to the end is extracted. This can be used
15447 with ACLs in order to check for the presence of some content in a buffer at
15448 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015449
Willy Tarreau74ca5042013-06-11 23:12:07 +020015450 ACL alternatives :
15451 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015452
Willy Tarreau74ca5042013-06-11 23:12:07 +020015453req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15454 This extracts a binary block whose size is specified at <offset1> for <length>
15455 bytes, and which starts at <offset2> if specified or just after the length in
15456 the request buffer. The <offset2> parameter also supports relative offsets if
15457 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015458
Willy Tarreau74ca5042013-06-11 23:12:07 +020015459 ACL alternatives :
15460 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015461
Willy Tarreau74ca5042013-06-11 23:12:07 +020015462 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015463
Willy Tarreau74ca5042013-06-11 23:12:07 +020015464req.proto_http : boolean
15465req_proto_http : boolean (deprecated)
15466 Returns true when data in the request buffer look like HTTP and correctly
15467 parses as such. It is the same parser as the common HTTP request parser which
15468 is used so there should be no surprises. The test does not match until the
15469 request is complete, failed or timed out. This test may be used to report the
15470 protocol in TCP logs, but the biggest use is to block TCP request analysis
15471 until a complete HTTP request is present in the buffer, for example to track
15472 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015473
Willy Tarreau74ca5042013-06-11 23:12:07 +020015474 Example:
15475 # track request counts per "base" (concatenation of Host+URL)
15476 tcp-request inspect-delay 10s
15477 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015478 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015479
Willy Tarreau74ca5042013-06-11 23:12:07 +020015480req.rdp_cookie([<name>]) : string
15481rdp_cookie([<name>]) : string (deprecated)
15482 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15483 <name>, or any cookie if unspecified. The parser only checks for the first
15484 cookie, as illustrated in the RDP protocol specification. The cookie name is
15485 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15486 contain the user name of the client connecting to the server if properly
15487 configured on the client. The "MSTSHASH" cookie is often used as well for
15488 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015489
Willy Tarreau74ca5042013-06-11 23:12:07 +020015490 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15491 used and thus the distribution of clients to backend servers is not linked to
15492 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15493 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15494 distribution of clients to backend servers than the hash used by "balance
15495 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015496
Willy Tarreau74ca5042013-06-11 23:12:07 +020015497 ACL derivatives :
15498 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015499
Willy Tarreau74ca5042013-06-11 23:12:07 +020015500 Example :
15501 listen tse-farm
15502 bind 0.0.0.0:3389
15503 # wait up to 5s for an RDP cookie in the request
15504 tcp-request inspect-delay 5s
15505 tcp-request content accept if RDP_COOKIE
15506 # apply RDP cookie persistence
15507 persist rdp-cookie
15508 # Persist based on the mstshash cookie
15509 # This is only useful makes sense if
15510 # balance rdp-cookie is not used
15511 stick-table type string size 204800
15512 stick on req.rdp_cookie(mstshash)
15513 server srv1 1.1.1.1:3389
15514 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015515
Willy Tarreau74ca5042013-06-11 23:12:07 +020015516 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15517 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015518
Willy Tarreau74ca5042013-06-11 23:12:07 +020015519req.rdp_cookie_cnt([name]) : integer
15520rdp_cookie_cnt([name]) : integer (deprecated)
15521 Tries to parse the request buffer as RDP protocol, then returns an integer
15522 corresponding to the number of RDP cookies found. If an optional cookie name
15523 is passed, only cookies matching this name are considered. This is mostly
15524 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015525
Willy Tarreau74ca5042013-06-11 23:12:07 +020015526 ACL derivatives :
15527 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015528
Alex Zorin4afdd132018-12-30 13:56:28 +110015529req.ssl_alpn : string
15530 Returns a string containing the values of the Application-Layer Protocol
15531 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
15532 ClientHello message. Note that this only applies to raw contents found in the
15533 request buffer and not to the contents deciphered via an SSL data layer, so
15534 this will not work with "bind" lines having the "ssl" option. This is useful
15535 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020015536 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110015537
15538 Examples :
15539 # Wait for a client hello for at most 5 seconds
15540 tcp-request inspect-delay 5s
15541 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020015542 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110015543 default_backend bk_default
15544
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015545req.ssl_ec_ext : boolean
15546 Returns a boolean identifying if client sent the Supported Elliptic Curves
15547 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015548 message. This can be used to present ECC compatible clients with EC
15549 certificate and to use RSA for all others, on the same IP address. Note that
15550 this only applies to raw contents found in the request buffer and not to
15551 contents deciphered via an SSL data layer, so this will not work with "bind"
15552 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015553
Willy Tarreau74ca5042013-06-11 23:12:07 +020015554req.ssl_hello_type : integer
15555req_ssl_hello_type : integer (deprecated)
15556 Returns an integer value containing the type of the SSL hello message found
15557 in the request buffer if the buffer contains data that parse as a complete
15558 SSL (v3 or superior) client hello message. Note that this only applies to raw
15559 contents found in the request buffer and not to contents deciphered via an
15560 SSL data layer, so this will not work with "bind" lines having the "ssl"
15561 option. This is mostly used in ACL to detect presence of an SSL hello message
15562 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015563
Willy Tarreau74ca5042013-06-11 23:12:07 +020015564req.ssl_sni : string
15565req_ssl_sni : string (deprecated)
15566 Returns a string containing the value of the Server Name TLS extension sent
15567 by a client in a TLS stream passing through the request buffer if the buffer
15568 contains data that parse as a complete SSL (v3 or superior) client hello
15569 message. Note that this only applies to raw contents found in the request
15570 buffer and not to contents deciphered via an SSL data layer, so this will not
15571 work with "bind" lines having the "ssl" option. SNI normally contains the
15572 name of the host the client tries to connect to (for recent browsers). SNI is
15573 useful for allowing or denying access to certain hosts when SSL/TLS is used
15574 by the client. This test was designed to be used with TCP request content
15575 inspection. If content switching is needed, it is recommended to first wait
15576 for a complete client hello (type 1), like in the example below. See also
15577 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015578
Willy Tarreau74ca5042013-06-11 23:12:07 +020015579 ACL derivatives :
15580 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015581
Willy Tarreau74ca5042013-06-11 23:12:07 +020015582 Examples :
15583 # Wait for a client hello for at most 5 seconds
15584 tcp-request inspect-delay 5s
15585 tcp-request content accept if { req_ssl_hello_type 1 }
15586 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15587 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015588
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015589req.ssl_st_ext : integer
15590 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15591 Returns 1 if the client sent SessionTicket TLS Extension
15592 Returns 2 if the client also sent non-zero length TLS SessionTicket
15593 Note that this only applies to raw contents found in the request buffer and
15594 not to contents deciphered via an SSL data layer, so this will not work with
15595 "bind" lines having the "ssl" option. This can for example be used to detect
15596 whether the client sent a SessionTicket or not and stick it accordingly, if
15597 no SessionTicket then stick on SessionID or don't stick as there's no server
15598 side state is there when SessionTickets are in use.
15599
Willy Tarreau74ca5042013-06-11 23:12:07 +020015600req.ssl_ver : integer
15601req_ssl_ver : integer (deprecated)
15602 Returns an integer value containing the version of the SSL/TLS protocol of a
15603 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15604 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15605 composed of the major version multiplied by 65536, added to the minor
15606 version. Note that this only applies to raw contents found in the request
15607 buffer and not to contents deciphered via an SSL data layer, so this will not
15608 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015609 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015610 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015611
Willy Tarreau74ca5042013-06-11 23:12:07 +020015612 ACL derivatives :
15613 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015614
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015615res.len : integer
15616 Returns an integer value corresponding to the number of bytes present in the
15617 response buffer. This is mostly used in ACL. It is important to understand
15618 that this test does not return false as long as the buffer is changing. This
15619 means that a check with equality to zero will almost always immediately match
15620 at the beginning of the session, while a test for more data will wait for
15621 that data to come in and return false only when haproxy is certain that no
15622 more data will come in. This test was designed to be used with TCP response
15623 content inspection.
15624
Willy Tarreau74ca5042013-06-11 23:12:07 +020015625res.payload(<offset>,<length>) : binary
15626 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015627 in the response buffer. As a special case, if the <length> argument is zero,
15628 the the whole buffer from <offset> to the end is extracted. This can be used
15629 with ACLs in order to check for the presence of some content in a buffer at
15630 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015631
Willy Tarreau74ca5042013-06-11 23:12:07 +020015632res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15633 This extracts a binary block whose size is specified at <offset1> for <length>
15634 bytes, and which starts at <offset2> if specified or just after the length in
15635 the response buffer. The <offset2> parameter also supports relative offsets
15636 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015637
Willy Tarreau74ca5042013-06-11 23:12:07 +020015638 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015639
Willy Tarreau971f7b62015-09-29 14:06:59 +020015640res.ssl_hello_type : integer
15641rep_ssl_hello_type : integer (deprecated)
15642 Returns an integer value containing the type of the SSL hello message found
15643 in the response buffer if the buffer contains data that parses as a complete
15644 SSL (v3 or superior) hello message. Note that this only applies to raw
15645 contents found in the response buffer and not to contents deciphered via an
15646 SSL data layer, so this will not work with "server" lines having the "ssl"
15647 option. This is mostly used in ACL to detect presence of an SSL hello message
15648 that is supposed to contain an SSL session ID usable for stickiness.
15649
Willy Tarreau74ca5042013-06-11 23:12:07 +020015650wait_end : boolean
15651 This fetch either returns true when the inspection period is over, or does
15652 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015653 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015654 actions, such as a delayed reject for some special addresses. Since it either
15655 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015656 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015657 "WAIT_END" is always usable without prior declaration. This test was designed
15658 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015659
Willy Tarreau74ca5042013-06-11 23:12:07 +020015660 Examples :
15661 # delay every incoming request by 2 seconds
15662 tcp-request inspect-delay 2s
15663 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015664
Willy Tarreau74ca5042013-06-11 23:12:07 +020015665 # don't immediately tell bad guys they are rejected
15666 tcp-request inspect-delay 10s
15667 acl goodguys src 10.0.0.0/24
15668 acl badguys src 10.0.1.0/24
15669 tcp-request content accept if goodguys
15670 tcp-request content reject if badguys WAIT_END
15671 tcp-request content reject
15672
15673
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200156747.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015675--------------------------------------
15676
15677It is possible to fetch samples from HTTP contents, requests and responses.
15678This application layer is also called layer 7. It is only possible to fetch the
15679data in this section when a full HTTP request or response has been parsed from
15680its respective request or response buffer. This is always the case with all
15681HTTP specific rules and for sections running with "mode http". When using TCP
15682content inspection, it may be necessary to support an inspection delay in order
15683to let the request or response come in first. These fetches may require a bit
15684more CPU resources than the layer 4 ones, but not much since the request and
15685response are indexed.
15686
15687base : string
15688 This returns the concatenation of the first Host header and the path part of
15689 the request, which starts at the first slash and ends before the question
15690 mark. It can be useful in virtual hosted environments to detect URL abuses as
15691 well as to improve shared caches efficiency. Using this with a limited size
15692 stick table also allows one to collect statistics about most commonly
15693 requested objects by host/path. With ACLs it can allow simple content
15694 switching rules involving the host and the path at the same time, such as
15695 "www.example.com/favicon.ico". See also "path" and "uri".
15696
15697 ACL derivatives :
15698 base : exact string match
15699 base_beg : prefix match
15700 base_dir : subdir match
15701 base_dom : domain match
15702 base_end : suffix match
15703 base_len : length match
15704 base_reg : regex match
15705 base_sub : substring match
15706
15707base32 : integer
15708 This returns a 32-bit hash of the value returned by the "base" fetch method
15709 above. This is useful to track per-URL activity on high traffic sites without
15710 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015711 memory. The output type is an unsigned integer. The hash function used is
15712 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15713 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015714
15715base32+src : binary
15716 This returns the concatenation of the base32 fetch above and the src fetch
15717 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15718 depending on the source address family. This can be used to track per-IP,
15719 per-URL counters.
