1 ## Copyright (C) 1996-2020 The Squid Software Foundation and contributors
3 ## Squid software is distributed under GPLv2+ license and includes
4 ## contributions from numerous individuals and organizations.
5 ## Please see the COPYING and CONTRIBUTORS files for details.
10 ----------------------------
12 This is the documentation for the Squid configuration file.
13 This documentation can also be found online at:
14 http://www.squid-cache.org/Doc/config/
16 You may wish to look at the Squid home page and wiki for the
17 FAQ and other documentation:
18 http://www.squid-cache.org/
19 http://wiki.squid-cache.org/SquidFaq
20 http://wiki.squid-cache.org/ConfigExamples
22 This documentation shows what the defaults for various directives
23 happen to be. If you don't need to change the default, you should
24 leave the line out of your squid.conf in most cases.
26 In some cases "none" refers to no default setting at all,
27 while in other cases it refers to the value of the option
28 - the comments for that keyword indicate if this is the case.
33 Configuration options can be included using the "include" directive.
34 Include takes a list of files to include. Quoting and wildcards are
39 include /path/to/included/file/squid.acl.config
41 Includes can be nested up to a hard-coded depth of 16 levels.
42 This arbitrary restriction is to prevent recursive include references
43 from causing Squid entering an infinite loop whilst trying to load
46 Values with byte units
48 Squid accepts size units on some size related directives. All
49 such directives are documented with a default value displaying
52 Units accepted by Squid are:
54 KB - Kilobyte (1024 bytes)
58 Values with spaces, quotes, and other special characters
60 Squid supports directive parameters with spaces, quotes, and other
61 special characters. Surround such parameters with "double quotes". Use
62 the configuration_includes_quoted_values directive to enable or
65 Squid supports reading configuration option parameters from external
66 files using the syntax:
67 parameters("/path/filename")
69 acl whitelist dstdomain parameters("/etc/squid/whitelist.txt")
71 Conditional configuration
73 If-statements can be used to make configuration directives
77 ... regular configuration directives ...
79 ... regular configuration directives ...]
82 The else part is optional. The keywords "if", "else", and "endif"
83 must be typed on their own lines, as if they were regular
84 configuration directives.
86 NOTE: An else-if condition is not supported.
88 These individual conditions types are supported:
91 Always evaluates to true.
93 Always evaluates to false.
95 Equality comparison of two integer numbers.
100 The following SMP-related preprocessor macros can be used.
102 ${process_name} expands to the current Squid process "name"
103 (e.g., squid1, squid2, or cache1).
105 ${process_number} expands to the current Squid process
106 identifier, which is an integer number (e.g., 1, 2, 3) unique
107 across all Squid processes of the current service instance.
109 ${service_name} expands into the current Squid service instance
110 name identifier which is provided by -n on the command line.
114 Logformat macros can be used in many places outside of the logformat
115 directive. In theory, all of the logformat codes can be used as %macros,
116 where they are supported. In practice, a %macro expands as a dash (-) when
117 the transaction does not yet have enough information and a value is needed.
119 There is no definitive list of what tokens are available at the various
120 stages of the transaction.
122 And some information may already be available to Squid but not yet
123 committed where the macro expansion code can access it (report
124 such instances!). The macro will be expanded into a single dash
125 ('-') in such cases. Not all macros have been tested.
129 # options still not yet ported from 2.7 to 3.x
130 NAME: broken_vary_encoding
133 This option is not yet supported by Squid-3.
139 This option is not yet supported by Squid-3.
145 This option is not yet supported by Squid-3.
148 NAME: external_refresh_check
151 This option is not yet supported by Squid-3.
154 NAME: location_rewrite_program location_rewrite_access location_rewrite_children location_rewrite_concurrency
157 This option is not yet supported by Squid-3.
160 NAME: refresh_stale_hit
163 This option is not yet supported by Squid-3.
166 # Options removed in 4.x
167 NAME: cache_peer_domain cache_host_domain
170 Replace with dstdomain ACLs and cache_peer_access.
176 Remove this line. The behaviour enabled by this is no longer needed.
179 NAME: sslproxy_cafile
182 Remove this line. Use tls_outgoing_options cafile= instead.
185 NAME: sslproxy_capath
188 Remove this line. Use tls_outgoing_options capath= instead.
191 NAME: sslproxy_cipher
194 Remove this line. Use tls_outgoing_options cipher= instead.
197 NAME: sslproxy_client_certificate
200 Remove this line. Use tls_outgoing_options cert= instead.
203 NAME: sslproxy_client_key
206 Remove this line. Use tls_outgoing_options key= instead.
212 Remove this line. Use tls_outgoing_options flags= instead.
215 NAME: sslproxy_options
218 Remove this line. Use tls_outgoing_options options= instead.
221 NAME: sslproxy_version
224 Remove this line. Use tls_outgoing_options options= instead.
227 # Options removed in 3.5
228 NAME: hierarchy_stoplist
231 Remove this line. Use always_direct or cache_peer_access ACLs instead if you need to prevent cache_peer use.
234 # Options removed in 3.4
238 Remove this line. Use acls with access_log directives to control access logging
244 Remove this line. Use acls with icap_log directives to control icap logging
247 # Options Removed in 3.3
248 NAME: ignore_ims_on_miss
251 Remove this line. The HTTP/1.1 feature is now configured by 'cache_miss_revalidate'.
254 # Options Removed in 3.2
255 NAME: balance_on_multiple_ip
258 Remove this line. Squid performs a 'Happy Eyeballs' algorithm, this multiple-IP algorithm is not longer relevant.
261 NAME: chunked_request_body_max_size
264 Remove this line. Squid is now HTTP/1.1 compliant.
267 NAME: dns_v4_fallback
270 Remove this line. Squid performs a 'Happy Eyeballs' algorithm, the 'fallback' algorithm is no longer relevant.
273 NAME: emulate_httpd_log
276 Replace this with an access_log directive using the format 'common' or 'combined'.
282 Use a regular access.log with ACL limiting it to MISS events.
288 Remove this line. Configure FTP page display using the CSS controls in errorpages.css instead.
291 NAME: ignore_expect_100
294 Remove this line. The HTTP/1.1 feature is now fully supported by default.
300 Remove this option from your config. To log FQDN use %>A in the log format.
303 NAME: log_ip_on_direct
306 Remove this option from your config. To log server or peer names use %<A in the log format.
309 NAME: maximum_single_addr_tries
312 Replaced by connect_retries. The behaviour has changed, please read the documentation before altering.
315 NAME: referer_log referrer_log
318 Replace this with an access_log directive using the format 'referrer'.
324 Remove this line. The feature is supported by default in storage types where update is implemented.
327 NAME: url_rewrite_concurrency
330 Remove this line. Set the 'concurrency=' option of url_rewrite_children instead.
336 Replace this with an access_log directive using the format 'useragent'.
339 # Options Removed in 3.1
343 Remove this line. DNS is no longer tested on startup.
346 NAME: extension_methods
349 Remove this line. All valid methods for HTTP are accepted by default.
352 # 2.7 Options Removed/Replaced in 3.2
357 # 2.7 Options Removed/Replaced in 3.1
365 Remove this line. HTTP/1.1 is supported by default.
368 NAME: upgrade_http0.9
371 Remove this line. ICY/1.0 streaming protocol is supported by default.
374 NAME: zph_local zph_mode zph_option zph_parent zph_sibling
377 Alter these entries. Use the qos_flows directive instead.
380 # Options Removed in 3.0
384 Since squid-3.0 replace with request_header_access or reply_header_access
385 depending on whether you wish to match client requests or server replies.
388 NAME: httpd_accel_no_pmtu_disc
391 Since squid-3.0 use the 'disable-pmtu-discovery' flag on http_port instead.
394 NAME: wais_relay_host
397 Replace this line with 'cache_peer' configuration.
400 NAME: wais_relay_port
403 Replace this line with 'cache_peer' configuration.
408 -----------------------------------------------------------------------------
415 DEFAULT_DOC: SMP support disabled.
417 Number of main Squid processes or "workers" to fork and maintain.
418 0: "no daemon" mode, like running "squid -N ..."
419 1: "no SMP" mode, start one main Squid process daemon (default)
420 N: start N main Squid process daemons (i.e., SMP mode)
422 In SMP mode, each worker does nearly all what a single Squid daemon
423 does (e.g., listen on http_port and forward HTTP requests).
426 NAME: cpu_affinity_map
428 LOC: Config.cpuAffinityMap
430 DEFAULT_DOC: Let operating system decide.
432 Usage: cpu_affinity_map process_numbers=P1,P2,... cores=C1,C2,...
434 Sets 1:1 mapping between Squid processes and CPU cores. For example,
436 cpu_affinity_map process_numbers=1,2,3,4 cores=1,3,5,7
438 affects processes 1 through 4 only and places them on the first
439 four even cores, starting with core #1.
441 CPU cores are numbered starting from 1. Requires support for
442 sched_getaffinity(2) and sched_setaffinity(2) system calls.
444 Multiple cpu_affinity_map options are merged.
449 NAME: shared_memory_locking
452 LOC: Config.shmLocking
455 Whether to ensure that all required shared memory is available by
456 "locking" that shared memory into RAM when Squid starts. The
457 alternative is faster startup time followed by slightly slower
458 performance and, if not enough RAM is actually available during
459 runtime, mysterious crashes.
461 SMP Squid uses many shared memory segments. These segments are
462 brought into Squid memory space using an mmap(2) system call. During
463 Squid startup, the mmap() call often succeeds regardless of whether
464 the system has enough RAM. In general, Squid cannot tell whether the
465 kernel applies this "optimistic" memory allocation policy (but
466 popular modern kernels usually use it).
468 Later, if Squid attempts to actually access the mapped memory
469 regions beyond what the kernel is willing to allocate, the
470 "optimistic" kernel simply kills Squid kid with a SIGBUS signal.