15720
William Lallemand65ad6e12014-01-31 15:08:02 +010015721capture.req.hdr(<idx>) : string
15722 This extracts the content of the header captured by the "capture request
15723 header", idx is the position of the capture keyword in the configuration.
15724 The first entry is an index of 0. See also: "capture request header".
15725
15726capture.req.method : string
15727 This extracts the METHOD of an HTTP request. It can be used in both request
15728 and response. Unlike "method", it can be used in both request and response
15729 because it's allocated.
15730
15731capture.req.uri : string
15732 This extracts the request's URI, which starts at the first slash and ends
15733 before the first space in the request (without the host part). Unlike "path"
15734 and "url", it can be used in both request and response because it's
15735 allocated.
15736
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015737capture.req.ver : string
15738 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15739 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15740 logs because it relies on a persistent flag.
15741
William Lallemand65ad6e12014-01-31 15:08:02 +010015742capture.res.hdr(<idx>) : string
15743 This extracts the content of the header captured by the "capture response
15744 header", idx is the position of the capture keyword in the configuration.
15745 The first entry is an index of 0.
15746 See also: "capture response header"
15747
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015748capture.res.ver : string
15749 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15750 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15751 persistent flag.
15752
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015753req.body : binary
15754 This returns the HTTP request's available body as a block of data. It
15755 requires that the request body has been buffered made available using
15756 "option http-buffer-request". In case of chunked-encoded body, currently only
15757 the first chunk is analyzed.
15758
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015759req.body_param([<name>) : string
15760 This fetch assumes that the body of the POST request is url-encoded. The user
15761 can check if the "content-type" contains the value
15762 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15763 parameter <name> in the body, which ends before '&'. The parameter name is
15764 case-sensitive. If no name is given, any parameter will match, and the first
15765 one will be returned. The result is a string corresponding to the value of the
15766 parameter <name> as presented in the request body (no URL decoding is
15767 performed). Note that the ACL version of this fetch iterates over multiple
15768 parameters and will iteratively report all parameters values if no name is
15769 given.
15770
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015771req.body_len : integer
15772 This returns the length of the HTTP request's available body in bytes. It may
15773 be lower than the advertised length if the body is larger than the buffer. It
15774 requires that the request body has been buffered made available using
15775 "option http-buffer-request".
15776
15777req.body_size : integer
15778 This returns the advertised length of the HTTP request's body in bytes. It
15779 will represent the advertised Content-Length header, or the size of the first
15780 chunk in case of chunked encoding. In order to parse the chunks, it requires
15781 that the request body has been buffered made available using
15782 "option http-buffer-request".
15783
Willy Tarreau74ca5042013-06-11 23:12:07 +020015784req.cook([<name>]) : string
15785cook([<name>]) : string (deprecated)
15786 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15787 header line from the request, and returns its value as string. If no name is
15788 specified, the first cookie value is returned. When used with ACLs, all
15789 matching cookies are evaluated. Spaces around the name and the value are
15790 ignored as requested by the Cookie header specification (RFC6265). The cookie
15791 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15792 well return an empty value if it is present. Use the "found" match to detect
15793 presence. Use the res.cook() variant for response cookies sent by the server.
15794
15795 ACL derivatives :
15796 cook([<name>]) : exact string match
15797 cook_beg([<name>]) : prefix match
15798 cook_dir([<name>]) : subdir match
15799 cook_dom([<name>]) : domain match
15800 cook_end([<name>]) : suffix match
15801 cook_len([<name>]) : length match
15802 cook_reg([<name>]) : regex match
15803 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015804
Willy Tarreau74ca5042013-06-11 23:12:07 +020015805req.cook_cnt([<name>]) : integer
15806cook_cnt([<name>]) : integer (deprecated)
15807 Returns an integer value representing the number of occurrences of the cookie
15808 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015809
Willy Tarreau74ca5042013-06-11 23:12:07 +020015810req.cook_val([<name>]) : integer
15811cook_val([<name>]) : integer (deprecated)
15812 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15813 header line from the request, and converts its value to an integer which is
15814 returned. If no name is specified, the first cookie value is returned. When
15815 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015816
Willy Tarreau74ca5042013-06-11 23:12:07 +020015817cookie([<name>]) : string (deprecated)
15818 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15819 header line from the request, or a "Set-Cookie" header from the response, and
15820 returns its value as a string. A typical use is to get multiple clients
15821 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015822 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015823 multi-peer synchronization and state keeping across restarts. If no name is
15824 specified, the first cookie value is returned. This fetch should not be used
15825 anymore and should be replaced by req.cook() or res.cook() instead as it
15826 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015827
Willy Tarreau74ca5042013-06-11 23:12:07 +020015828hdr([<name>[,<occ>]]) : string
15829 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15830 used on responses. Please refer to these respective fetches for more details.
15831 In case of doubt about the fetch direction, please use the explicit ones.
15832 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015833 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015834
Willy Tarreau74ca5042013-06-11 23:12:07 +020015835req.fhdr(<name>[,<occ>]) : string
15836 This extracts the last occurrence of header <name> in an HTTP request. When
15837 used from an ACL, all occurrences are iterated over until a match is found.
15838 Optionally, a specific occurrence might be specified as a position number.
15839 Positive values indicate a position from the first occurrence, with 1 being
15840 the first one. Negative values indicate positions relative to the last one,
15841 with -1 being the last one. It differs from req.hdr() in that any commas
15842 present in the value are returned and are not used as delimiters. This is
15843 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015844
Willy Tarreau74ca5042013-06-11 23:12:07 +020015845req.fhdr_cnt([<name>]) : integer
15846 Returns an integer value representing the number of occurrences of request
15847 header field name <name>, or the total number of header fields if <name> is
15848 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15849 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015850
Willy Tarreau74ca5042013-06-11 23:12:07 +020015851req.hdr([<name>[,<occ>]]) : string
15852 This extracts the last occurrence of header <name> in an HTTP request. When
15853 used from an ACL, all occurrences are iterated over until a match is found.
15854 Optionally, a specific occurrence might be specified as a position number.
15855 Positive values indicate a position from the first occurrence, with 1 being
15856 the first one. Negative values indicate positions relative to the last one,
15857 with -1 being the last one. A typical use is with the X-Forwarded-For header
15858 once converted to IP, associated with an IP stick-table. The function
15859 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015860 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015861 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015862 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015863
Willy Tarreau74ca5042013-06-11 23:12:07 +020015864 ACL derivatives :
15865 hdr([<name>[,<occ>]]) : exact string match
15866 hdr_beg([<name>[,<occ>]]) : prefix match
15867 hdr_dir([<name>[,<occ>]]) : subdir match
15868 hdr_dom([<name>[,<occ>]]) : domain match
15869 hdr_end([<name>[,<occ>]]) : suffix match
15870 hdr_len([<name>[,<occ>]]) : length match
15871 hdr_reg([<name>[,<occ>]]) : regex match
15872 hdr_sub([<name>[,<occ>]]) : substring match
15873
15874req.hdr_cnt([<name>]) : integer
15875hdr_cnt([<header>]) : integer (deprecated)
15876 Returns an integer value representing the number of occurrences of request
15877 header field name <name>, or the total number of header field values if
15878 <name> is not specified. It is important to remember that one header line may
15879 count as several headers if it has several values. The function considers any
15880 comma as a delimiter for distinct values. If full-line headers are desired
15881 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15882 detect presence, absence or abuse of a specific header, as well as to block
15883 request smuggling attacks by rejecting requests which contain more than one
15884 of certain headers. See "req.hdr" for more information on header matching.
15885
15886req.hdr_ip([<name>[,<occ>]]) : ip
15887hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15888 This extracts the last occurrence of header <name> in an HTTP request,
15889 converts it to an IPv4 or IPv6 address and returns this address. When used
15890 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15891 of every header is checked. Optionally, a specific occurrence might be
15892 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015893 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015894 positions relative to the last one, with -1 being the last one. A typical use
15895 is with the X-Forwarded-For and X-Client-IP headers.
15896
15897req.hdr_val([<name>[,<occ>]]) : integer
15898hdr_val([<name>[,<occ>]]) : integer (deprecated)
15899 This extracts the last occurrence of header <name> in an HTTP request, and
15900 converts it to an integer value. When used with ACLs, all occurrences are
15901 checked, and if <name> is omitted, every value of every header is checked.
15902 Optionally, a specific occurrence might be specified as a position number.
15903 Positive values indicate a position from the first occurrence, with 1 being
15904 the first one. Negative values indicate positions relative to the last one,
15905 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15906
Frédéric Lécailleec891192019-02-26 15:02:35 +010015907
15908
Willy Tarreau74ca5042013-06-11 23:12:07 +020015909http_auth(<userlist>) : boolean
15910 Returns a boolean indicating whether the authentication data received from
15911 the client match a username & password stored in the specified userlist. This
15912 fetch function is not really useful outside of ACLs. Currently only http
15913 basic auth is supported.
15914
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015915http_auth_group(<userlist>) : string
15916 Returns a string corresponding to the user name found in the authentication
15917 data received from the client if both the user name and password are valid
15918 according to the specified userlist. The main purpose is to use it in ACLs
15919 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015920 This fetch function is not really useful outside of ACLs. Currently only http
15921 basic auth is supported.
15922
15923 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015924 http_auth_group(<userlist>) : group ...
15925 Returns true when the user extracted from the request and whose password is
15926 valid according to the specified userlist belongs to at least one of the
15927 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015928
15929http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015930 Returns true when the request being processed is the first one of the
15931 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015932 from some requests when a request is not the first one, or to help grouping
15933 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015934
Willy Tarreau74ca5042013-06-11 23:12:07 +020015935method : integer + string
15936 Returns an integer value corresponding to the method in the HTTP request. For
15937 example, "GET" equals 1 (check sources to establish the matching). Value 9
15938 means "other method" and may be converted to a string extracted from the
15939 stream. This should not be used directly as a sample, this is only meant to
15940 be used from ACLs, which transparently convert methods from patterns to these
15941 integer + string values. Some predefined ACL already check for most common
15942 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015943
Willy Tarreau74ca5042013-06-11 23:12:07 +020015944 ACL derivatives :
15945 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015946
Willy Tarreau74ca5042013-06-11 23:12:07 +020015947 Example :
15948 # only accept GET and HEAD requests
15949 acl valid_method method GET HEAD
15950 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020015951
Willy Tarreau74ca5042013-06-11 23:12:07 +020015952path : string
15953 This extracts the request's URL path, which starts at the first slash and
15954 ends before the question mark (without the host part). A typical use is with
15955 prefetch-capable caches, and with portals which need to aggregate multiple
15956 information from databases and keep them in caches. Note that with outgoing
15957 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010015958 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020015959 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015960
Willy Tarreau74ca5042013-06-11 23:12:07 +020015961 ACL derivatives :
15962 path : exact string match
15963 path_beg : prefix match
15964 path_dir : subdir match
15965 path_dom : domain match
15966 path_end : suffix match
15967 path_len : length match
15968 path_reg : regex match
15969 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015970
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015971query : string
15972 This extracts the request's query string, which starts after the first
15973 question mark. If no question mark is present, this fetch returns nothing. If
15974 a question mark is present but nothing follows, it returns an empty string.
15975 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010015976 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015977 which stops before the question mark.
15978
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015979req.hdr_names([<delim>]) : string
15980 This builds a string made from the concatenation of all header names as they
15981 appear in the request when the rule is evaluated. The default delimiter is
15982 the comma (',') but it may be overridden as an optional argument <delim>. In
15983 this case, only the first character of <delim> is considered.