471 Some of the memory limits enforced by the kernel are currently
472 poorly understood: We do not know how to detect and check them. This
473 option ensures that the mapped memory will be available.
475 This option may have a positive performance side-effect: Locking
476 memory at start avoids runtime paging I/O. Paging slows Squid down.
478 Locking memory may require a large enough RLIMIT_MEMLOCK OS limit,
479 CAP_IPC_LOCK capability, or equivalent.
482 NAME: hopeless_kid_revival_delay
485 LOC: Config.hopelessKidRevivalDelay
488 Normally, when a kid process dies, Squid immediately restarts the
489 kid. A kid experiencing frequent deaths is marked as "hopeless" for
490 the duration specified by this directive. Hopeless kids are not
491 automatically restarted.
493 Currently, zero values are not supported because they result in
494 misconfigured SMP Squid instances running forever, endlessly
495 restarting each dying kid. To effectively disable hopeless kids
496 revival, set the delay to a huge value (e.g., 1 year).
498 Reconfiguration also clears all hopeless kids designations, allowing
499 for manual revival of hopeless kids.
503 OPTIONS FOR AUTHENTICATION
504 -----------------------------------------------------------------------------
510 LOC: Auth::TheConfig.schemes
513 This is used to define parameters for the various authentication
514 schemes supported by Squid.
516 format: auth_param scheme parameter [setting]
518 The order in which authentication schemes are presented to the client is
519 dependent on the order the scheme first appears in config file. IE
520 has a bug (it's not RFC 2617 compliant) in that it will use the basic
521 scheme if basic is the first entry presented, even if more secure
522 schemes are presented. For now use the order in the recommended
523 settings section below. If other browsers have difficulties (don't
524 recognize the schemes offered even if you are using basic) either
525 put basic first, or disable the other schemes (by commenting out their
528 Once an authentication scheme is fully configured, it can only be
529 shutdown by shutting squid down and restarting. Changes can be made on
530 the fly and activated with a reconfigure. I.E. You can change to a
531 different helper, but not unconfigure the helper completely.
533 Please note that while this directive defines how Squid processes
534 authentication it does not automatically activate authentication.
535 To use authentication you must in addition make use of ACLs based
536 on login name in http_access (proxy_auth, proxy_auth_regex or
537 external with %LOGIN used in the format tag). The browser will be
538 challenged for authentication on the first such acl encountered
539 in http_access processing and will also be re-challenged for new
540 login credentials if the request is being denied by a proxy_auth
543 WARNING: authentication can't be used in a transparently intercepting
544 proxy as the client then thinks it is talking to an origin server and
545 not the proxy. This is a limitation of bending the TCP/IP protocol to
546 transparently intercepting port 80, not a limitation in Squid.
547 Ports flagged 'transparent', 'intercept', or 'tproxy' have
548 authentication disabled.
550 === Parameters common to all schemes. ===
553 Specifies the command for the external authenticator.
555 By default, each authentication scheme is not used unless a
556 program is specified.
558 See http://wiki.squid-cache.org/Features/AddonHelpers for
559 more details on helper operations and creating your own.
562 Specifies a string to be append to request line format for
563 the authentication helper. "Quoted" format values may contain
564 spaces and logformat %macros. In theory, any logformat %macro
565 can be used. In practice, a %macro expands as a dash (-) if
566 the helper request is sent before the required macro
567 information is available to Squid.
569 By default, Squid uses request formats provided in
570 scheme-specific examples below (search for %credentials).
572 The expanded key_extras value is added to the Squid credentials
573 cache and, hence, will affect authentication. It can be used to
574 autenticate different users with identical user names (e.g.,
575 when user authentication depends on http_port).
577 Avoid adding frequently changing information to key_extras. For
578 example, if you add user source IP, and it changes frequently
579 in your environment, then max_user_ip ACL is going to treat
580 every user+IP combination as a unique "user", breaking the ACL
581 and wasting a lot of memory on those user records. It will also
582 force users to authenticate from scratch whenever their IP
586 Specifies the protection scope (aka realm name) which is to be
587 reported to the client for the authentication scheme. It is
588 commonly part of the text the user will see when prompted for
589 their username and password.
591 For Basic the default is "Squid proxy-caching web server".
592 For Digest there is no default, this parameter is mandatory.
593 For NTLM and Negotiate this parameter is ignored.
595 "children" numberofchildren [startup=N] [idle=N] [concurrency=N]
596 [queue-size=N] [on-persistent-overload=action]
597 [reservation-timeout=seconds]
599 The maximum number of authenticator processes to spawn. If
600 you start too few Squid will have to wait for them to process
601 a backlog of credential verifications, slowing it down. When
602 password verifications are done via a (slow) network you are
603 likely to need lots of authenticator processes.
605 The startup= and idle= options permit some skew in the exact
606 amount run. A minimum of startup=N will begin during startup
607 and reconfigure. Squid will start more in groups of up to
608 idle=N in an attempt to meet traffic needs and to keep idle=N
609 free above those traffic needs up to the maximum.
611 The concurrency= option sets the number of concurrent requests
612 the helper can process. The default of 0 is used for helpers
613 who only supports one request at a time. Setting this to a
614 number greater than 0 changes the protocol used to include a
615 channel ID field first on the request/response line, allowing
616 multiple requests to be sent to the same helper in parallel
617 without waiting for the response.
619 Concurrency must not be set unless it's known the helper
620 supports the input format with channel-ID fields.
622 The queue-size option sets the maximum number of queued
623 requests. A request is queued when no existing child can
624 accept it due to concurrency limit and no new child can be
625 started due to numberofchildren limit. The default maximum is
626 2*numberofchildren. Squid is allowed to temporarily exceed the
627 configured maximum, marking the affected helper as
628 "overloaded". If the helper overload lasts more than 3
629 minutes, the action prescribed by the on-persistent-overload
632 The on-persistent-overload=action option specifies Squid
633 reaction to a new helper request arriving when the helper
634 has been overloaded for more that 3 minutes already. The number
635 of queued requests determines whether the helper is overloaded
636 (see the queue-size option).
638 Two actions are supported:
640 die Squid worker quits. This is the default behavior.
642 ERR Squid treats the helper request as if it was
643 immediately submitted, and the helper immediately
644 replied with an ERR response. This action has no effect
645 on the already queued and in-progress helper requests.
647 NOTE: NTLM and Negotiate schemes do not support concurrency
648 in the Squid code module even though some helpers can.
650 The reservation-timeout=seconds option allows NTLM and Negotiate
651 helpers to forget about clients that abandon their in-progress
652 connection authentication without closing the connection. The
653 timeout is measured since the last helper response received by
654 Squid for the client. Fractional seconds are not supported.
656 After the timeout, the helper will be used for other clients if
657 there are no unreserved helpers available. In the latter case,
658 the old client attempt to resume authentication will not be
659 forwarded to the helper (and the client should open a new HTTP
660 connection and retry authentication from scratch).
662 By default, reservations do not expire and clients that keep
663 their connections open without completing authentication may
664 exhaust all NTLM and Negotiate helpers.
667 If you experience problems with PUT/POST requests when using
668 the NTLM or Negotiate schemes then you can try setting this
669 to off. This will cause Squid to forcibly close the connection
670 on the initial request where the browser asks which schemes
671 are supported by the proxy.
673 For Basic and Digest this parameter is ignored.
676 Useful for sending credentials to authentication backends that
677 expect UTF-8 encoding (e.g., LDAP).
679 When this option is enabled, Squid uses HTTP Accept-Language
680 request header to guess the received credentials encoding
681 (ISO-Latin-1, CP1251, or UTF-8) and then converts the first
682 two encodings into UTF-8.
684 When this option is disabled and by default, Squid sends
685 credentials in their original (i.e. received) encoding.
687 This parameter is only honored for Basic and Digest schemes.
688 For Basic, the entire username:password credentials are
689 checked and, if necessary, re-encoded. For Digest -- just the
690 username component. For NTLM and Negotiate schemes, this
691 parameter is ignored.
693 IF HAVE_AUTH_MODULE_BASIC
694 === Basic authentication parameters ===
696 "credentialsttl" timetolive
697 Specifies how long squid assumes an externally validated
698 username:password pair is valid for - in other words how
699 often the helper program is called for that user. Set this
700 low to force revalidation with short lived passwords.
702 NOTE: setting this high does not impact your susceptibility
703 to replay attacks unless you are using an one-time password
704 system (such as SecureID). If you are using such a system,
705 you will be vulnerable to replay attacks unless you also
706 use the max_user_ip ACL in an http_access rule.
708 "casesensitive" on|off
709 Specifies if usernames are case sensitive. Most user databases
710 are case insensitive allowing the same username to be spelled
711 using both lower and upper case letters, but some are case
712 sensitive. This makes a big difference for user_max_ip ACL
713 processing and similar.
716 IF HAVE_AUTH_MODULE_DIGEST
717 === Digest authentication parameters ===
719 "nonce_garbage_interval" timeinterval
720 Specifies the interval that nonces that have been issued
721 to client_agent's are checked for validity.
723 "nonce_max_duration" timeinterval
724 Specifies the maximum length of time a given nonce will be
727 "nonce_max_count" number
728 Specifies the maximum number of times a given nonce can be
731 "nonce_strictness" on|off
732 Determines if squid requires strict increment-by-1 behavior
733 for nonce counts, or just incrementing (off - for use when
734 user agents generate nonce counts that occasionally miss 1
735 (ie, 1,2,4,6)). Default off.
737 "check_nonce_count" on|off
738 This directive if set to off can disable the nonce count check
739 completely to work around buggy digest qop implementations in
740 certain mainstream browser versions. Default on to check the
741 nonce count to protect from authentication replay attacks.
743 "post_workaround" on|off
744 This is a workaround to certain buggy browsers who send an
745 incorrect request digest in POST requests when reusing the
746 same nonce as acquired earlier on a GET request.