15984
Willy Tarreau74ca5042013-06-11 23:12:07 +020015985req.ver : string
15986req_ver : string (deprecated)
15987 Returns the version string from the HTTP request, for example "1.1". This can
15988 be useful for logs, but is mostly there for ACL. Some predefined ACL already
15989 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015990
Willy Tarreau74ca5042013-06-11 23:12:07 +020015991 ACL derivatives :
15992 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020015993
Willy Tarreau74ca5042013-06-11 23:12:07 +020015994res.comp : boolean
15995 Returns the boolean "true" value if the response has been compressed by
15996 HAProxy, otherwise returns boolean "false". This may be used to add
15997 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015998
Willy Tarreau74ca5042013-06-11 23:12:07 +020015999res.comp_algo : string
16000 Returns a string containing the name of the algorithm used if the response
16001 was compressed by HAProxy, for example : "deflate". This may be used to add
16002 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016003
Willy Tarreau74ca5042013-06-11 23:12:07 +020016004res.cook([<name>]) : string
16005scook([<name>]) : string (deprecated)
16006 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16007 header line from the response, and returns its value as string. If no name is
16008 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016009
Willy Tarreau74ca5042013-06-11 23:12:07 +020016010 ACL derivatives :
16011 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016012
Willy Tarreau74ca5042013-06-11 23:12:07 +020016013res.cook_cnt([<name>]) : integer
16014scook_cnt([<name>]) : integer (deprecated)
16015 Returns an integer value representing the number of occurrences of the cookie
16016 <name> in the response, or all cookies if <name> is not specified. This is
16017 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016018
Willy Tarreau74ca5042013-06-11 23:12:07 +020016019res.cook_val([<name>]) : integer
16020scook_val([<name>]) : integer (deprecated)
16021 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16022 header line from the response, and converts its value to an integer which is
16023 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016024
Willy Tarreau74ca5042013-06-11 23:12:07 +020016025res.fhdr([<name>[,<occ>]]) : string
16026 This extracts the last occurrence of header <name> in an HTTP response, or of
16027 the last header if no <name> is specified. Optionally, a specific occurrence
16028 might be specified as a position number. Positive values indicate a position
16029 from the first occurrence, with 1 being the first one. Negative values
16030 indicate positions relative to the last one, with -1 being the last one. It
16031 differs from res.hdr() in that any commas present in the value are returned
16032 and are not used as delimiters. If this is not desired, the res.hdr() fetch
16033 should be used instead. This is sometimes useful with headers such as Date or
16034 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016035
Willy Tarreau74ca5042013-06-11 23:12:07 +020016036res.fhdr_cnt([<name>]) : integer
16037 Returns an integer value representing the number of occurrences of response
16038 header field name <name>, or the total number of header fields if <name> is
16039 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
16040 the number of full line headers and does not stop on commas. If this is not
16041 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016042
Willy Tarreau74ca5042013-06-11 23:12:07 +020016043res.hdr([<name>[,<occ>]]) : string
16044shdr([<name>[,<occ>]]) : string (deprecated)
16045 This extracts the last occurrence of header <name> in an HTTP response, or of
16046 the last header if no <name> is specified. Optionally, a specific occurrence
16047 might be specified as a position number. Positive values indicate a position
16048 from the first occurrence, with 1 being the first one. Negative values
16049 indicate positions relative to the last one, with -1 being the last one. This
16050 can be useful to learn some data into a stick-table. The function considers
16051 any comma as a delimiter for distinct values. If this is not desired, the
16052 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016053
Willy Tarreau74ca5042013-06-11 23:12:07 +020016054 ACL derivatives :
16055 shdr([<name>[,<occ>]]) : exact string match
16056 shdr_beg([<name>[,<occ>]]) : prefix match
16057 shdr_dir([<name>[,<occ>]]) : subdir match
16058 shdr_dom([<name>[,<occ>]]) : domain match
16059 shdr_end([<name>[,<occ>]]) : suffix match
16060 shdr_len([<name>[,<occ>]]) : length match
16061 shdr_reg([<name>[,<occ>]]) : regex match
16062 shdr_sub([<name>[,<occ>]]) : substring match
16063
16064res.hdr_cnt([<name>]) : integer
16065shdr_cnt([<name>]) : integer (deprecated)
16066 Returns an integer value representing the number of occurrences of response
16067 header field name <name>, or the total number of header fields if <name> is
16068 not specified. The function considers any comma as a delimiter for distinct
16069 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16070 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016071
Willy Tarreau74ca5042013-06-11 23:12:07 +020016072res.hdr_ip([<name>[,<occ>]]) : ip
16073shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16074 This extracts the last occurrence of header <name> in an HTTP response,
16075 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16076 specific occurrence might be specified as a position number. Positive values
16077 indicate a position from the first occurrence, with 1 being the first one.
16078 Negative values indicate positions relative to the last one, with -1 being
16079 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016080
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016081res.hdr_names([<delim>]) : string
16082 This builds a string made from the concatenation of all header names as they
16083 appear in the response when the rule is evaluated. The default delimiter is
16084 the comma (',') but it may be overridden as an optional argument <delim>. In
16085 this case, only the first character of <delim> is considered.
16086
Willy Tarreau74ca5042013-06-11 23:12:07 +020016087res.hdr_val([<name>[,<occ>]]) : integer
16088shdr_val([<name>[,<occ>]]) : integer (deprecated)
16089 This extracts the last occurrence of header <name> in an HTTP response, and
16090 converts it to an integer value. Optionally, a specific occurrence might be
16091 specified as a position number. Positive values indicate a position from the
16092 first occurrence, with 1 being the first one. Negative values indicate
16093 positions relative to the last one, with -1 being the last one. This can be
16094 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016095
Willy Tarreau74ca5042013-06-11 23:12:07 +020016096res.ver : string
16097resp_ver : string (deprecated)
16098 Returns the version string from the HTTP response, for example "1.1". This
16099 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016100
Willy Tarreau74ca5042013-06-11 23:12:07 +020016101 ACL derivatives :
16102 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016103
Willy Tarreau74ca5042013-06-11 23:12:07 +020016104set-cookie([<name>]) : string (deprecated)
16105 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16106 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016107 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016108 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016109
Willy Tarreau74ca5042013-06-11 23:12:07 +020016110 This fetch function is deprecated and has been superseded by the "res.cook"
16111 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016112
Willy Tarreau74ca5042013-06-11 23:12:07 +020016113status : integer
16114 Returns an integer containing the HTTP status code in the HTTP response, for
16115 example, 302. It is mostly used within ACLs and integer ranges, for example,
16116 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016117
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016118unique-id : string
16119 Returns the unique-id attached to the request. The directive
16120 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16121 fails. Note that the unique-id is usually used with HTTP requests, however this
16122 sample fetch can be used with other protocols. Obviously, if it is used with
16123 other protocols than HTTP, the unique-id-format directive must not contain
16124 HTTP parts. See: unique-id-format and unique-id-header
16125
Willy Tarreau74ca5042013-06-11 23:12:07 +020016126url : string
16127 This extracts the request's URL as presented in the request. A typical use is
16128 with prefetch-capable caches, and with portals which need to aggregate
16129 multiple information from databases and keep them in caches. With ACLs, using
16130 "path" is preferred over using "url", because clients may send a full URL as
16131 is normally done with proxies. The only real use is to match "*" which does
16132 not match in "path", and for which there is already a predefined ACL. See
16133 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016134
Willy Tarreau74ca5042013-06-11 23:12:07 +020016135 ACL derivatives :
16136 url : exact string match
16137 url_beg : prefix match
16138 url_dir : subdir match
16139 url_dom : domain match
16140 url_end : suffix match
16141 url_len : length match
16142 url_reg : regex match
16143 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016144
Willy Tarreau74ca5042013-06-11 23:12:07 +020016145url_ip : ip
16146 This extracts the IP address from the request's URL when the host part is
16147 presented as an IP address. Its use is very limited. For instance, a
16148 monitoring system might use this field as an alternative for the source IP in
16149 order to test what path a given source address would follow, or to force an
16150 entry in a table for a given source address. With ACLs it can be used to
16151 restrict access to certain systems through a proxy, for example when combined
16152 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016153
Willy Tarreau74ca5042013-06-11 23:12:07 +020016154url_port : integer
16155 This extracts the port part from the request's URL. Note that if the port is
16156 not specified in the request, port 80 is assumed. With ACLs it can be used to
16157 restrict access to certain systems through a proxy, for example when combined
16158 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016159
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016160urlp([<name>[,<delim>]]) : string
16161url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016162 This extracts the first occurrence of the parameter <name> in the query
16163 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016164 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16165 any parameter will match, and the first one will be returned. The result is
16166 a string corresponding to the value of the parameter <name> as presented in
16167 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016168 stickiness based on a client ID, to extract an application cookie passed as a
16169 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016170 this fetch iterates over multiple parameters and will iteratively report all
16171 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016172
Willy Tarreau74ca5042013-06-11 23:12:07 +020016173 ACL derivatives :
16174 urlp(<name>[,<delim>]) : exact string match
16175 urlp_beg(<name>[,<delim>]) : prefix match
16176 urlp_dir(<name>[,<delim>]) : subdir match
16177 urlp_dom(<name>[,<delim>]) : domain match
16178 urlp_end(<name>[,<delim>]) : suffix match
16179 urlp_len(<name>[,<delim>]) : length match
16180 urlp_reg(<name>[,<delim>]) : regex match
16181 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016182
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016183
Willy Tarreau74ca5042013-06-11 23:12:07 +020016184 Example :
16185 # match http://example.com/foo?PHPSESSIONID=some_id
16186 stick on urlp(PHPSESSIONID)
16187 # match http://example.com/foo;JSESSIONID=some_id
16188 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016189
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016190urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016191 See "urlp" above. This one extracts the URL parameter <name> in the request
16192 and converts it to an integer value. This can be used for session stickiness
16193 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016194
Dragan Dosen0070cd52016-06-16 12:19:49 +020016195url32 : integer
16196 This returns a 32-bit hash of the value obtained by concatenating the first
16197 Host header and the whole URL including parameters (not only the path part of
16198 the request, as in the "base32" fetch above). This is useful to track per-URL
16199 activity. A shorter hash is stored, saving a lot of memory. The output type
16200 is an unsigned integer.
16201
16202url32+src : binary
16203 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16204 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16205 the source address family. This can be used to track per-IP, per-URL counters.
16206
Willy Tarreau198a7442008-01-17 12:05:32 +010016207
Willy Tarreau74ca5042013-06-11 23:12:07 +0200162087.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016209---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016210
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016211Some predefined ACLs are hard-coded so that they do not have to be declared in
16212every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016213order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016214
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016215ACL name Equivalent to Usage
16216---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016217FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016218HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016219HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16220HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016221HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16222HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16223HTTP_URL_SLASH url_beg / match URL beginning with "/"
16224HTTP_URL_STAR url * match URL equal to "*"
16225LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016226METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016227METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016228METH_GET method GET HEAD match HTTP GET or HEAD method
16229METH_HEAD method HEAD match HTTP HEAD method
16230METH_OPTIONS method OPTIONS match HTTP OPTIONS method
16231METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016232METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016233METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020016234RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016235REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016236TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016237WAIT_END wait_end wait for end of content analysis
16238---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016239
Willy Tarreaub937b7e2010-01-12 15:27:54 +010016240
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162418. Logging
16242----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010016243
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016244One of HAProxy's strong points certainly lies is its precise logs. It probably
16245provides the finest level of information available for such a product, which is
16246very important for troubleshooting complex environments. Standard information
16247provided in logs include client ports, TCP/HTTP state timers, precise session
16248state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016249to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016250headers.
16251
16252In order to improve administrators reactivity, it offers a great transparency
16253about encountered problems, both internal and external, and it is possible to
16254send logs to different sources at the same time with different level filters :
16255
16256 - global process-level logs (system errors, start/stop, etc..)
16257 - per-instance system and internal errors (lack of resource, bugs, ...)
16258 - per-instance external troubles (servers up/down, max connections)
16259 - per-instance activity (client connections), either at the establishment or
16260 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016261 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016262 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016263
16264The ability to distribute different levels of logs to different log servers
16265allow several production teams to interact and to fix their problems as soon
16266as possible. For example, the system team might monitor system-wide errors,
16267while the application team might be monitoring the up/down for their servers in
16268real time, and the security team might analyze the activity logs with one hour
16269delay.
16270
16271
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162728.1. Log levels
16273---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016274
Simon Hormandf791f52011-05-29 15:01:10 +090016275TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016276source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016277HTTP request, HTTP return code, number of bytes transmitted, conditions
16278in which the session ended, and even exchanged cookies values. For example
16279track a particular user's problems. All messages may be sent to up to two
16280syslog servers. Check the "log" keyword in section 4.2 for more information
16281about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016282
16283
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162848.2. Log formats
16285----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016286
William Lallemand48940402012-01-30 16:47:22 +010016287HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016288and will be detailed in the following sections. A few of them may vary
16289slightly with the configuration, due to indicators specific to certain
16290options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016291
16292 - the default format, which is very basic and very rarely used. It only
16293 provides very basic information about the incoming connection at the moment
16294 it is accepted : source IP:port, destination IP:port, and frontend-name.
16295 This mode will eventually disappear so it will not be described to great
16296 extents.
16297
16298 - the TCP format, which is more advanced. This format is enabled when "option
16299 tcplog" is set on the frontend. HAProxy will then usually wait for the
16300 connection to terminate before logging. This format provides much richer
16301 information, such as timers, connection counts, queue size, etc... This
16302 format is recommended for pure TCP proxies.
16303
16304 - the HTTP format, which is the most advanced for HTTP proxying. This format
16305 is enabled when "option httplog" is set on the frontend. It provides the
16306 same information as the TCP format with some HTTP-specific fields such as
16307 the request, the status code, and captures of headers and cookies. This
16308 format is recommended for HTTP proxies.
16309
Emeric Brun3a058f32009-06-30 18:26:00 +020016310 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16311 fields arranged in the same order as the CLF format. In this mode, all
16312 timers, captures, flags, etc... appear one per field after the end of the
16313 common fields, in the same order they appear in the standard HTTP format.
16314
William Lallemand48940402012-01-30 16:47:22 +010016315 - the custom log format, allows you to make your own log line.
16316
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016317Next sections will go deeper into details for each of these formats. Format
16318specification will be performed on a "field" basis. Unless stated otherwise, a
16319field is a portion of text delimited by any number of spaces. Since syslog
16320servers are susceptible of inserting fields at the beginning of a line, it is
16321always assumed that the first field is the one containing the process name and
16322identifier.
16323
16324Note : Since log lines may be quite long, the log examples in sections below
16325 might be broken into multiple lines. The example log lines will be
16326 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16327 broken into multiple lines, each non-final line will end with a
16328 backslash ('\') and the next line will start indented by two characters.
16329
16330
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163318.2.1. Default log format
16332-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016333
16334This format is used when no specific option is set. The log is emitted as soon
16335as the connection is accepted. One should note that this currently is the only
16336format which logs the request's destination IP and ports.
16337
16338 Example :
16339 listen www
16340 mode http
16341 log global
16342 server srv1 127.0.0.1:8000
16343
16344 >>> Feb 6 12:12:09 localhost \
16345 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16346 (www/HTTP)
16347
16348 Field Format Extract from the example above
16349 1 process_name '[' pid ']:' haproxy[14385]:
16350 2 'Connect from' Connect from
16351 3 source_ip ':' source_port 10.0.1.2:33312
16352 4 'to' to
16353 5 destination_ip ':' destination_port 10.0.3.31:8012
16354 6 '(' frontend_name '/' mode ')' (www/HTTP)
16355
16356Detailed fields description :
16357 - "source_ip" is the IP address of the client which initiated the connection.
16358 - "source_port" is the TCP port of the client which initiated the connection.
16359 - "destination_ip" is the IP address the client connected to.
16360 - "destination_port" is the TCP port the client connected to.
16361 - "frontend_name" is the name of the frontend (or listener) which received
16362 and processed the connection.
16363 - "mode is the mode the frontend is operating (TCP or HTTP).
16364
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016365In case of a UNIX socket, the source and destination addresses are marked as
16366"unix:" and the ports reflect the internal ID of the socket which accepted the
16367connection (the same ID as reported in the stats).
16368
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016369It is advised not to use this deprecated format for newer installations as it
16370will eventually disappear.
16371
16372
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163738.2.2. TCP log format
16374---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016375
16376The TCP format is used when "option tcplog" is specified in the frontend, and
16377is the recommended format for pure TCP proxies. It provides a lot of precious
16378information for troubleshooting. Since this format includes timers and byte
16379counts, the log is normally emitted at the end of the session. It can be
16380emitted earlier if "option logasap" is specified, which makes sense in most
16381environments with long sessions such as remote terminals. Sessions which match
16382the "monitor" rules are never logged. It is also possible not to emit logs for
16383sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016384specifying "option dontlognull" in the frontend. Successful connections will
16385not be logged if "option dontlog-normal" is specified in the frontend. A few
16386fields may slightly vary depending on some configuration options, those are
16387marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016388
16389 Example :
16390 frontend fnt
16391 mode tcp
16392 option tcplog
16393 log global
16394 default_backend bck
16395
16396 backend bck
16397 server srv1 127.0.0.1:8000
16398
16399 >>> Feb 6 12:12:56 localhost \
16400 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
16401 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
16402
16403 Field Format Extract from the example above
16404 1 process_name '[' pid ']:' haproxy[14387]:
16405 2 client_ip ':' client_port 10.0.1.2:33313
16406 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
16407 4 frontend_name fnt
16408 5 backend_name '/' server_name bck/srv1
16409 6 Tw '/' Tc '/' Tt* 0/0/5007
16410 7 bytes_read* 212
16411 8 termination_state --
16412 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16413 10 srv_queue '/' backend_queue 0/0
16414
16415Detailed fields description :
16416 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016417 connection to haproxy. If the connection was accepted on a UNIX socket
16418 instead, the IP address would be replaced with the word "unix". Note that
16419 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016420 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016421 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016422 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016423
16424 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016425 If the connection was accepted on a UNIX socket instead, the port would be
16426 replaced with the ID of the accepting socket, which is also reported in the
16427 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016428
16429 - "accept_date" is the exact date when the connection was received by haproxy
16430 (which might be very slightly different from the date observed on the
16431 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016432 the same date which may appear in any upstream firewall's log. When used in
16433 HTTP mode, the accept_date field will be reset to the first moment the
16434 connection is ready to receive a new request (end of previous response for
16435 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016436
16437 - "frontend_name" is the name of the frontend (or listener) which received
16438 and processed the connection.
16439
16440 - "backend_name" is the name of the backend (or listener) which was selected
16441 to manage the connection to the server. This will be the same as the
16442 frontend if no switching rule has been applied, which is common for TCP
16443 applications.
16444
16445 - "server_name" is the name of the last server to which the connection was
16446 sent, which might differ from the first one if there were connection errors
16447 and a redispatch occurred. Note that this server belongs to the backend
16448 which processed the request. If the connection was aborted before reaching
16449 a server, "<NOSRV>" is indicated instead of a server name.
16450
16451 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16452 It can be "-1" if the connection was aborted before reaching the queue.
16453 See "Timers" below for more details.
16454
16455 - "Tc" is the total time in milliseconds spent waiting for the connection to
16456 establish to the final server, including retries. It can be "-1" if the
16457 connection was aborted before a connection could be established. See
16458 "Timers" below for more details.
16459
16460 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016461 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016462 "option logasap" was specified, then the time counting stops at the moment
16463 the log is emitted. In this case, a '+' sign is prepended before the value,
16464 indicating that the final one will be larger. See "Timers" below for more
16465 details.
16466
16467 - "bytes_read" is the total number of bytes transmitted from the server to
16468 the client when the log is emitted. If "option logasap" is specified, the
16469 this value will be prefixed with a '+' sign indicating that the final one
16470 may be larger. Please note that this value is a 64-bit counter, so log
16471 analysis tools must be able to handle it without overflowing.
16472
16473 - "termination_state" is the condition the session was in when the session
16474 ended. This indicates the session state, which side caused the end of
16475 session to happen, and for what reason (timeout, error, ...). The normal
16476 flags should be "--", indicating the session was closed by either end with
16477 no data remaining in buffers. See below "Session state at disconnection"
16478 for more details.
16479
16480 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016481 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016482 limits have been reached. For instance, if actconn is close to 512 when
16483 multiple connection errors occur, chances are high that the system limits
16484 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016485 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016486
16487 - "feconn" is the total number of concurrent connections on the frontend when
16488 the session was logged. It is useful to estimate the amount of resource
16489 required to sustain high loads, and to detect when the frontend's "maxconn"
16490 has been reached. Most often when this value increases by huge jumps, it is
16491 because there is congestion on the backend servers, but sometimes it can be
16492 caused by a denial of service attack.
16493
16494 - "beconn" is the total number of concurrent connections handled by the
16495 backend when the session was logged. It includes the total number of
16496 concurrent connections active on servers as well as the number of
16497 connections pending in queues. It is useful to estimate the amount of
16498 additional servers needed to support high loads for a given application.
16499 Most often when this value increases by huge jumps, it is because there is
16500 congestion on the backend servers, but sometimes it can be caused by a
16501 denial of service attack.
16502
16503 - "srv_conn" is the total number of concurrent connections still active on
16504 the server when the session was logged. It can never exceed the server's
16505 configured "maxconn" parameter. If this value is very often close or equal
16506 to the server's "maxconn", it means that traffic regulation is involved a
16507 lot, meaning that either the server's maxconn value is too low, or that
16508 there aren't enough servers to process the load with an optimal response
16509 time. When only one of the server's "srv_conn" is high, it usually means
16510 that this server has some trouble causing the connections to take longer to
16511 be processed than on other servers.
16512
16513 - "retries" is the number of connection retries experienced by this session
16514 when trying to connect to the server. It must normally be zero, unless a
16515 server is being stopped at the same moment the connection was attempted.
16516 Frequent retries generally indicate either a network problem between
16517 haproxy and the server, or a misconfigured system backlog on the server
16518 preventing new connections from being queued. This field may optionally be
16519 prefixed with a '+' sign, indicating that the session has experienced a
16520 redispatch after the maximal retry count has been reached on the initial
16521 server. In this case, the server name appearing in the log is the one the
16522 connection was redispatched to, and not the first one, though both may
16523 sometimes be the same in case of hashing for instance. So as a general rule
16524 of thumb, when a '+' is present in front of the retry count, this count
16525 should not be attributed to the logged server.
16526
16527 - "srv_queue" is the total number of requests which were processed before
16528 this one in the server queue. It is zero when the request has not gone
16529 through the server queue. It makes it possible to estimate the approximate
16530 server's response time by dividing the time spent in queue by the number of
16531 requests in the queue. It is worth noting that if a session experiences a
16532 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016533 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016534 backend queue unless a redispatch occurs.
16535
16536 - "backend_queue" is the total number of requests which were processed before
16537 this one in the backend's global queue. It is zero when the request has not
16538 gone through the global queue. It makes it possible to estimate the average
16539 queue length, which easily translates into a number of missing servers when
16540 divided by a server's "maxconn" parameter. It is worth noting that if a
16541 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016542 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016543 through both the server queue and the backend queue unless a redispatch
16544 occurs.
16545
16546
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165478.2.3. HTTP log format
16548----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016549
16550The HTTP format is the most complete and the best suited for HTTP proxies. It
16551is enabled by when "option httplog" is specified in the frontend. It provides
16552the same level of information as the TCP format with additional features which
16553are specific to the HTTP protocol. Just like the TCP format, the log is usually
16554emitted at the end of the session, unless "option logasap" is specified, which
16555generally only makes sense for download sites. A session which matches the
16556"monitor" rules will never logged. It is also possible not to log sessions for
16557which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016558frontend. Successful connections will not be logged if "option dontlog-normal"
16559is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016560
16561Most fields are shared with the TCP log, some being different. A few fields may
16562slightly vary depending on some configuration options. Those ones are marked
16563with a star ('*') after the field name below.