750 === Example Configuration ===
752 This configuration displays the recommended authentication scheme
753 order from most to least secure with recommended minimum configuration
754 settings for each scheme:
756 #auth_param negotiate program <uncomment and complete this line to activate>
757 #auth_param negotiate children 20 startup=0 idle=1
759 #auth_param digest program <uncomment and complete this line to activate>
760 #auth_param digest children 20 startup=0 idle=1
761 #auth_param digest realm Squid proxy-caching web server
762 #auth_param digest nonce_garbage_interval 5 minutes
763 #auth_param digest nonce_max_duration 30 minutes
764 #auth_param digest nonce_max_count 50
766 #auth_param ntlm program <uncomment and complete this line to activate>
767 #auth_param ntlm children 20 startup=0 idle=1
769 #auth_param basic program <uncomment and complete this line>
770 #auth_param basic children 5 startup=5 idle=1
771 #auth_param basic credentialsttl 2 hours
774 NAME: authenticate_cache_garbage_interval
778 LOC: Auth::TheConfig.garbageCollectInterval
780 The time period between garbage collection across the username cache.
781 This is a trade-off between memory utilization (long intervals - say
782 2 days) and CPU (short intervals - say 1 minute). Only change if you
786 NAME: authenticate_ttl
790 LOC: Auth::TheConfig.credentialsTtl
792 The time a user & their credentials stay in the logged in
793 user cache since their last request. When the garbage
794 interval passes, all user credentials that have passed their
795 TTL are removed from memory.
798 NAME: authenticate_ip_ttl
801 LOC: Auth::TheConfig.ipTtl
804 If you use proxy authentication and the 'max_user_ip' ACL,
805 this directive controls how long Squid remembers the IP
806 addresses associated with each user. Use a small value
807 (e.g., 60 seconds) if your users might change addresses
808 quickly, as is the case with dialup. You might be safe
809 using a larger value (e.g., 2 hours) in a corporate LAN
810 environment with relatively static address assignments.
815 -----------------------------------------------------------------------------
818 NAME: external_acl_type
819 TYPE: externalAclHelper
820 LOC: Config.externalAclHelperList
823 This option defines external acl classes using a helper program
824 to look up the status
826 external_acl_type name [options] FORMAT /path/to/helper [helper arguments]
830 ttl=n TTL in seconds for cached results (defaults to 3600
834 TTL for cached negative lookups (default same
837 grace=n Percentage remaining of TTL where a refresh of a
838 cached entry should be initiated without needing to
839 wait for a new reply. (default is for no grace period)
841 cache=n The maximum number of entries in the result cache. The
842 default limit is 262144 entries. Each cache entry usually
843 consumes at least 256 bytes. Squid currently does not remove
844 expired cache entries until the limit is reached, so a proxy
845 will sooner or later reach the limit. The expanded FORMAT
846 value is used as the cache key, so if the details in FORMAT
847 are highly variable, a larger cache may be needed to produce
848 reduction in helper load.
851 Maximum number of acl helper processes spawned to service
852 external acl lookups of this type. (default 5)
855 Minimum number of acl helper processes to spawn during
856 startup and reconfigure to service external acl lookups
857 of this type. (default 0)
860 Number of acl helper processes to keep ahead of traffic
861 loads. Squid will spawn this many at once whenever load
862 rises above the capabilities of existing processes.
863 Up to the value of children-max. (default 1)
865 concurrency=n concurrency level per process. Only used with helpers
866 capable of processing more than one query at a time.
868 queue-size=N The queue-size option sets the maximum number of
869 queued requests. A request is queued when no existing
870 helper can accept it due to concurrency limit and no
871 new helper can be started due to children-max limit.
872 If the queued requests exceed queue size, the acl is
873 ignored. The default value is set to 2*children-max.
875 protocol=2.5 Compatibility mode for Squid-2.5 external acl helpers.
877 ipv4 / ipv6 IP protocol used to communicate with this helper.
878 The default is to auto-detect IPv6 and use it when available.
881 FORMAT is a series of %macro codes. See logformat directive for a full list
882 of the accepted codes. Although note that at the time of any external ACL
883 being tested data may not be available and thus some %macro expand to '-'.
885 In addition to the logformat codes; when processing external ACLs these
886 additional macros are made available:
888 %ACL The name of the ACL being tested.
890 %DATA The ACL arguments specified in the referencing config
891 'acl ... external' line, separated by spaces (an
892 "argument string"). see acl external.
894 If there are no ACL arguments %DATA expands to '-'.
896 If you do not specify a DATA macro inside FORMAT,
897 Squid automatically appends %DATA to your FORMAT.
898 Note that Squid-3.x may expand %DATA to whitespace
899 or nothing in this case.
901 By default, Squid applies URL-encoding to each ACL
902 argument inside the argument string. If an explicit
903 encoding modifier is used (e.g., %#DATA), then Squid
904 encodes the whole argument string as a single token
905 (e.g., with %#DATA, spaces between arguments become
908 If SSL is enabled, the following formating codes become available:
910 %USER_CERT SSL User certificate in PEM format
911 %USER_CERTCHAIN SSL User certificate chain in PEM format
912 %USER_CERT_xx SSL User certificate subject attribute xx
913 %USER_CA_CERT_xx SSL User certificate issuer attribute xx
916 NOTE: all other format codes accepted by older Squid versions
920 General request syntax:
922 [channel-ID] FORMAT-values
925 FORMAT-values consists of transaction details expanded with
926 whitespace separation per the config file FORMAT specification
927 using the FORMAT macros listed above.
929 Request values sent to the helper are URL escaped to protect
930 each value in requests against whitespaces.
932 If using protocol=2.5 then the request sent to the helper is not
933 URL escaped to protect against whitespace.
935 NOTE: protocol=3.0 is deprecated as no longer necessary.
937 When using the concurrency= option the protocol is changed by
938 introducing a query channel tag in front of the request/response.
939 The query channel tag is a number between 0 and concurrency-1.
940 This value must be echoed back unchanged to Squid as the first part
941 of the response relating to its request.
944 The helper receives lines expanded per the above format specification
945 and for each input line returns 1 line starting with OK/ERR/BH result
946 code and optionally followed by additional keywords with more details.
949 General result syntax:
951 [channel-ID] result keyword=value ...
953 Result consists of one of the codes:
956 the ACL test produced a match.
959 the ACL test does not produce a match.
962 An internal error occurred in the helper, preventing
963 a result being identified.
965 The meaning of 'a match' is determined by your squid.conf
966 access control configuration. See the Squid wiki for details.
970 user= The users name (login)
972 password= The users password (for login= cache_peer option)
974 message= Message describing the reason for this response.
975 Available as %o in error pages.
976 Useful on (ERR and BH results).
978 tag= Apply a tag to a request. Only sets a tag once,
979 does not alter existing tags.
981 log= String to be logged in access.log. Available as
982 %ea in logformat specifications.
984 clt_conn_tag= Associates a TAG with the client TCP connection.
985 Please see url_rewrite_program related documentation
988 Any keywords may be sent on any response whether OK, ERR or BH.
990 All response keyword values need to be a single token with URL
991 escaping, or enclosed in double quotes (") and escaped using \ on
992 any double quotes or \ characters within the value. The wrapping
993 double quotes are removed before the value is interpreted by Squid.
994 \r and \n are also replace by CR and LF.
996 Some example key values:
1000 user="J. \"Bob\" Smith"
1007 DEFAULT: ssl::certHasExpired ssl_error X509_V_ERR_CERT_HAS_EXPIRED
1008 DEFAULT: ssl::certNotYetValid ssl_error X509_V_ERR_CERT_NOT_YET_VALID
1009 DEFAULT: ssl::certDomainMismatch ssl_error SQUID_X509_V_ERR_DOMAIN_MISMATCH
1010 DEFAULT: ssl::certUntrusted ssl_error X509_V_ERR_INVALID_CA X509_V_ERR_SELF_SIGNED_CERT_IN_CHAIN X509_V_ERR_UNABLE_TO_VERIFY_LEAF_SIGNATURE X509_V_ERR_UNABLE_TO_GET_ISSUER_CERT X509_V_ERR_UNABLE_TO_GET_ISSUER_CERT_LOCALLY X509_V_ERR_CERT_UNTRUSTED
1011 DEFAULT: ssl::certSelfSigned ssl_error X509_V_ERR_DEPTH_ZERO_SELF_SIGNED_CERT
1013 DEFAULT: all src all
1014 DEFAULT: manager url_regex -i ^cache_object:// +i ^https?://[^/]+/squid-internal-mgr/
1015 DEFAULT: localhost src 127.0.0.1/32 ::1
1016 DEFAULT: to_localhost dst 127.0.0.0/8 0.0.0.0/32 ::1/128 ::/128
1017 DEFAULT: CONNECT method CONNECT
1018 DEFAULT_DOC: ACLs all, manager, localhost, to_localhost, and CONNECT are predefined.
1020 Defining an Access List
1022 Every access list definition must begin with an aclname and acltype,
1023 followed by either type-specific arguments or a quoted filename that
1026 acl aclname acltype argument ...
1027 acl aclname acltype "file" ...
1029 When using "file", the file should contain one item per line.
1034 Some acl types supports options which changes their default behaviour:
1036 -i,+i By default, regular expressions are CASE-SENSITIVE. To make them
1037 case-insensitive, use the -i option. To return case-sensitive
1038 use the +i option between patterns, or make a new ACL line
1041 -n Disable lookups and address type conversions. If lookup or
1042 conversion is required because the parameter type (IP or
1043 domain name) does not match the message address type (domain
1044 name or IP), then the ACL would immediately declare a mismatch
1045 without any warnings or lookups.
1048 Perform a list membership test, interpreting values as
1049 comma-separated token lists and matching against individual
1050 tokens instead of whole values.
1051 The optional "delimiters" parameter specifies one or more
1052 alternative non-alphanumeric delimiter characters.
1053 non-alphanumeric delimiter characters.