16564
16565 Example :
16566 frontend http-in
16567 mode http
16568 option httplog
16569 log global
16570 default_backend bck
16571
16572 backend static
16573 server srv1 127.0.0.1:8000
16574
16575 >>> Feb 6 12:14:14 localhost \
16576 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16577 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 +010016578 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016579
16580 Field Format Extract from the example above
16581 1 process_name '[' pid ']:' haproxy[14389]:
16582 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016583 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016584 4 frontend_name http-in
16585 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016586 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016587 7 status_code 200
16588 8 bytes_read* 2750
16589 9 captured_request_cookie -
16590 10 captured_response_cookie -
16591 11 termination_state ----
16592 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16593 13 srv_queue '/' backend_queue 0/0
16594 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16595 15 '{' captured_response_headers* '}' {}
16596 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016597
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016598Detailed fields description :
16599 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016600 connection to haproxy. If the connection was accepted on a UNIX socket
16601 instead, the IP address would be replaced with the word "unix". Note that
16602 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016603 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016604 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016605 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016606
16607 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016608 If the connection was accepted on a UNIX socket instead, the port would be
16609 replaced with the ID of the accepting socket, which is also reported in the
16610 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016611
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016612 - "request_date" is the exact date when the first byte of the HTTP request
16613 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016614
16615 - "frontend_name" is the name of the frontend (or listener) which received
16616 and processed the connection.
16617
16618 - "backend_name" is the name of the backend (or listener) which was selected
16619 to manage the connection to the server. This will be the same as the
16620 frontend if no switching rule has been applied.
16621
16622 - "server_name" is the name of the last server to which the connection was
16623 sent, which might differ from the first one if there were connection errors
16624 and a redispatch occurred. Note that this server belongs to the backend
16625 which processed the request. If the request was aborted before reaching a
16626 server, "<NOSRV>" is indicated instead of a server name. If the request was
16627 intercepted by the stats subsystem, "<STATS>" is indicated instead.
16628
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016629 - "TR" is the total time in milliseconds spent waiting for a full HTTP
16630 request from the client (not counting body) after the first byte was
16631 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050016632 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016633 always be very small because a request generally fits in one single packet.
16634 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020016635 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
16636 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016637
16638 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16639 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020016640 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016641
16642 - "Tc" is the total time in milliseconds spent waiting for the connection to
16643 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020016644 request was aborted before a connection could be established. See section
16645 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016646
16647 - "Tr" is the total time in milliseconds spent waiting for the server to send
16648 a full HTTP response, not counting data. It can be "-1" if the request was
16649 aborted before a complete response could be received. It generally matches
16650 the server's processing time for the request, though it may be altered by
16651 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020016652 "GET" requests generally indicate an overloaded server. See section 8.4
16653 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016654
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016655 - "Ta" is the time the request remained active in haproxy, which is the total
16656 time in milliseconds elapsed between the first byte of the request was
16657 received and the last byte of response was sent. It covers all possible
16658 processing except the handshake (see Th) and idle time (see Ti). There is
16659 one exception, if "option logasap" was specified, then the time counting
16660 stops at the moment the log is emitted. In this case, a '+' sign is
16661 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020016662 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016663
16664 - "status_code" is the HTTP status code returned to the client. This status
16665 is generally set by the server, but it might also be set by haproxy when
16666 the server cannot be reached or when its response is blocked by haproxy.
16667
16668 - "bytes_read" is the total number of bytes transmitted to the client when
16669 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050016670 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016671 the final one may be larger. Please note that this value is a 64-bit
16672 counter, so log analysis tools must be able to handle it without
16673 overflowing.
16674
16675 - "captured_request_cookie" is an optional "name=value" entry indicating that
16676 the client had this cookie in the request. The cookie name and its maximum
16677 length are defined by the "capture cookie" statement in the frontend
16678 configuration. The field is a single dash ('-') when the option is not
16679 set. Only one cookie may be captured, it is generally used to track session
16680 ID exchanges between a client and a server to detect session crossing
16681 between clients due to application bugs. For more details, please consult
16682 the section "Capturing HTTP headers and cookies" below.
16683
16684 - "captured_response_cookie" is an optional "name=value" entry indicating
16685 that the server has returned a cookie with its response. The cookie name
16686 and its maximum length are defined by the "capture cookie" statement in the
16687 frontend configuration. The field is a single dash ('-') when the option is
16688 not set. Only one cookie may be captured, it is generally used to track
16689 session ID exchanges between a client and a server to detect session
16690 crossing between clients due to application bugs. For more details, please
16691 consult the section "Capturing HTTP headers and cookies" below.
16692
16693 - "termination_state" is the condition the session was in when the session
16694 ended. This indicates the session state, which side caused the end of
16695 session to happen, for what reason (timeout, error, ...), just like in TCP
16696 logs, and information about persistence operations on cookies in the last
16697 two characters. The normal flags should begin with "--", indicating the
16698 session was closed by either end with no data remaining in buffers. See
16699 below "Session state at disconnection" for more details.
16700
16701 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016702 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016703 limits have been reached. For instance, if actconn is close to 512 or 1024
16704 when multiple connection errors occur, chances are high that the system
16705 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016706 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016707 system.
16708
16709 - "feconn" is the total number of concurrent connections on the frontend when
16710 the session was logged. It is useful to estimate the amount of resource
16711 required to sustain high loads, and to detect when the frontend's "maxconn"
16712 has been reached. Most often when this value increases by huge jumps, it is
16713 because there is congestion on the backend servers, but sometimes it can be
16714 caused by a denial of service attack.
16715
16716 - "beconn" is the total number of concurrent connections handled by the
16717 backend when the session was logged. It includes the total number of
16718 concurrent connections active on servers as well as the number of
16719 connections pending in queues. It is useful to estimate the amount of
16720 additional servers needed to support high loads for a given application.
16721 Most often when this value increases by huge jumps, it is because there is
16722 congestion on the backend servers, but sometimes it can be caused by a
16723 denial of service attack.
16724
16725 - "srv_conn" is the total number of concurrent connections still active on
16726 the server when the session was logged. It can never exceed the server's
16727 configured "maxconn" parameter. If this value is very often close or equal
16728 to the server's "maxconn", it means that traffic regulation is involved a
16729 lot, meaning that either the server's maxconn value is too low, or that
16730 there aren't enough servers to process the load with an optimal response
16731 time. When only one of the server's "srv_conn" is high, it usually means
16732 that this server has some trouble causing the requests to take longer to be
16733 processed than on other servers.
16734
16735 - "retries" is the number of connection retries experienced by this session
16736 when trying to connect to the server. It must normally be zero, unless a
16737 server is being stopped at the same moment the connection was attempted.
16738 Frequent retries generally indicate either a network problem between
16739 haproxy and the server, or a misconfigured system backlog on the server
16740 preventing new connections from being queued. This field may optionally be
16741 prefixed with a '+' sign, indicating that the session has experienced a
16742 redispatch after the maximal retry count has been reached on the initial
16743 server. In this case, the server name appearing in the log is the one the
16744 connection was redispatched to, and not the first one, though both may
16745 sometimes be the same in case of hashing for instance. So as a general rule
16746 of thumb, when a '+' is present in front of the retry count, this count
16747 should not be attributed to the logged server.
16748
16749 - "srv_queue" is the total number of requests which were processed before
16750 this one in the server queue. It is zero when the request has not gone
16751 through the server queue. It makes it possible to estimate the approximate
16752 server's response time by dividing the time spent in queue by the number of
16753 requests in the queue. It is worth noting that if a session experiences a
16754 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016755 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016756 backend queue unless a redispatch occurs.
16757
16758 - "backend_queue" is the total number of requests which were processed before
16759 this one in the backend's global queue. It is zero when the request has not
16760 gone through the global queue. It makes it possible to estimate the average
16761 queue length, which easily translates into a number of missing servers when
16762 divided by a server's "maxconn" parameter. It is worth noting that if a
16763 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016764 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016765 through both the server queue and the backend queue unless a redispatch
16766 occurs.
16767
16768 - "captured_request_headers" is a list of headers captured in the request due
16769 to the presence of the "capture request header" statement in the frontend.
16770 Multiple headers can be captured, they will be delimited by a vertical bar
16771 ('|'). When no capture is enabled, the braces do not appear, causing a
16772 shift of remaining fields. It is important to note that this field may
16773 contain spaces, and that using it requires a smarter log parser than when
16774 it's not used. Please consult the section "Capturing HTTP headers and
16775 cookies" below for more details.
16776
16777 - "captured_response_headers" is a list of headers captured in the response
16778 due to the presence of the "capture response header" statement in the
16779 frontend. Multiple headers can be captured, they will be delimited by a
16780 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16781 causing a shift of remaining fields. It is important to note that this
16782 field may contain spaces, and that using it requires a smarter log parser
16783 than when it's not used. Please consult the section "Capturing HTTP headers
16784 and cookies" below for more details.
16785
16786 - "http_request" is the complete HTTP request line, including the method,
16787 request and HTTP version string. Non-printable characters are encoded (see
16788 below the section "Non-printable characters"). This is always the last
16789 field, and it is always delimited by quotes and is the only one which can
16790 contain quotes. If new fields are added to the log format, they will be
16791 added before this field. This field might be truncated if the request is
16792 huge and does not fit in the standard syslog buffer (1024 characters). This
16793 is the reason why this field must always remain the last one.
16794
16795
Cyril Bontédc4d9032012-04-08 21:57:39 +0200167968.2.4. Custom log format
16797------------------------
William Lallemand48940402012-01-30 16:47:22 +010016798
Willy Tarreau2beef582012-12-20 17:22:52 +010016799The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016800mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016801
Davor Ocelice9ed2812017-12-25 17:49:28 +010016802HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016803Variables can take arguments using braces ('{}'), and multiple arguments are
16804separated by commas within the braces. Flags may be added or removed by
16805prefixing them with a '+' or '-' sign.
16806
16807Special variable "%o" may be used to propagate its flags to all other
16808variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016809("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016810
Willy Tarreauc8368452012-12-21 00:09:23 +010016811If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016812as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016813less common information such as the client's SSL certificate's DN, or to log
16814the key that would be used to store an entry into a stick table.
16815
William Lallemand48940402012-01-30 16:47:22 +010016816Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016817In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016818in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016819
Dragan Dosen835b9212016-02-12 13:23:03 +010016820Note: when using the RFC5424 syslog message format, the characters '"',
16821'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16822https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16823such cases, the use of the flag "E" should be considered.