1055 -- Used to stop processing all options, in the case the first acl
1056 value has '-' character as first character (for example the '-'
1057 is a valid domain name)
1059 Some acl types require suspending the current request in order
1060 to access some external data source.
1061 Those which do are marked with the tag [slow], those which
1062 don't are marked as [fast].
1063 See http://wiki.squid-cache.org/SquidFaq/SquidAcl
1064 for further information
1066 ***** ACL TYPES AVAILABLE *****
1068 acl aclname src ip-address/mask ... # clients IP address [fast]
1069 acl aclname src addr1-addr2/mask ... # range of addresses [fast]
1070 acl aclname dst [-n] ip-address/mask ... # URL host's IP address [slow]
1071 acl aclname localip ip-address/mask ... # IP address the client connected to [fast]
1074 acl aclname arp mac-address ...
1075 acl aclname eui64 eui64-address ...
1077 # MAC (EUI-48) and EUI-64 addresses use xx:xx:xx:xx:xx:xx notation.
1079 # The 'arp' ACL code is not portable to all operating systems.
1080 # It works on Linux, Solaris, Windows, FreeBSD, and some other
1083 # The eui_lookup directive is required to be 'on' (the default)
1084 # and Squid built with --enable-eui for MAC/EUI addresses to be
1085 # available for this ACL.
1087 # Squid can only determine the MAC/EUI address for IPv4
1088 # clients that are on the same subnet. If the client is on a
1089 # different subnet, then Squid cannot find out its address.
1091 # IPv6 protocol does not contain ARP. MAC/EUI is either
1092 # encoded directly in the IPv6 address or not available.
1094 acl aclname clientside_mark mark[/mask] ...
1095 # matches CONNMARK of an accepted connection [fast]
1096 # DEPRECATED. Use the 'client_connection_mark' instead.
1098 acl aclname client_connection_mark mark[/mask] ...
1099 # matches CONNMARK of an accepted connection [fast]
1101 # mark and mask are unsigned integers (hex, octal, or decimal).
1102 # If multiple marks are given, then the ACL matches if at least
1105 # Uses netfilter-conntrack library.
1106 # Requires building Squid with --enable-linux-netfilter.
1108 # The client, various intermediaries, and Squid itself may set
1109 # CONNMARK at various times. The last CONNMARK set wins. This ACL
1110 # checks the mark present on an accepted connection or set by
1111 # Squid afterwards, depending on the ACL check timing. This ACL
1112 # effectively ignores any mark set by other agents after Squid has
1113 # accepted the connection.
1115 acl aclname srcdomain .foo.com ...
1116 # reverse lookup, from client IP [slow]
1117 acl aclname dstdomain [-n] .foo.com ...
1118 # Destination server from URL [fast]
1119 acl aclname srcdom_regex [-i] \.foo\.com ...
1120 # regex matching client name [slow]
1121 acl aclname dstdom_regex [-n] [-i] \.foo\.com ...
1122 # regex matching server [fast]
1124 # For dstdomain and dstdom_regex a reverse lookup is tried if a IP
1125 # based URL is used and no match is found. The name "none" is used
1126 # if the reverse lookup fails.
1128 acl aclname src_as number ...
1129 acl aclname dst_as number ...
1131 # Except for access control, AS numbers can be used for
1132 # routing of requests to specific caches. Here's an
1133 # example for routing all requests for AS#1241 and only
1134 # those to mycache.mydomain.net:
1135 # acl asexample dst_as 1241
1136 # cache_peer_access mycache.mydomain.net allow asexample
1137 # cache_peer_access mycache_mydomain.net deny all
1139 acl aclname peername myPeer ...
1140 acl aclname peername_regex [-i] regex-pattern ...
1142 # match against a named cache_peer entry
1143 # set unique name= on cache_peer lines for reliable use.
1145 acl aclname time [day-abbrevs] [h1:m1-h2:m2]
1155 # h1:m1 must be less than h2:m2
1157 acl aclname url_regex [-i] ^http:// ...
1158 # regex matching on whole URL [fast]
1159 acl aclname urllogin [-i] [^a-zA-Z0-9] ...
1160 # regex matching on URL login field
1161 acl aclname urlpath_regex [-i] \.gif$ ...
1162 # regex matching on URL path [fast]
1164 acl aclname port 80 70 21 0-1024... # destination TCP port [fast]
1166 acl aclname localport 3128 ... # TCP port the client connected to [fast]
1167 # NP: for interception mode this is usually '80'
1169 acl aclname myportname 3128 ... # *_port name [fast]
1171 acl aclname proto HTTP FTP ... # request protocol [fast]
1173 acl aclname method GET POST ... # HTTP request method [fast]
1175 acl aclname http_status 200 301 500- 400-403 ...
1176 # status code in reply [fast]
1178 acl aclname browser [-i] regexp ...
1179 # pattern match on User-Agent header (see also req_header below) [fast]
1181 acl aclname referer_regex [-i] regexp ...
1182 # pattern match on Referer header [fast]
1183 # Referer is highly unreliable, so use with care
1185 acl aclname ident [-i] username ...
1186 acl aclname ident_regex [-i] pattern ...
1187 # string match on ident output [slow]
1188 # use REQUIRED to accept any non-null ident.
1190 acl aclname proxy_auth [-i] username ...
1191 acl aclname proxy_auth_regex [-i] pattern ...
1192 # perform http authentication challenge to the client and match against
1193 # supplied credentials [slow]
1195 # takes a list of allowed usernames.
1196 # use REQUIRED to accept any valid username.
1198 # Will use proxy authentication in forward-proxy scenarios, and plain
1199 # http authenticaiton in reverse-proxy scenarios
1201 # NOTE: when a Proxy-Authentication header is sent but it is not
1202 # needed during ACL checking the username is NOT logged
1205 # NOTE: proxy_auth requires a EXTERNAL authentication program
1206 # to check username/password combinations (see
1207 # auth_param directive).
1209 # NOTE: proxy_auth can't be used in a transparent/intercepting proxy
1210 # as the browser needs to be configured for using a proxy in order
1211 # to respond to proxy authentication.
1213 acl aclname snmp_community string ...
1214 # A community string to limit access to your SNMP Agent [fast]
1217 # acl snmppublic snmp_community public
1219 acl aclname maxconn number
1220 # This will be matched when the client's IP address has
1221 # more than <number> TCP connections established. [fast]
1222 # NOTE: This only measures direct TCP links so X-Forwarded-For
1223 # indirect clients are not counted.
1225 acl aclname max_user_ip [-s] number
1226 # This will be matched when the user attempts to log in from more
1227 # than <number> different ip addresses. The authenticate_ip_ttl
1228 # parameter controls the timeout on the ip entries. [fast]
1229 # If -s is specified the limit is strict, denying browsing
1230 # from any further IP addresses until the ttl has expired. Without
1231 # -s Squid will just annoy the user by "randomly" denying requests.
1232 # (the counter is reset each time the limit is reached and a
1233 # request is denied)
1234 # NOTE: in acceleration mode or where there is mesh of child proxies,
1235 # clients may appear to come from multiple addresses if they are
1236 # going through proxy farms, so a limit of 1 may cause user problems.
1238 acl aclname random probability
1239 # Pseudo-randomly match requests. Based on the probability given.
1240 # Probability may be written as a decimal (0.333), fraction (1/3)
1241 # or ratio of matches:non-matches (3:5).
1243 acl aclname req_mime_type [-i] mime-type ...
1244 # regex match against the mime type of the request generated
1245 # by the client. Can be used to detect file upload or some
1246 # types HTTP tunneling requests [fast]
1247 # NOTE: This does NOT match the reply. You cannot use this
1248 # to match the returned file type.
1250 acl aclname req_header header-name [-i] any\.regex\.here
1251 # regex match against any of the known request headers. May be
1252 # thought of as a superset of "browser", "referer" and "mime-type"
1255 acl aclname rep_mime_type [-i] mime-type ...
1256 # regex match against the mime type of the reply received by
1257 # squid. Can be used to detect file download or some
1258 # types HTTP tunneling requests. [fast]
1259 # NOTE: This has no effect in http_access rules. It only has
1260 # effect in rules that affect the reply data stream such as
1261 # http_reply_access.
1263 acl aclname rep_header header-name [-i] any\.regex\.here
1264 # regex match against any of the known reply headers. May be
1265 # thought of as a superset of "browser", "referer" and "mime-type"
1268 acl aclname external class_name [arguments...]
1269 # external ACL lookup via a helper class defined by the
1270 # external_acl_type directive [slow]
1272 acl aclname user_cert attribute values...
1273 # match against attributes in a user SSL certificate
1274 # attribute is one of DN/C/O/CN/L/ST or a numerical OID [fast]
1276 acl aclname ca_cert attribute values...
1277 # match against attributes a users issuing CA SSL certificate
1278 # attribute is one of DN/C/O/CN/L/ST or a numerical OID [fast]
1280 acl aclname ext_user [-i] username ...
1281 acl aclname ext_user_regex [-i] pattern ...
1282 # string match on username returned by external acl helper [slow]
1283 # use REQUIRED to accept any non-null user name.
1285 acl aclname tag tagvalue ...
1286 # string match on tag returned by external acl helper [fast]
1287 # DEPRECATED. Only the first tag will match with this ACL.
1288 # Use the 'note' ACL instead for handling multiple tag values.
1290 acl aclname hier_code codename ...
1291 # string match against squid hierarchy code(s); [fast]
1292 # e.g., DIRECT, PARENT_HIT, NONE, etc.
1294 # NOTE: This has no effect in http_access rules. It only has
1295 # effect in rules that affect the reply data stream such as
1296 # http_reply_access.
1298 acl aclname note [-m[=delimiters]] name [value ...]
1299 # match transaction annotation [fast]
1300 # Without values, matches any annotation with a given name.
1301 # With value(s), matches any annotation with a given name that
1302 # also has one of the given values.