16824
William Lallemand48940402012-01-30 16:47:22 +010016825Flags are :
16826 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016827 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016828 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16829 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016830
16831 Example:
16832
16833 log-format %T\ %t\ Some\ Text
16834 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16835
Dragan Dosen835b9212016-02-12 13:23:03 +010016836 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16837
William Lallemand48940402012-01-30 16:47:22 +010016838At the moment, the default HTTP format is defined this way :
16839
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016840 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16841 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016842
William Lallemandbddd4fd2012-02-27 11:23:10 +010016843the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016844
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016845 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16846 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16847 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016848
William Lallemandbddd4fd2012-02-27 11:23:10 +010016849and the default TCP format is defined this way :
16850
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016851 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16852 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016853
William Lallemand48940402012-01-30 16:47:22 +010016854Please refer to the table below for currently defined variables :
16855
William Lallemandbddd4fd2012-02-27 11:23:10 +010016856 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016857 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016858 +---+------+-----------------------------------------------+-------------+
16859 | | %o | special variable, apply flags on all next var | |
16860 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016861 | | %B | bytes_read (from server to client) | numeric |
16862 | H | %CC | captured_request_cookie | string |
16863 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016864 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016865 | H | %HM | HTTP method (ex: POST) | string |
16866 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016867 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016868 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16869 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016870 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016871 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016872 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016873 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016874 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016875 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016876 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016877 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16878 | H | %Ti | idle time before the HTTP request | numeric |
16879 | H | %Tq | Th + Ti + TR | numeric |
16880 | H | %TR | time to receive the full request from 1st byte| numeric |
16881 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016882 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016883 | | %Tt | Tt | numeric |
16884 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016885 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016886 | | %ac | actconn | numeric |
16887 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016888 | | %bc | beconn (backend concurrent connections) | numeric |
16889 | | %bi | backend_source_ip (connecting address) | IP |
16890 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016891 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016892 | | %ci | client_ip (accepted address) | IP |
16893 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016894 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016895 | | %fc | feconn (frontend concurrent connections) | numeric |
16896 | | %fi | frontend_ip (accepting address) | IP |
16897 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016898 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016899 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016900 | | %hr | captured_request_headers default style | string |
16901 | | %hrl | captured_request_headers CLF style | string list |
16902 | | %hs | captured_response_headers default style | string |
16903 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016904 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016905 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016906 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016907 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016908 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016909 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016910 | | %sc | srv_conn (server concurrent connections) | numeric |
16911 | | %si | server_IP (target address) | IP |
16912 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016913 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016914 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16915 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016916 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016917 | H | %tr | date_time of HTTP request | date |
16918 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020016919 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016920 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016921 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016922 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016923
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016924 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016925
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016926
169278.2.5. Error log format
16928-----------------------
16929
16930When an incoming connection fails due to an SSL handshake or an invalid PROXY
16931protocol header, haproxy will log the event using a shorter, fixed line format.
16932By default, logs are emitted at the LOG_INFO level, unless the option
16933"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016934will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016935logged if the "dontlognull" option is set.
16936
16937The format looks like this :
16938
16939 >>> Dec 3 18:27:14 localhost \
16940 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
16941 Connection error during SSL handshake
16942
16943 Field Format Extract from the example above
16944 1 process_name '[' pid ']:' haproxy[6103]:
16945 2 client_ip ':' client_port 127.0.0.1:56059
16946 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
16947 4 frontend_name "/" bind_name ":" frt/f1:
16948 5 message Connection error during SSL handshake
16949
16950These fields just provide minimal information to help debugging connection
16951failures.
16952
16953
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169548.3. Advanced logging options
16955-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016956
16957Some advanced logging options are often looked for but are not easy to find out
16958just by looking at the various options. Here is an entry point for the few
16959options which can enable better logging. Please refer to the keywords reference
16960for more information about their usage.
16961
16962
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169638.3.1. Disabling logging of external tests
16964------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016965
16966It is quite common to have some monitoring tools perform health checks on
16967haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
16968commercial load-balancer, and sometimes it will simply be a more complete
16969monitoring system such as Nagios. When the tests are very frequent, users often
16970ask how to disable logging for those checks. There are three possibilities :
16971
16972 - if connections come from everywhere and are just TCP probes, it is often
16973 desired to simply disable logging of connections without data exchange, by
16974 setting "option dontlognull" in the frontend. It also disables logging of
16975 port scans, which may or may not be desired.
16976
16977 - if the connection come from a known source network, use "monitor-net" to
16978 declare this network as monitoring only. Any host in this network will then
16979 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016980 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016981 such as other load-balancers.
16982
16983 - if the tests are performed on a known URI, use "monitor-uri" to declare
16984 this URI as dedicated to monitoring. Any host sending this request will
16985 only get the result of a health-check, and the request will not be logged.
16986
16987
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169888.3.2. Logging before waiting for the session to terminate
16989----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016990
16991The problem with logging at end of connection is that you have no clue about
16992what is happening during very long sessions, such as remote terminal sessions
16993or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010016994"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016995just before data transfer begins. This means that in case of TCP, it will still
16996log the connection status to the server, and in case of HTTP, it will log just
16997after processing the server headers. In this case, the number of bytes reported
16998is the number of header bytes sent to the client. In order to avoid confusion
16999with normal logs, the total time field and the number of bytes are prefixed
17000with a '+' sign which means that real numbers are certainly larger.
17001
17002
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170038.3.3. Raising log level upon errors
17004------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017005
17006Sometimes it is more convenient to separate normal traffic from errors logs,
17007for instance in order to ease error monitoring from log files. When the option
17008"log-separate-errors" is used, connections which experience errors, timeouts,
17009retries, redispatches or HTTP status codes 5xx will see their syslog level
17010raised from "info" to "err". This will help a syslog daemon store the log in
17011a separate file. It is very important to keep the errors in the normal traffic
17012file too, so that log ordering is not altered. You should also be careful if
17013you already have configured your syslog daemon to store all logs higher than
17014"notice" in an "admin" file, because the "err" level is higher than "notice".
17015
17016
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170178.3.4. Disabling logging of successful connections
17018--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017019
17020Although this may sound strange at first, some large sites have to deal with
17021multiple thousands of logs per second and are experiencing difficulties keeping
17022them intact for a long time or detecting errors within them. If the option
17023"dontlog-normal" is set on the frontend, all normal connections will not be
17024logged. In this regard, a normal connection is defined as one without any
17025error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
17026and a response with a status 5xx is not considered normal and will be logged
17027too. Of course, doing is is really discouraged as it will remove most of the
17028useful information from the logs. Do this only if you have no other
17029alternative.
17030
17031
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170328.4. Timing events
17033------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017034
17035Timers provide a great help in troubleshooting network problems. All values are
17036reported in milliseconds (ms). These timers should be used in conjunction with
17037the session termination flags. In TCP mode with "option tcplog" set on the
17038frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017039mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
17040addition, three other measures are provided, "Th", "Ti", and "Tq".
17041
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010017042Timings events in HTTP mode:
17043
17044 first request 2nd request
17045 |<-------------------------------->|<-------------- ...
17046 t tr t tr ...
17047 ---|----|----|----|----|----|----|----|----|--
17048 : Th Ti TR Tw Tc Tr Td : Ti ...
17049 :<---- Tq ---->: :
17050 :<-------------- Tt -------------->:
17051 :<--------- Ta --------->:
17052
17053Timings events in TCP mode:
17054
17055 TCP session
17056 |<----------------->|
17057 t t
17058 ---|----|----|----|----|---
17059 | Th Tw Tc Td |
17060 |<------ Tt ------->|
17061
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017062 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017063 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017064 only happen once during the whole connection's lifetime. A large time here
17065 may indicate that the client only pre-established the connection without
17066 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017067 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017068 SSL handshake was very expensive to compute. Please note that this time is
17069 reported only before the first request, so it is safe to average it over
17070 all request to calculate the amortized value. The second and subsequent
17071 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017072
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017073 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17074 counts between the end of the handshakes and the first byte of the HTTP
17075 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017076 to count after the end of the transmission the previous response. When a
17077 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17078 after the previous request. Some browsers pre-establish connections to a
17079 server in order to reduce the latency of a future request, and keep them
17080 pending until they need it. This delay will be reported as the idle time. A
17081 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017082
17083 - TR: total time to get the client request (HTTP mode only). It's the time
17084 elapsed between the first bytes received and the moment the proxy received
17085 the empty line marking the end of the HTTP headers. The value "-1"
17086 indicates that the end of headers has never been seen. This happens when
17087 the client closes prematurely or times out. This time is usually very short
17088 since most requests fit in a single packet. A large time may indicate a
17089 request typed by hand during a test.
17090
17091 - Tq: total time to get the client request from the accept date or since the
17092 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017093 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 +020017094 returns -1 as well. This timer used to be very useful before the arrival of
17095 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17096 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17097 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017098
17099 - Tw: total time spent in the queues waiting for a connection slot. It
17100 accounts for backend queue as well as the server queues, and depends on the
17101 queue size, and the time needed for the server to complete previous
17102 requests. The value "-1" means that the request was killed before reaching
17103 the queue, which is generally what happens with invalid or denied requests.
17104
17105 - Tc: total time to establish the TCP connection to the server. It's the time
17106 elapsed between the moment the proxy sent the connection request, and the
17107 moment it was acknowledged by the server, or between the TCP SYN packet and
17108 the matching SYN/ACK packet in return. The value "-1" means that the
17109 connection never established.
17110
17111 - Tr: server response time (HTTP mode only). It's the time elapsed between
17112 the moment the TCP connection was established to the server and the moment
17113 the server sent its complete response headers. It purely shows its request
17114 processing time, without the network overhead due to the data transmission.
17115 It is worth noting that when the client has data to send to the server, for
17116 instance during a POST request, the time already runs, and this can distort
17117 apparent response time. For this reason, it's generally wise not to trust
17118 too much this field for POST requests initiated from clients behind an
17119 untrusted network. A value of "-1" here means that the last the response
17120 header (empty line) was never seen, most likely because the server timeout
17121 stroke before the server managed to process the request.
17122
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017123 - Ta: total active time for the HTTP request, between the moment the proxy
17124 received the first byte of the request header and the emission of the last
17125 byte of the response body. The exception is when the "logasap" option is
17126 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17127 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17128 by subtracting other timers when valid :
17129
17130 Td = Ta - (TR + Tw + Tc + Tr)
17131
17132 Timers with "-1" values have to be excluded from this equation. Note that
17133 "Ta" can never be negative.
17134
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017135 - Tt: total session duration time, between the moment the proxy accepted it
17136 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017137 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17138 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017139 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017140
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017141 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017142
17143 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017144 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17145 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017146
17147These timers provide precious indications on trouble causes. Since the TCP
17148protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17149that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017150due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17151"Tt" is close to a timeout value specified in the configuration, it often means
17152that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017153
17154Most common cases :
17155
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017156 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17157 the client and the proxy. This is very rare on local networks but might
17158 happen when clients are on far remote networks and send large requests. It
17159 may happen that values larger than usual appear here without any network
17160 cause. Sometimes, during an attack or just after a resource starvation has
17161 ended, haproxy may accept thousands of connections in a few milliseconds.
17162 The time spent accepting these connections will inevitably slightly delay
17163 processing of other connections, and it can happen that request times in the
17164 order of a few tens of milliseconds are measured after a few thousands of
17165 new connections have been accepted at once. Using one of the keep-alive
17166 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017167 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017168
17169 - If "Tc" is close to 3000, a packet has probably been lost between the
17170 server and the proxy during the server connection phase. This value should
17171 always be very low, such as 1 ms on local networks and less than a few tens
17172 of ms on remote networks.
17173
Willy Tarreau55165fe2009-05-10 12:02:55 +020017174 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17175 to be the average majored by 3000, there are probably some packets lost
17176 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017177
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017178 - If "Ta" is large even for small byte counts, it generally is because
17179 neither the client nor the server decides to close the connection while
17180 haproxy is running in tunnel mode and both have agreed on a keep-alive
17181 connection mode. In order to solve this issue, it will be needed to specify
17182 one of the HTTP options to manipulate keep-alive or close options on either
17183 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17184 important when connection regulation is used with the "maxconn" option on
17185 the servers, since no new connection will be sent to the server until
17186 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017187
17188Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17189
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017190 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017191 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017192 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017193
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017194 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017195 or it aborted too early. Check the session termination flags
17196 then "timeout http-request" and "timeout client" settings.
17197
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017198 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017199 servers were out of order, because the request was invalid
17200 or forbidden by ACL rules. Check the session termination
17201 flags.
17202
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017203 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17204 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017205 Check the session termination flags, then check the
17206 "timeout connect" setting. Note that the tarpit action might
17207 return similar-looking patterns, with "Tw" equal to the time
17208 the client connection was maintained open.
17209
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017210 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017211 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017212 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017213 termination flags, then check the "timeout server" setting.
17214
17215
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172168.5. Session state at disconnection
17217-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017218
17219TCP and HTTP logs provide a session termination indicator in the
17220"termination_state" field, just before the number of active connections. It is
172212-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17222each of which has a special meaning :
17223
17224 - On the first character, a code reporting the first event which caused the
17225 session to terminate :
17226
17227 C : the TCP session was unexpectedly aborted by the client.
17228
17229 S : the TCP session was unexpectedly aborted by the server, or the
17230 server explicitly refused it.
17231
17232 P : the session was prematurely aborted by the proxy, because of a
17233 connection limit enforcement, because a DENY filter was matched,
17234 because of a security check which detected and blocked a dangerous
17235 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010017236 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020017237
17238 L : the session was locally processed by haproxy and was not passed to
17239 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017240
17241 R : a resource on the proxy has been exhausted (memory, sockets, source
17242 ports, ...). Usually, this appears during the connection phase, and
17243 system logs should contain a copy of the precise error. If this
17244 happens, it must be considered as a very serious anomaly which
17245 should be fixed as soon as possible by any means.
17246
17247 I : an internal error was identified by the proxy during a self-check.
17248 This should NEVER happen, and you are encouraged to report any log
17249 containing this, because this would almost certainly be a bug. It
17250 would be wise to preventively restart the process after such an
17251 event too, in case it would be caused by memory corruption.
17252
Simon Horman752dc4a2011-06-21 14:34:59 +090017253 D : the session was killed by haproxy because the server was detected
17254 as down and was configured to kill all connections when going down.
17255
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070017256 U : the session was killed by haproxy on this backup server because an
17257 active server was detected as up and was configured to kill all
17258 backup connections when going up.
17259
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017260 K : the session was actively killed by an admin operating on haproxy.
17261
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017262 c : the client-side timeout expired while waiting for the client to
17263 send or receive data.
17264
17265 s : the server-side timeout expired while waiting for the server to
17266 send or receive data.
17267
17268 - : normal session completion, both the client and the server closed
17269 with nothing left in the buffers.
17270
17271 - on the second character, the TCP or HTTP session state when it was closed :
17272
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017273 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017274 (HTTP mode only). Nothing was sent to any server.
17275
17276 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17277 only happen when servers have a 'maxconn' parameter set. It can
17278 also happen in the global queue after a redispatch consecutive to
17279 a failed attempt to connect to a dying server. If no redispatch is
17280 reported, then no connection attempt was made to any server.
17281
17282 C : the proxy was waiting for the CONNECTION to establish on the
17283 server. The server might at most have noticed a connection attempt.
17284
17285 H : the proxy was waiting for complete, valid response HEADERS from the
17286 server (HTTP only).
17287
17288 D : the session was in the DATA phase.
17289
17290 L : the proxy was still transmitting LAST data to the client while the
17291 server had already finished. This one is very rare as it can only
17292 happen when the client dies while receiving the last packets.
17293
17294 T : the request was tarpitted. It has been held open with the client
17295 during the whole "timeout tarpit" duration or until the client
17296 closed, both of which will be reported in the "Tw" timer.
17297
17298 - : normal session completion after end of data transfer.
17299
17300 - the third character tells whether the persistence cookie was provided by
17301 the client (only in HTTP mode) :
17302
17303 N : the client provided NO cookie. This is usually the case for new
17304 visitors, so counting the number of occurrences of this flag in the
17305 logs generally indicate a valid trend for the site frequentation.
17306
17307 I : the client provided an INVALID cookie matching no known server.
17308 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017309 cookies between HTTP/HTTPS sites, persistence conditionally
17310 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017311
17312 D : the client provided a cookie designating a server which was DOWN,
17313 so either "option persist" was used and the client was sent to
17314 this server, or it was not set and the client was redispatched to
17315 another server.
17316
Willy Tarreau996a92c2010-10-13 19:30:47 +020017317 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017318 server.
17319
Willy Tarreau996a92c2010-10-13 19:30:47 +020017320 E : the client provided a valid cookie, but with a last date which was
17321 older than what is allowed by the "maxidle" cookie parameter, so
17322 the cookie is consider EXPIRED and is ignored. The request will be
17323 redispatched just as if there was no cookie.
17324
17325 O : the client provided a valid cookie, but with a first date which was
17326 older than what is allowed by the "maxlife" cookie parameter, so
17327 the cookie is consider too OLD and is ignored. The request will be
17328 redispatched just as if there was no cookie.
17329
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017330 U : a cookie was present but was not used to select the server because
17331 some other server selection mechanism was used instead (typically a
17332 "use-server" rule).
17333
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017334 - : does not apply (no cookie set in configuration).
17335
17336 - the last character reports what operations were performed on the persistence
17337 cookie returned by the server (only in HTTP mode) :
17338
17339 N : NO cookie was provided by the server, and none was inserted either.
17340
17341 I : no cookie was provided by the server, and the proxy INSERTED one.
17342 Note that in "cookie insert" mode, if the server provides a cookie,
17343 it will still be overwritten and reported as "I" here.
17344
Willy Tarreau996a92c2010-10-13 19:30:47 +020017345 U : the proxy UPDATED the last date in the cookie that was presented by
17346 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017347 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017348 date indicated in the cookie. If any other change happens, such as
17349 a redispatch, then the cookie will be marked as inserted instead.
17350
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017351 P : a cookie was PROVIDED by the server and transmitted as-is.
17352
17353 R : the cookie provided by the server was REWRITTEN by the proxy, which
17354 happens in "cookie rewrite" or "cookie prefix" modes.
17355
17356 D : the cookie provided by the server was DELETED by the proxy.
17357
17358 - : does not apply (no cookie set in configuration).
17359
Willy Tarreau996a92c2010-10-13 19:30:47 +020017360The combination of the two first flags gives a lot of information about what
17361was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017362helpful to detect server saturation, network troubles, local system resource
17363starvation, attacks, etc...
17364
17365The most common termination flags combinations are indicated below. They are
17366alphabetically sorted, with the lowercase set just after the upper case for
17367easier finding and understanding.
17368
17369 Flags Reason
17370
17371 -- Normal termination.
17372
17373 CC The client aborted before the connection could be established to the
17374 server. This can happen when haproxy tries to connect to a recently
17375 dead (or unchecked) server, and the client aborts while haproxy is
17376 waiting for the server to respond or for "timeout connect" to expire.
17377
17378 CD The client unexpectedly aborted during data transfer. This can be
17379 caused by a browser crash, by an intermediate equipment between the
17380 client and haproxy which decided to actively break the connection,
17381 by network routing issues between the client and haproxy, or by a
17382 keep-alive session between the server and the client terminated first
17383 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010017384
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017385 cD The client did not send nor acknowledge any data for as long as the
17386 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020017387 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017388
17389 CH The client aborted while waiting for the server to start responding.
17390 It might be the server taking too long to respond or the client
17391 clicking the 'Stop' button too fast.
17392
17393 cH The "timeout client" stroke while waiting for client data during a
17394 POST request. This is sometimes caused by too large TCP MSS values
17395 for PPPoE networks which cannot transport full-sized packets. It can
17396 also happen when client timeout is smaller than server timeout and
17397 the server takes too long to respond.
17398
17399 CQ The client aborted while its session was queued, waiting for a server
17400 with enough empty slots to accept it. It might be that either all the
17401 servers were saturated or that the assigned server was taking too
17402 long a time to respond.
17403
17404 CR The client aborted before sending a full HTTP request. Most likely
17405 the request was typed by hand using a telnet client, and aborted
17406 too early. The HTTP status code is likely a 400 here. Sometimes this
17407 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017408 and the client. "option http-ignore-probes" can be used to ignore
17409 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017410
17411 cR The "timeout http-request" stroke before the client sent a full HTTP
17412 request. This is sometimes caused by too large TCP MSS values on the
17413 client side for PPPoE networks which cannot transport full-sized
17414 packets, or by clients sending requests by hand and not typing fast
17415 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017416 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017417 some browsers started to implement a "pre-connect" feature consisting
17418 in speculatively connecting to some recently visited web sites just
17419 in case the user would like to visit them. This results in many
17420 connections being established to web sites, which end up in 408
17421 Request Timeout if the timeout strikes first, or 400 Bad Request when
17422 the browser decides to close them first. These ones pollute the log
17423 and feed the error counters. Some versions of some browsers have even
17424 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017425 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017426 http-ignore-probes" in the frontend, resulting in connections with
17427 zero data transfer to be totally ignored. This will definitely hide
17428 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017429
17430 CT The client aborted while its session was tarpitted. It is important to
17431 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017432 wrong tarpit rules have been written. If a lot of them happen, it
17433 might make sense to lower the "timeout tarpit" value to something
17434 closer to the average reported "Tw" timer, in order not to consume
17435 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017436
Willy Tarreau570f2212013-06-10 16:42:09 +020017437 LR The request was intercepted and locally handled by haproxy. Generally
17438 it means that this was a redirect or a stats request.
17439
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017440 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017441 the TCP connection (the proxy received a TCP RST or an ICMP message
17442 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017443 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017444 or no ARP response on local network). When this happens in HTTP mode,
17445 the status code is likely a 502 or 503 here.
17446
17447 sC The "timeout connect" stroke before a connection to the server could
17448 complete. When this happens in HTTP mode, the status code is likely a
17449 503 or 504 here.
17450
17451 SD The connection to the server died with an error during the data
17452 transfer. This usually means that haproxy has received an RST from
17453 the server or an ICMP message from an intermediate equipment while
17454 exchanging data with the server. This can be caused by a server crash
17455 or by a network issue on an intermediate equipment.
17456
17457 sD The server did not send nor acknowledge any data for as long as the
17458 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017459 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017460 load-balancers, ...), as well as keep-alive sessions maintained
17461 between the client and the server expiring first on haproxy.
17462
17463 SH The server aborted before sending its full HTTP response headers, or
17464 it crashed while processing the request. Since a server aborting at
17465 this moment is very rare, it would be wise to inspect its logs to
17466 control whether it crashed and why. The logged request may indicate a
17467 small set of faulty requests, demonstrating bugs in the application.
17468 Sometimes this might also be caused by an IDS killing the connection
17469 between haproxy and the server.
17470
17471 sH The "timeout server" stroke before the server could return its
17472 response headers. This is the most common anomaly, indicating too
17473 long transactions, probably caused by server or database saturation.
17474 The immediate workaround consists in increasing the "timeout server"
17475 setting, but it is important to keep in mind that the user experience
17476 will suffer from these long response times. The only long term
17477 solution is to fix the application.
17478
17479 sQ The session spent too much time in queue and has been expired. See
17480 the "timeout queue" and "timeout connect" settings to find out how to
17481 fix this if it happens too often. If it often happens massively in
17482 short periods, it may indicate general problems on the affected
17483 servers due to I/O or database congestion, or saturation caused by
17484 external attacks.
17485
17486 PC The proxy refused to establish a connection to the server because the
17487 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017488 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017489 so that it does not happen anymore. This status is very rare and
17490 might happen when the global "ulimit-n" parameter is forced by hand.
17491
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017492 PD The proxy blocked an incorrectly formatted chunked encoded message in
17493 a request or a response, after the server has emitted its headers. In
17494 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017495 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017496 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017497
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017498 PH The proxy blocked the server's response, because it was invalid,
17499 incomplete, dangerous (cache control), or matched a security filter.
17500 In any case, an HTTP 502 error is sent to the client. One possible
17501 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017502 containing unauthorized characters. It is also possible but quite
17503 rare, that the proxy blocked a chunked-encoding request from the
17504 client due to an invalid syntax, before the server responded. In this
17505 case, an HTTP 400 error is sent to the client and reported in the
17506 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017507
17508 PR The proxy blocked the client's HTTP request, either because of an
17509 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17510 the client, or because a deny filter matched, in which case it
17511 returned an HTTP 403 error.
17512
17513 PT The proxy blocked the client's request and has tarpitted its
17514 connection before returning it a 500 server error. Nothing was sent
17515 to the server. The connection was maintained open for as long as
17516 reported by the "Tw" timer field.
17517
17518 RC A local resource has been exhausted (memory, sockets, source ports)
17519 preventing the connection to the server from establishing. The error
17520 logs will tell precisely what was missing. This is very rare and can
17521 only be solved by proper system tuning.