1303 # If the -m flag is used, then the value of the named
1304 # annotation is interpreted as a list of tokens, and the ACL
1305 # matches individual name=token pairs rather than whole
1306 # name=value pairs. See "ACL Options" above for more info.
1307 # Annotation sources include note and adaptation_meta directives
1308 # as well as helper and eCAP responses.
1310 acl aclname annotate_transaction [-m[=delimiters]] key=value ...
1311 acl aclname annotate_transaction [-m[=delimiters]] key+=value ...
1312 # Always matches. [fast]
1313 # Used for its side effect: This ACL immediately adds a
1314 # key=value annotation to the current master transaction.
1315 # The added annotation can then be tested using note ACL and
1316 # logged (or sent to helpers) using %note format code.
1318 # Annotations can be specified using replacement and addition
1319 # formats. The key=value form replaces old same-key annotation
1320 # value(s). The key+=value form appends a new value to the old
1321 # same-key annotation. Both forms create a new key=value
1322 # annotation if no same-key annotation exists already. If
1323 # -m flag is used, then the value is interpreted as a list
1324 # and the annotation will contain key=token pair(s) instead of the
1325 # whole key=value pair.
1327 # This ACL is especially useful for recording complex multi-step
1328 # ACL-driven decisions. For example, the following configuration
1329 # avoids logging transactions accepted after aclX matched:
1331 # # First, mark transactions accepted after aclX matched
1332 # acl markSpecial annotate_transaction special=true
1333 # http_access allow acl001
1335 # http_access deny acl100
1336 # http_access allow aclX markSpecial
1338 # # Second, do not log marked transactions:
1339 # acl markedSpecial note special true
1340 # access_log ... deny markedSpecial
1342 # # Note that the following would not have worked because aclX
1343 # # alone does not determine whether the transaction was allowed:
1344 # access_log ... deny aclX # Wrong!
1346 # Warning: This ACL annotates the transaction even when negated
1347 # and even if subsequent ACLs fail to match. For example, the
1348 # following three rules will have exactly the same effect as far
1349 # as annotations set by the "mark" ACL are concerned:
1351 # some_directive acl1 ... mark # rule matches if mark is reached
1352 # some_directive acl1 ... !mark # rule never matches
1353 # some_directive acl1 ... mark !all # rule never matches
1355 acl aclname annotate_client [-m[=delimiters]] key=value ...
1356 acl aclname annotate_client [-m[=delimiters]] key+=value ...
1358 # Always matches. [fast]
1359 # Used for its side effect: This ACL immediately adds a
1360 # key=value annotation to the current client-to-Squid
1361 # connection. Connection annotations are propagated to the current
1362 # and all future master transactions on the annotated connection.
1363 # See the annotate_transaction ACL for details.
1365 # For example, the following configuration avoids rewriting URLs
1366 # of transactions bumped by SslBump:
1368 # # First, mark bumped connections:
1369 # acl markBumped annotate_client bumped=true
1370 # ssl_bump peek acl1
1371 # ssl_bump stare acl2
1372 # ssl_bump bump acl3 markBumped
1373 # ssl_bump splice all
1375 # # Second, do not send marked transactions to the redirector:
1376 # acl markedBumped note bumped true
1377 # url_rewrite_access deny markedBumped
1379 # # Note that the following would not have worked because acl3 alone
1380 # # does not determine whether the connection is going to be bumped:
1381 # url_rewrite_access deny acl3 # Wrong!
1383 acl aclname adaptation_service service ...
1384 # Matches the name of any icap_service, ecap_service,
1385 # adaptation_service_set, or adaptation_service_chain that Squid
1386 # has used (or attempted to use) for the master transaction.
1387 # This ACL must be defined after the corresponding adaptation
1388 # service is named in squid.conf. This ACL is usable with
1389 # adaptation_meta because it starts matching immediately after
1390 # the service has been selected for adaptation.
1392 acl aclname transaction_initiator initiator ...
1393 # Matches transaction's initiator [fast]
1395 # Supported initiators are:
1396 # esi: matches transactions fetching ESI resources
1397 # certificate-fetching: matches transactions fetching
1398 # a missing intermediate TLS certificate
1399 # cache-digest: matches transactions fetching Cache Digests
1401 # htcp: matches HTCP requests from peers
1402 # icp: matches ICP requests to peers
1403 # icmp: matches ICMP RTT database (NetDB) requests to peers
1404 # asn: matches asns db requests
1405 # internal: matches any of the above
1406 # client: matches transactions containing an HTTP or FTP
1407 # client request received at a Squid *_port
1408 # all: matches any transaction, including internal transactions
1409 # without a configurable initiator and hopefully rare
1410 # transactions without a known-to-Squid initiator
1412 # Multiple initiators are ORed.
1414 acl aclname has component
1415 # matches a transaction "component" [fast]
1417 # Supported transaction components are:
1418 # request: transaction has a request header (at least)
1419 # response: transaction has a response header (at least)
1420 # ALE: transaction has an internally-generated Access Log Entry
1421 # structure; bugs notwithstanding, all transaction have it
1423 # For example, the following configuration helps when dealing with HTTP
1424 # clients that close connections without sending a request header:
1426 # acl hasRequest has request
1427 # acl logMe note important_transaction
1428 # # avoid "logMe ACL is used in context without an HTTP request" warnings
1429 # access_log ... logformat=detailed hasRequest logMe
1430 # # log request-less transactions, instead of ignoring them
1431 # access_log ... logformat=brief !hasRequest
1433 # Multiple components are not supported for one "acl" rule, but
1434 # can be specified (and are ORed) using multiple same-name rules:
1436 # # OK, this strange logging daemon needs request or response,
1437 # # but can work without either a request or a response:
1438 # acl hasWhatMyLoggingDaemonNeeds has request
1439 # acl hasWhatMyLoggingDaemonNeeds has response
1441 acl aclname at_step step
1442 # match against the current request processing step [fast]
1444 # GeneratingCONNECT: Generating HTTP CONNECT request headers
1446 # The following ssl_bump processing steps are recognized:
1447 # SslBump1: After getting TCP-level and HTTP CONNECT info.
1448 # SslBump2: After getting SSL Client Hello info.
1449 # SslBump3: After getting SSL Server Hello info.
1453 acl aclname ssl_error errorname
1454 # match against SSL certificate validation error [fast]
1456 # For valid error names see in @DEFAULT_ERROR_DIR@/templates/error-details.txt
1459 # The following can be used as shortcuts for certificate properties:
1460 # [ssl::]certHasExpired: the "not after" field is in the past
1461 # [ssl::]certNotYetValid: the "not before" field is in the future
1462 # [ssl::]certUntrusted: The certificate issuer is not to be trusted.
1463 # [ssl::]certSelfSigned: The certificate is self signed.
1464 # [ssl::]certDomainMismatch: The certificate CN domain does not
1465 # match the name the name of the host we are connecting to.
1467 # The ssl::certHasExpired, ssl::certNotYetValid, ssl::certDomainMismatch,
1468 # ssl::certUntrusted, and ssl::certSelfSigned can also be used as
1469 # predefined ACLs, just like the 'all' ACL.
1471 # NOTE: The ssl_error ACL is only supported with sslproxy_cert_error,
1472 # sslproxy_cert_sign, and sslproxy_cert_adapt options.
1474 acl aclname server_cert_fingerprint [-sha1] fingerprint
1475 # match against server SSL certificate fingerprint [fast]
1477 # The fingerprint is the digest of the DER encoded version
1478 # of the whole certificate. The user should use the form: XX:XX:...
1479 # Optional argument specifies the digest algorithm to use.
1480 # The SHA1 digest algorithm is the default and is currently
1481 # the only algorithm supported (-sha1).
1483 acl aclname ssl::server_name [option] .foo.com ...
1484 # matches server name obtained from various sources [fast]
1486 # The ACL computes server name(s) using such information sources as
1487 # CONNECT request URI, TLS client SNI, and TLS server certificate
1488 # subject (CN and SubjectAltName). The computed server name(s) usually
1489 # change with each SslBump step, as more info becomes available:
1490 # * SNI is used as the server name instead of the request URI,
1491 # * subject name(s) from the server certificate (CN and
1492 # SubjectAltName) are used as the server names instead of SNI.
1494 # When the ACL computes multiple server names, matching any single
1495 # computed name is sufficient for the ACL to match.
1497 # The "none" name can be used to match transactions where the ACL
1498 # could not compute the server name using any information source
1499 # that was both available and allowed to be used by the ACL options at
1500 # the ACL evaluation time.
1502 # Unlike dstdomain, this ACL does not perform DNS lookups.
1504 # An ACL option below may be used to restrict what information
1505 # sources are used to extract the server names from:
1507 # --client-requested
1508 # The server name is SNI regardless of what the server says.
1510 # The server name(s) are the certificate subject name(s), regardless
1511 # of what the client has requested. If the server certificate is
1512 # unavailable, then the name is "none".
1514 # The server name is either SNI (if SNI matches at least one of the
1515 # certificate subject names) or "none" (otherwise). When the server
1516 # certificate is unavailable, the consensus server name is SNI.
1518 # Combining multiple options in one ACL is a fatal configuration
1521 # For all options: If no SNI is available, then the CONNECT request
1522 # target (a.k.a. URI) is used instead of SNI (for an intercepted
1523 # connection, this target is the destination IP address).
1525 acl aclname ssl::server_name_regex [-i] \.foo\.com ...
1526 # regex matches server name obtained from various sources [fast]
1528 acl aclname connections_encrypted
1529 # matches transactions with all HTTP messages received over TLS
1530 # transport connections. [fast]
1532 # The master transaction deals with HTTP messages received from
1533 # various sources. All sources used by the master transaction in the
1534 # past are considered by the ACL. The following rules define whether
1535 # a given message source taints the entire master transaction,
1536 # resulting in ACL mismatches:
1538 # * The HTTP client transport connection is not TLS.
1539 # * An adaptation service connection-encryption flag is off.
1540 # * The peer or origin server transport connection is not TLS.
1542 # Caching currently does not affect these rules. This cache ignorance
1543 # implies that only the current HTTP client transport and REQMOD
1544 # services status determine whether this ACL matches a from-cache
1545 # transaction. The source of the cached response does not have any
1546 # effect on future transaction that use the cached response without
1547 # revalidation. This may change.
1549 # DNS, ICP, and HTCP exchanges during the master transaction do not
1550 # affect these rules.
1552 acl aclname any-of acl1 acl2 ...
1553 # match any one of the acls [fast or slow]
1554 # The first matching ACL stops further ACL evaluation.
1556 # ACLs from multiple any-of lines with the same name are ORed.
1557 # For example, A = (a1 or a2) or (a3 or a4) can be written as
1558 # acl A any-of a1 a2
1559 # acl A any-of a3 a4
1561 # This group ACL is fast if all evaluated ACLs in the group are fast
1562 # and slow otherwise.
1564 acl aclname all-of acl1 acl2 ...
1565 # match all of the acls [fast or slow]
1566 # The first mismatching ACL stops further ACL evaluation.
1568 # ACLs from multiple all-of lines with the same name are ORed.
1569 # For example, B = (b1 and b2) or (b3 and b4) can be written as
1570 # acl B all-of b1 b2
1571 # acl B all-of b3 b4
1573 # This group ACL is fast if all evaluated ACLs in the group are fast
1574 # and slow otherwise.
1577 acl macaddress arp 09:00:2b:23:45:67
1578 acl myexample dst_as 1241
1579 acl password proxy_auth REQUIRED
1580 acl fileupload req_mime_type -i ^multipart/form-data$
1581 acl javascript rep_mime_type -i ^application/x-javascript$
1585 # Recommended minimum configuration:
1588 # Example rule allowing access from your local networks.
1589 # Adapt to list your (internal) IP networks from where browsing
1591 acl localnet src 0.0.0.1-0.255.255.255 # RFC 1122 "this" network (LAN)
1592 acl localnet src 10.0.0.0/8 # RFC 1918 local private network (LAN)
1593 acl localnet src 100.64.0.0/10 # RFC 6598 shared address space (CGN)
1594 acl localnet src 169.254.0.0/16 # RFC 3927 link-local (directly plugged) machines
1595 acl localnet src 172.16.0.0/12 # RFC 1918 local private network (LAN)
1596 acl localnet src 192.168.0.0/16 # RFC 1918 local private network (LAN)
1597 acl localnet src fc00::/7 # RFC 4193 local private network range
1598 acl localnet src fe80::/10 # RFC 4291 link-local (directly plugged) machines
1600 acl SSL_ports port 443
1601 acl Safe_ports port 80 # http
1602 acl Safe_ports port 21 # ftp
1603 acl Safe_ports port 443 # https
1604 acl Safe_ports port 70 # gopher
1605 acl Safe_ports port 210 # wais
1606 acl Safe_ports port 1025-65535 # unregistered ports
1607 acl Safe_ports port 280 # http-mgmt
1608 acl Safe_ports port 488 # gss-http
1609 acl Safe_ports port 591 # filemaker
1610 acl Safe_ports port 777 # multiling http
1614 NAME: proxy_protocol_access
1616 LOC: Config.accessList.proxyProtocol
1618 DEFAULT_DOC: all TCP connections to ports with require-proxy-header will be denied
1620 Determine which client proxies can be trusted to provide correct
1621 information regarding real client IP address using PROXY protocol.
1623 Requests may pass through a chain of several other proxies
1624 before reaching us. The original source details may by sent in:
1625 * HTTP message Forwarded header, or
1626 * HTTP message X-Forwarded-For header, or
1627 * PROXY protocol connection header.
1629 This directive is solely for validating new PROXY protocol
1630 connections received from a port flagged with require-proxy-header.
1631 It is checked only once after TCP connection setup.
1633 A deny match results in TCP connection closure.
1635 An allow match is required for Squid to permit the corresponding
1636 TCP connection, before Squid even looks for HTTP request headers.
1637 If there is an allow match, Squid starts using PROXY header information
1638 to determine the source address of the connection for all future ACL
1639 checks, logging, etc.
1641 SECURITY CONSIDERATIONS:
1643 Any host from which we accept client IP details can place
1644 incorrect information in the relevant header, and Squid
1645 will use the incorrect information as if it were the
1646 source address of the request. This may enable remote
1647 hosts to bypass any access control restrictions that are
1648 based on the client's source addresses.
1650 This clause only supports fast acl types.
1651 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1654 NAME: follow_x_forwarded_for
1656 IFDEF: FOLLOW_X_FORWARDED_FOR
1657 LOC: Config.accessList.followXFF
1658 DEFAULT_IF_NONE: deny all
1659 DEFAULT_DOC: X-Forwarded-For header will be ignored.
1661 Determine which client proxies can be trusted to provide correct
1662 information regarding real client IP address.
1664 Requests may pass through a chain of several other proxies
1665 before reaching us. The original source details may by sent in:
1666 * HTTP message Forwarded header, or
1667 * HTTP message X-Forwarded-For header, or
1668 * PROXY protocol connection header.
1670 PROXY protocol connections are controlled by the proxy_protocol_access
1671 directive which is checked before this.
1673 If a request reaches us from a source that is allowed by this
1674 directive, then we trust the information it provides regarding
1675 the IP of the client it received from (if any).
1677 For the purpose of ACLs used in this directive the src ACL type always
1678 matches the address we are testing and srcdomain matches its rDNS.
1680 On each HTTP request Squid checks for X-Forwarded-For header fields.
1681 If found the header values are iterated in reverse order and an allow
1682 match is required for Squid to continue on to the next value.
1683 The verification ends when a value receives a deny match, cannot be
1684 tested, or there are no more values to test.
1685 NOTE: Squid does not yet follow the Forwarded HTTP header.
1687 The end result of this process is an IP address that we will
1688 refer to as the indirect client address. This address may
1689 be treated as the client address for access control, ICAP, delay
1690 pools and logging, depending on the acl_uses_indirect_client,
1691 icap_uses_indirect_client, delay_pool_uses_indirect_client,
1692 log_uses_indirect_client and tproxy_uses_indirect_client options.
1694 This clause only supports fast acl types.
1695 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1697 SECURITY CONSIDERATIONS:
1699 Any host from which we accept client IP details can place
1700 incorrect information in the relevant header, and Squid
1701 will use the incorrect information as if it were the
1702 source address of the request. This may enable remote
1703 hosts to bypass any access control restrictions that are
1704 based on the client's source addresses.
1708 acl localhost src 127.0.0.1
1709 acl my_other_proxy srcdomain .proxy.example.com
1710 follow_x_forwarded_for allow localhost
1711 follow_x_forwarded_for allow my_other_proxy
1714 NAME: acl_uses_indirect_client
1717 IFDEF: FOLLOW_X_FORWARDED_FOR
1719 LOC: Config.onoff.acl_uses_indirect_client
1721 Controls whether the indirect client address
1722 (see follow_x_forwarded_for) is used instead of the
1723 direct client address in acl matching.
1725 NOTE: maxconn ACL considers direct TCP links and indirect
1726 clients will always have zero. So no match.
1729 NAME: delay_pool_uses_indirect_client
1732 IFDEF: FOLLOW_X_FORWARDED_FOR&&USE_DELAY_POOLS
1734 LOC: Config.onoff.delay_pool_uses_indirect_client
1736 Controls whether the indirect client address
1737 (see follow_x_forwarded_for) is used instead of the
1738 direct client address in delay pools.
1741 NAME: log_uses_indirect_client
1744 IFDEF: FOLLOW_X_FORWARDED_FOR
1746 LOC: Config.onoff.log_uses_indirect_client
1748 Controls whether the indirect client address
1749 (see follow_x_forwarded_for) is used instead of the
1750 direct client address in the access log.
1753 NAME: tproxy_uses_indirect_client
1756 IFDEF: FOLLOW_X_FORWARDED_FOR&&LINUX_NETFILTER
1758 LOC: Config.onoff.tproxy_uses_indirect_client
1760 Controls whether the indirect client address
1761 (see follow_x_forwarded_for) is used instead of the
1762 direct client address when spoofing the outgoing client.
1764 This has no effect on requests arriving in non-tproxy
1767 SECURITY WARNING: Usage of this option is dangerous
1768 and should not be used trivially. Correct configuration
1769 of follow_x_forwarded_for with a limited set of trusted
1770 sources is required to prevent abuse of your proxy.
1773 NAME: spoof_client_ip
1775 LOC: Config.accessList.spoof_client_ip
1777 DEFAULT_DOC: Allow spoofing on all TPROXY traffic.
1779 Control client IP address spoofing of TPROXY traffic based on
1780 defined access lists.
1782 spoof_client_ip allow|deny [!]aclname ...
1784 If there are no "spoof_client_ip" lines present, the default
1785 is to "allow" spoofing of any suitable request.
1787 Note that the cache_peer "no-tproxy" option overrides this ACL.
1789 This clause supports fast acl types.
1790 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1795 LOC: Config.accessList.http
1796 DEFAULT_IF_NONE: deny all
1797 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
1799 Allowing or Denying access based on defined access lists
1801 To allow or deny a message received on an HTTP, HTTPS, or FTP port:
1802 http_access allow|deny [!]aclname ...
1804 NOTE on default values:
1806 If there are no "access" lines present, the default is to deny
1809 If none of the "access" lines cause a match, the default is the
1810 opposite of the last line in the list. If the last line was
1811 deny, the default is allow. Conversely, if the last line
1812 is allow, the default will be deny. For these reasons, it is a
1813 good idea to have an "deny all" entry at the end of your access
1814 lists to avoid potential confusion.