17522
Willy Tarreau996a92c2010-10-13 19:30:47 +020017523The combination of the two last flags gives a lot of information about how
17524persistence was handled by the client, the server and by haproxy. This is very
17525important to troubleshoot disconnections, when users complain they have to
17526re-authenticate. The commonly encountered flags are :
17527
17528 -- Persistence cookie is not enabled.
17529
17530 NN No cookie was provided by the client, none was inserted in the
17531 response. For instance, this can be in insert mode with "postonly"
17532 set on a GET request.
17533
17534 II A cookie designating an invalid server was provided by the client,
17535 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017536 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017537 value can be presented by a client when no other server knows it.
17538
17539 NI No cookie was provided by the client, one was inserted in the
17540 response. This typically happens for first requests from every user
17541 in "insert" mode, which makes it an easy way to count real users.
17542
17543 VN A cookie was provided by the client, none was inserted in the
17544 response. This happens for most responses for which the client has
17545 already got a cookie.
17546
17547 VU A cookie was provided by the client, with a last visit date which is
17548 not completely up-to-date, so an updated cookie was provided in
17549 response. This can also happen if there was no date at all, or if
17550 there was a date but the "maxidle" parameter was not set, so that the
17551 cookie can be switched to unlimited time.
17552
17553 EI A cookie was provided by the client, with a last visit date which is
17554 too old for the "maxidle" parameter, so the cookie was ignored and a
17555 new cookie was inserted in the response.
17556
17557 OI A cookie was provided by the client, with a first visit date which is
17558 too old for the "maxlife" parameter, so the cookie was ignored and a
17559 new cookie was inserted in the response.
17560
17561 DI The server designated by the cookie was down, a new server was
17562 selected and a new cookie was emitted in the response.
17563
17564 VI The server designated by the cookie was not marked dead but could not
17565 be reached. A redispatch happened and selected another one, which was
17566 then advertised in the response.
17567
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017568
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175698.6. Non-printable characters
17570-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017571
17572In order not to cause trouble to log analysis tools or terminals during log
17573consulting, non-printable characters are not sent as-is into log files, but are
17574converted to the two-digits hexadecimal representation of their ASCII code,
17575prefixed by the character '#'. The only characters that can be logged without
17576being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17577escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17578is the same for the character '"' which becomes "#22", as well as '{', '|' and
17579'}' when logging headers.
17580
17581Note that the space character (' ') is not encoded in headers, which can cause
17582issues for tools relying on space count to locate fields. A typical header
17583containing spaces is "User-Agent".
17584
17585Last, it has been observed that some syslog daemons such as syslog-ng escape
17586the quote ('"') with a backslash ('\'). The reverse operation can safely be
17587performed since no quote may appear anywhere else in the logs.
17588
17589
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175908.7. Capturing HTTP cookies
17591---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017592
17593Cookie capture simplifies the tracking a complete user session. This can be
17594achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017595section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017596cookie will simultaneously be checked in the request ("Cookie:" header) and in
17597the response ("Set-Cookie:" header). The respective values will be reported in
17598the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017599locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017600not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17601user switches to a new session for example, because the server will reassign it
17602a new cookie. It is also possible to detect if a server unexpectedly sets a
17603wrong cookie to a client, leading to session crossing.
17604
17605 Examples :
17606 # capture the first cookie whose name starts with "ASPSESSION"
17607 capture cookie ASPSESSION len 32
17608
17609 # capture the first cookie whose name is exactly "vgnvisitor"
17610 capture cookie vgnvisitor= len 32
17611
17612
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176138.8. Capturing HTTP headers
17614---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017615
17616Header captures are useful to track unique request identifiers set by an upper
17617proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
17618the response, one can search for information about the response length, how the
17619server asked the cache to behave, or an object location during a redirection.
17620
17621Header captures are performed using the "capture request header" and "capture
17622response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017623section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017624
17625It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017626time. Non-existent headers are logged as empty strings, and if one header
17627appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017628are grouped within braces '{' and '}' in the same order as they were declared,
17629and delimited with a vertical bar '|' without any space. Response headers
17630follow the same representation, but are displayed after a space following the
17631request headers block. These blocks are displayed just before the HTTP request
17632in the logs.
17633
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017634As a special case, it is possible to specify an HTTP header capture in a TCP
17635frontend. The purpose is to enable logging of headers which will be parsed in
17636an HTTP backend if the request is then switched to this HTTP backend.
17637
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017638 Example :
17639 # This instance chains to the outgoing proxy
17640 listen proxy-out
17641 mode http
17642 option httplog
17643 option logasap
17644 log global
17645 server cache1 192.168.1.1:3128
17646
17647 # log the name of the virtual server
17648 capture request header Host len 20
17649
17650 # log the amount of data uploaded during a POST
17651 capture request header Content-Length len 10
17652
17653 # log the beginning of the referrer
17654 capture request header Referer len 20
17655
17656 # server name (useful for outgoing proxies only)
17657 capture response header Server len 20
17658
17659 # logging the content-length is useful with "option logasap"
17660 capture response header Content-Length len 10
17661
Davor Ocelice9ed2812017-12-25 17:49:28 +010017662 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017663 capture response header Cache-Control len 8
17664
17665 # the Via header will report the next proxy's name
17666 capture response header Via len 20
17667
17668 # log the URL location during a redirection
17669 capture response header Location len 20
17670
17671 >>> Aug 9 20:26:09 localhost \
17672 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
17673 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
17674 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
17675 "GET http://fr.adserver.yahoo.com/"
17676
17677 >>> Aug 9 20:30:46 localhost \
17678 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
17679 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
17680 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017681 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017682
17683 >>> Aug 9 20:30:46 localhost \
17684 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
17685 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
17686 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
17687 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017688 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017689
17690
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176918.9. Examples of logs
17692---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017693
17694These are real-world examples of logs accompanied with an explanation. Some of
17695them have been made up by hand. The syslog part has been removed for better
17696reading. Their sole purpose is to explain how to decipher them.
17697
17698 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17699 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17700 "HEAD / HTTP/1.0"
17701
17702 => long request (6.5s) entered by hand through 'telnet'. The server replied
17703 in 147 ms, and the session ended normally ('----')
17704
17705 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17706 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17707 0/9 "HEAD / HTTP/1.0"
17708
17709 => Idem, but the request was queued in the global queue behind 9 other
17710 requests, and waited there for 1230 ms.
17711
17712 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17713 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17714 "GET /image.iso HTTP/1.0"
17715
17716 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017717 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017718 14 ms, 243 bytes of headers were sent to the client, and total time from
17719 accept to first data byte is 30 ms.
17720
17721 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17722 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17723 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17724
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020017725 => the proxy blocked a server response either because of an "http-response
17726 deny" rule, or because the response was improperly formatted and not
17727 HTTP-compliant, or because it blocked sensitive information which risked
17728 being cached. In this case, the response is replaced with a "502 bad
17729 gateway". The flags ("PH--") tell us that it was haproxy who decided to
17730 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017731
17732 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017733 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 +010017734
17735 => the client never completed its request and aborted itself ("C---") after
17736 8.5s, while the proxy was waiting for the request headers ("-R--").
17737 Nothing was sent to any server.
17738
17739 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17740 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17741
17742 => The client never completed its request, which was aborted by the
17743 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017744 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017745 send a 408 return code to the client.
17746
17747 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17748 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17749
17750 => This log was produced with "option tcplog". The client timed out after
17751 5 seconds ("c----").
17752
17753 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17754 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017755 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017756
17757 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017758 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017759 (config says 'retries 3'), and no redispatch (otherwise we would have
17760 seen "/+3"). Status code 503 was returned to the client. There were 115
17761 connections on this server, 202 connections on this proxy, and 205 on
17762 the global process. It is possible that the server refused the
17763 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017764
Willy Tarreau52b2d222011-09-07 23:48:48 +020017765
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200177669. Supported filters
17767--------------------
17768
17769Here are listed officially supported filters with the list of parameters they
17770accept. Depending on compile options, some of these filters might be
17771unavailable. The list of available filters is reported in haproxy -vv.
17772
17773See also : "filter"
17774
177759.1. Trace
17776----------
17777
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017778filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017779
17780 Arguments:
17781 <name> is an arbitrary name that will be reported in
17782 messages. If no name is provided, "TRACE" is used.
17783
17784 <random-parsing> enables the random parsing of data exchanged between
17785 the client and the server. By default, this filter
17786 parses all available data. With this parameter, it
17787 only parses a random amount of the available data.
17788
Davor Ocelice9ed2812017-12-25 17:49:28 +010017789 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017790 default, this filter forwards all previously parsed
17791 data. With this parameter, it only forwards a random
17792 amount of the parsed data.
17793
Davor Ocelice9ed2812017-12-25 17:49:28 +010017794 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017795
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017796This filter can be used as a base to develop new filters. It defines all
17797callbacks and print a message on the standard error stream (stderr) with useful
17798information for all of them. It may be useful to debug the activity of other
17799filters or, quite simply, HAProxy's activity.
17800
17801Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17802tests the behavior of a filter that parses data exchanged between a client and
17803a server by adding some latencies in the processing.
17804
17805
178069.2. HTTP compression
17807---------------------
17808
17809filter compression
17810
17811The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17812keyword must still be used to enable and configure the HTTP compression. And
Christopher Faulet27d93c32018-12-15 22:32:02 +010017813when no other filter is used, it is enough. When used with the cache enabled,
17814it is also enough. In this case, the compression is always done after the
17815response is stored in the cache. But it is mandatory to explicitly use a filter
17816line to enable the HTTP compression when at least one filter other than the
17817cache is used for the same listener/frontend/backend. This is important to know
17818the filters evaluation order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017819
Christopher Faulet27d93c32018-12-15 22:32:02 +010017820See also : "compression" and section 9.4 about the cache filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017821
17822
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200178239.3. Stream Processing Offload Engine (SPOE)
17824--------------------------------------------
17825
17826filter spoe [engine <name>] config <file>
17827
17828 Arguments :
17829
17830 <name> is the engine name that will be used to find the right scope in
17831 the configuration file. If not provided, all the file will be
17832 parsed.
17833
17834 <file> is the path of the engine configuration file. This file can
17835 contain configuration of several engines. In this case, each
17836 part must be placed in its own scope.
17837
17838The Stream Processing Offload Engine (SPOE) is a filter communicating with
17839external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017840streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017841exchanged with them are configured in dedicated files, for the main part. It
17842also requires dedicated backends, defined in HAProxy configuration.
17843
17844SPOE communicates with external components using an in-house binary protocol,
17845the Stream Processing Offload Protocol (SPOP).
17846
Tim Düsterhus4896c442016-11-29 02:15:19 +010017847For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017848"doc/SPOE.txt".
17849
17850Important note:
17851 The SPOE filter is highly experimental for now and was not heavily
17852 tested. It is really not production ready. So use it carefully.
17853
Christopher Faulet99a17a22018-12-11 09:18:27 +0100178549.4. Cache
17855----------
17856
17857filter cache <name>
17858
17859 Arguments :
17860
17861 <name> is name of the cache section this filter will use.
17862
17863The cache uses a filter to store cacheable responses. The HTTP rules
17864"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050017865cache. By default the corresponding filter is implicitly defined. And when no
Christopher Faulet27d93c32018-12-15 22:32:02 +010017866other filters than cache or compression are used, it is enough. In such case,
17867the compression filter is always evaluated after the cache filter. But it is
17868mandatory to explicitly use a filter line to use a cache when at least one
17869filter other than the compression is used for the same
17870listener/frontend/backend. This is important to know the filters evaluation
17871order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010017872
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020017873See also : section 9.2 about the compression filter and section 6 about cache.
William Lallemand86d0df02017-11-24 21:36:45 +010017874
Willy Tarreau0ba27502007-12-24 16:55:16 +010017875/*
17876 * Local variables:
17877 * fill-column: 79
17878 * End:
17879 */