1816 This clause supports both fast and slow acl types.
1817 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1822 # Recommended minimum Access Permission configuration:
1824 # Deny requests to certain unsafe ports
1825 http_access deny !Safe_ports
1827 # Deny CONNECT to other than secure SSL ports
1828 http_access deny CONNECT !SSL_ports
1830 # Only allow cachemgr access from localhost
1831 http_access allow localhost manager
1832 http_access deny manager
1834 # We strongly recommend the following be uncommented to protect innocent
1835 # web applications running on the proxy server who think the only
1836 # one who can access services on "localhost" is a local user
1837 #http_access deny to_localhost
1840 # INSERT YOUR OWN RULE(S) HERE TO ALLOW ACCESS FROM YOUR CLIENTS
1843 # Example rule allowing access from your local networks.
1844 # Adapt localnet in the ACL section to list your (internal) IP networks
1845 # from where browsing should be allowed
1846 http_access allow localnet
1847 http_access allow localhost
1849 # And finally deny all other access to this proxy
1850 http_access deny all
1854 NAME: adapted_http_access http_access2
1856 LOC: Config.accessList.adapted_http
1858 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
1860 Allowing or Denying access based on defined access lists
1862 Essentially identical to http_access, but runs after redirectors
1863 and ICAP/eCAP adaptation. Allowing access control based on their
1866 If not set then only http_access is used.
1869 NAME: http_reply_access
1871 LOC: Config.accessList.reply
1873 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
1875 Allow replies to client requests. This is complementary to http_access.
1877 http_reply_access allow|deny [!] aclname ...
1879 NOTE: if there are no access lines present, the default is to allow
1882 If none of the access lines cause a match the opposite of the
1883 last line will apply. Thus it is good practice to end the rules
1884 with an "allow all" or "deny all" entry.
1886 This clause supports both fast and slow acl types.
1887 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1892 LOC: Config.accessList.icp
1894 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
1896 Allowing or Denying access to the ICP port based on defined
1899 icp_access allow|deny [!]aclname ...
1901 NOTE: The default if no icp_access lines are present is to
1902 deny all traffic. This default may cause problems with peers
1905 This clause only supports fast acl types.
1906 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1908 # Allow ICP queries from local networks only
1909 #icp_access allow localnet
1910 #icp_access deny all
1916 LOC: Config.accessList.htcp
1918 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
1920 Allowing or Denying access to the HTCP port based on defined
1923 htcp_access allow|deny [!]aclname ...
1925 See also htcp_clr_access for details on access control for
1926 cache purge (CLR) HTCP messages.
1928 NOTE: The default if no htcp_access lines are present is to
1929 deny all traffic. This default may cause problems with peers
1930 using the htcp option.
1932 This clause only supports fast acl types.
1933 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1935 # Allow HTCP queries from local networks only
1936 #htcp_access allow localnet
1937 #htcp_access deny all
1940 NAME: htcp_clr_access
1943 LOC: Config.accessList.htcp_clr
1945 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
1947 Allowing or Denying access to purge content using HTCP based
1948 on defined access lists.
1949 See htcp_access for details on general HTCP access control.
1951 htcp_clr_access allow|deny [!]aclname ...
1953 This clause only supports fast acl types.
1954 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1956 # Allow HTCP CLR requests from trusted peers
1957 acl htcp_clr_peer src 192.0.2.2 2001:DB8::2
1958 htcp_clr_access allow htcp_clr_peer
1959 htcp_clr_access deny all
1964 LOC: Config.accessList.miss
1966 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
1968 Determines whether network access is permitted when satisfying a request.
1971 to force your neighbors to use you as a sibling instead of
1974 acl localclients src 192.0.2.0/24 2001:DB8::a:0/64
1975 miss_access deny !localclients
1976 miss_access allow all
1978 This means only your local clients are allowed to fetch relayed/MISS
1979 replies from the network and all other clients can only fetch cached
1982 The default for this setting allows all clients who passed the
1983 http_access rules to relay via this proxy.
1985 This clause only supports fast acl types.
1986 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1989 NAME: ident_lookup_access
1993 DEFAULT_DOC: Unless rules exist in squid.conf, IDENT is not fetched.
1994 LOC: Ident::TheConfig.identLookup
1996 A list of ACL elements which, if matched, cause an ident
1997 (RFC 931) lookup to be performed for this request. For
1998 example, you might choose to always perform ident lookups
1999 for your main multi-user Unix boxes, but not for your Macs
2000 and PCs. By default, ident lookups are not performed for
2003 To enable ident lookups for specific client addresses, you
2004 can follow this example:
2006 acl ident_aware_hosts src 198.168.1.0/24
2007 ident_lookup_access allow ident_aware_hosts
2008 ident_lookup_access deny all
2010 Only src type ACL checks are fully supported. A srcdomain
2011 ACL might work at times, but it will not always provide
2014 This clause only supports fast acl types.
2015 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2018 NAME: reply_body_max_size
2019 COMMENT: size [acl acl...]
2022 DEFAULT_DOC: No limit is applied.
2023 LOC: Config.ReplyBodySize
2025 This option specifies the maximum size of a reply body. It can be
2026 used to prevent users from downloading very large files, such as
2027 MP3's and movies. When the reply headers are received, the
2028 reply_body_max_size lines are processed, and the first line where
2029 all (if any) listed ACLs are true is used as the maximum body size
2032 This size is checked twice. First when we get the reply headers,
2033 we check the content-length value. If the content length value exists
2034 and is larger than the allowed size, the request is denied and the
2035 user receives an error message that says "the request or reply
2036 is too large." If there is no content-length, and the reply
2037 size exceeds this limit, the client's connection is just closed
2038 and they will receive a partial reply.
2040 WARNING: downstream caches probably can not detect a partial reply
2041 if there is no content-length header, so they will cache
2042 partial responses and give them out as hits. You should NOT
2043 use this option if you have downstream caches.
2045 WARNING: A maximum size smaller than the size of squid's error messages
2046 will cause an infinite loop and crash squid. Ensure that the smallest
2047 non-zero value you use is greater that the maximum header size plus
2048 the size of your largest error page.
2050 If you set this parameter none (the default), there will be
2053 Configuration Format is:
2054 reply_body_max_size SIZE UNITS [acl ...]
2056 reply_body_max_size 10 MB
2060 NAME: on_unsupported_protocol
2061 TYPE: on_unsupported_protocol
2062 LOC: Config.accessList.on_unsupported_protocol
2064 DEFAULT_DOC: Respond with an error message to unidentifiable traffic
2066 Determines Squid behavior when encountering strange requests at the
2067 beginning of an accepted TCP connection or the beginning of a bumped
2068 CONNECT tunnel. Controlling Squid reaction to unexpected traffic is
2069 especially useful in interception environments where Squid is likely
2070 to see connections for unsupported protocols that Squid should either
2071 terminate or tunnel at TCP level.
2073 on_unsupported_protocol <action> [!]acl ...
2075 The first matching action wins. Only fast ACLs are supported.
2077 Supported actions are:
2079 tunnel: Establish a TCP connection with the intended server and
2080 blindly shovel TCP packets between the client and server.
2082 respond: Respond with an error message, using the transfer protocol
2083 for the Squid port that received the request (e.g., HTTP
2084 for connections intercepted at the http_port). This is the
2087 Squid expects the following traffic patterns:
2089 http_port: a plain HTTP request
2090 https_port: SSL/TLS handshake followed by an [encrypted] HTTP request
2091 ftp_port: a plain FTP command (no on_unsupported_protocol support yet!)
2092 CONNECT tunnel on http_port: same as https_port
2093 CONNECT tunnel on https_port: same as https_port
2095 Currently, this directive has effect on intercepted connections and
2096 bumped tunnels only. Other cases are not supported because Squid
2097 cannot know the intended destination of other traffic.
2100 # define what Squid errors indicate receiving non-HTTP traffic:
2101 acl foreignProtocol squid_error ERR_PROTOCOL_UNKNOWN ERR_TOO_BIG
2102 # define what Squid errors indicate receiving nothing:
2103 acl serverTalksFirstProtocol squid_error ERR_REQUEST_START_TIMEOUT
2104 # tunnel everything that does not look like HTTP:
2105 on_unsupported_protocol tunnel foreignProtocol
2106 # tunnel if we think the client waits for the server to talk first:
2107 on_unsupported_protocol tunnel serverTalksFirstProtocol
2108 # in all other error cases, just send an HTTP "error page" response:
2109 on_unsupported_protocol respond all
2111 See also: squid_error ACL
2117 LOC: Auth::TheConfig.schemeAccess
2119 DEFAULT_DOC: use all auth_param schemes in their configuration order
2121 Use this directive to customize authentication schemes presence and
2122 order in Squid's Unauthorized and Authentication Required responses.
2124 auth_schemes scheme1,scheme2,... [!]aclname ...
2126 where schemeN is the name of one of the authentication schemes
2127 configured using auth_param directives. At least one scheme name is
2128 required. Multiple scheme names are separated by commas. Either
2129 avoid whitespace or quote the entire schemes list.
2131 A special "ALL" scheme name expands to all auth_param-configured
2132 schemes in their configuration order. This directive cannot be used
2133 to configure Squid to offer no authentication schemes at all.
2135 The first matching auth_schemes rule determines the schemes order
2136 for the current Authentication Required transaction. Note that the
2137 future response is not yet available during auth_schemes evaluation.
2139 If this directive is not used or none of its rules match, then Squid
2140 responds with all configured authentication schemes in the order of
2141 auth_param directives in the configuration file.
2143 This directive does not determine when authentication is used or
2144 how each authentication scheme authenticates clients.
2146 The following example sends basic and negotiate authentication
2147 schemes, in that order, when requesting authentication of HTTP
2148 requests matching the isIE ACL (not shown) while sending all
2149 auth_param schemes in their configuration order to other clients:
2151 auth_schemes basic,negotiate isIE
2152 auth_schemes ALL all # explicit default
2154 This directive supports fast ACLs only.
2156 See also: auth_param.
2161 -----------------------------------------------------------------------------
2164 NAME: http_port ascii_port
2169 Usage: port [mode] [options]
2170 hostname:port [mode] [options]
2171 1.2.3.4:port [mode] [options]
2173 The socket addresses where Squid will listen for HTTP client
2174 requests. You may specify multiple socket addresses.
2175 There are three forms: port alone, hostname with port, and
2176 IP address with port. If you specify a hostname or IP
2177 address, Squid binds the socket to that specific
2178 address. Most likely, you do not need to bind to a specific
2179 address, so you can use the port number alone.
2181 If you are running Squid in accelerator mode, you
2182 probably want to listen on port 80 also, or instead.
2184 The -a command line option may be used to specify additional
2185 port(s) where Squid listens for proxy request. Such ports will
2186 be plain proxy ports with no options.
2188 You may specify multiple socket addresses on multiple lines.
2192 intercept Support for IP-Layer NAT interception delivering
2193 traffic to this Squid port.
2194 NP: disables authentication on the port.
2196 tproxy Support Linux TPROXY (or BSD divert-to) with spoofing
2197 of outgoing connections using the client IP address.
2198 NP: disables authentication on the port.
2200 accel Accelerator / reverse proxy mode
2202 ssl-bump For each CONNECT request allowed by ssl_bump ACLs,
2203 establish secure connection with the client and with
2204 the server, decrypt HTTPS messages as they pass through
2205 Squid, and treat them as unencrypted HTTP messages,
2206 becoming the man-in-the-middle.
2208 The ssl_bump option is required to fully enable
2209 bumping of CONNECT requests.
2211 Omitting the mode flag causes default forward proxy mode to be used.
2214 Accelerator Mode Options:
2216 defaultsite=domainname
2217 What to use for the Host: header if it is not present
2218 in a request. Determines what site (not origin server)
2219 accelerators should consider the default.
2221 no-vhost Disable using HTTP/1.1 Host header for virtual domain support.
2223 protocol= Protocol to reconstruct accelerated and intercepted
2224 requests with. Defaults to HTTP/1.1 for http_port and
2225 HTTPS/1.1 for https_port.
2226 When an unsupported value is configured Squid will
2227 produce a FATAL error.
2228 Values: HTTP or HTTP/1.1, HTTPS or HTTPS/1.1
2230 vport Virtual host port support. Using the http_port number
2231 instead of the port passed on Host: headers.
2233 vport=NN Virtual host port support. Using the specified port
2234 number instead of the port passed on Host: headers.
2237 Act as if this Squid is the origin server.
2238 This currently means generate new Date: and Expires:
2239 headers on HIT instead of adding Age:.
2241 ignore-cc Ignore request Cache-Control headers.
2243 WARNING: This option violates HTTP specifications if
2244 used in non-accelerator setups.
2246 allow-direct Allow direct forwarding in accelerator mode. Normally
2247 accelerated requests are denied direct forwarding as if
2248 never_direct was used.
2250 WARNING: this option opens accelerator mode to security
2251 vulnerabilities usually only affecting in interception
2252 mode. Make sure to protect forwarding with suitable
2253 http_access rules when using this.
2256 SSL Bump Mode Options:
2257 In addition to these options ssl-bump requires TLS/SSL options.
2259 generate-host-certificates[=<on|off>]
2260 Dynamically create SSL server certificates for the
2261 destination hosts of bumped CONNECT requests.When
2262 enabled, the cert and key options are used to sign
2263 generated certificates. Otherwise generated
2264 certificate will be selfsigned.
2265 If there is a CA certificate lifetime of the generated
2266 certificate equals lifetime of the CA certificate. If
2267 generated certificate is selfsigned lifetime is three
2269 This option is enabled by default when ssl-bump is used.
2270 See the ssl-bump option above for more information.
2272 dynamic_cert_mem_cache_size=SIZE
2273 Approximate total RAM size spent on cached generated
2274 certificates. If set to zero, caching is disabled. The
2275 default value is 4MB.
2279 tls-cert= Path to file containing an X.509 certificate (PEM format)
2280 to be used in the TLS handshake ServerHello.
2282 If this certificate is constrained by KeyUsage TLS
2283 feature it must allow HTTP server usage, along with
2284 any additional restrictions imposed by your choice
2285 of options= settings.
2287 When OpenSSL is used this file may also contain a
2288 chain of intermediate CA certificates to send in the
2291 When GnuTLS is used this option (and any paired
2292 tls-key= option) may be repeated to load multiple
2293 certificates for different domains.
2295 Also, when generate-host-certificates=on is configured
2296 the first tls-cert= option must be a CA certificate
2297 capable of signing the automatically generated
2300 tls-key= Path to a file containing private key file (PEM format)
2301 for the previous tls-cert= option.
2303 If tls-key= is not specified tls-cert= is assumed to
2304 reference a PEM file containing both the certificate
2307 cipher= Colon separated list of supported ciphers.
2308 NOTE: some ciphers such as EDH ciphers depend on
2309 additional settings. If those settings are
2310 omitted the ciphers may be silently ignored
2311 by the OpenSSL library.
2313 options= Various SSL implementation options. The most important
2316 NO_SSLv3 Disallow the use of SSLv3
2318 NO_TLSv1 Disallow the use of TLSv1.0
2320 NO_TLSv1_1 Disallow the use of TLSv1.1
2322 NO_TLSv1_2 Disallow the use of TLSv1.2
2325 Always create a new key when using
2326 temporary/ephemeral DH key exchanges
2329 Enable ephemeral ECDH key exchange.
2330 The adopted curve should be specified
2331 using the tls-dh option.
2334 Disable use of RFC5077 session tickets.
2335 Some servers may have problems
2336 understanding the TLS extension due
2337 to ambiguous specification in RFC4507.
2339 ALL Enable various bug workarounds
2340 suggested as "harmless" by OpenSSL
2341 Be warned that this reduces SSL/TLS
2342 strength to some attacks.
2344 See the OpenSSL SSL_CTX_set_options documentation for a
2347 clientca= File containing the list of CAs to use when
2348 requesting a client certificate.
2350 tls-cafile= PEM file containing CA certificates to use when verifying
2351 client certificates. If not configured clientca will be
2352 used. May be repeated to load multiple files.
2354 capath= Directory containing additional CA certificates
2355 and CRL lists to use when verifying client certificates.
2356 Requires OpenSSL or LibreSSL.
2358 crlfile= File of additional CRL lists to use when verifying
2359 the client certificate, in addition to CRLs stored in
2360 the capath. Implies VERIFY_CRL flag below.
2363 File containing DH parameters for temporary/ephemeral DH key
2364 exchanges, optionally prefixed by a curve for ephemeral ECDH
2366 See OpenSSL documentation for details on how to create the
2367 DH parameter file. Supported curves for ECDH can be listed
2368 using the "openssl ecparam -list_curves" command.
2369 WARNING: EDH and EECDH ciphers will be silently disabled if
2370 this option is not set.
2372 sslflags= Various flags modifying the use of SSL:
2374 Don't request client certificates
2375 immediately, but wait until acl processing
2376 requires a certificate (not yet implemented).
2378 Don't allow for session reuse. Each connection
2379 will result in a new SSL session.
2381 Verify CRL lists when accepting client
2384 Verify CRL lists for all certificates in the
2385 client certificate chain.
2387 tls-default-ca[=off]
2388 Whether to use the system Trusted CAs. Default is OFF.
2390 tls-no-npn Do not use the TLS NPN extension to advertise HTTP/1.1.
2392 sslcontext= SSL session ID context identifier.
2396 connection-auth[=on|off]
2397 use connection-auth=off to tell Squid to prevent
2398 forwarding Microsoft connection oriented authentication
2399 (NTLM, Negotiate and Kerberos)
2401 disable-pmtu-discovery=
2402 Control Path-MTU discovery usage:
2403 off lets OS decide on what to do (default).
2404 transparent disable PMTU discovery when transparent
2406 always disable always PMTU discovery.
2408 In many setups of transparently intercepting proxies
2409 Path-MTU discovery can not work on traffic towards the
2410 clients. This is the case when the intercepting device
2411 does not fully track connections and fails to forward
2412 ICMP must fragment messages to the cache server. If you
2413 have such setup and experience that certain clients
2414 sporadically hang or never complete requests set
2415 disable-pmtu-discovery option to 'transparent'.
2417 name= Specifies a internal name for the port. Defaults to
2418 the port specification (port or addr:port)
2420 tcpkeepalive[=idle,interval,timeout]
2421 Enable TCP keepalive probes of idle connections.
2422 In seconds; idle is the initial time before TCP starts
2423 probing the connection, interval how often to probe, and
2424 timeout the time before giving up.
2426 require-proxy-header
2427 Require PROXY protocol version 1 or 2 connections.
2428 The proxy_protocol_access is required to whitelist
2429 downstream proxies which can be trusted.
2431 If you run Squid on a dual-homed machine with an internal
2432 and an external interface we recommend you to specify the
2433 internal address:port in http_port. This way Squid will only be
2434 visible on the internal address.
2438 # Squid normally listens to port 3128
2439 http_port @DEFAULT_HTTP_PORT@
2444 IFDEF: USE_GNUTLS||USE_OPENSSL
2449 Usage: [ip:]port [mode] tls-cert=certificate.pem [options]
2451 The socket address where Squid will listen for client requests made
2452 over TLS or SSL connections. Commonly referred to as HTTPS.
2454 This is most useful for situations where you are running squid in
2455 accelerator mode and you want to do the TLS work at the accelerator
2458 You may specify multiple socket addresses on multiple lines,
2459 each with their own certificate and/or options.
2461 The tls-cert= option is mandatory on HTTPS ports.
2463 See http_port for a list of modes and options.
2471 Enables Native FTP proxy by specifying the socket address where Squid
2472 listens for FTP client requests. See http_port directive for various
2473 ways to specify the listening address and mode.
2475 Usage: ftp_port address [mode] [options]
2477 WARNING: This is a new, experimental, complex feature that has seen
2478 limited production exposure. Some Squid modules (e.g., caching) do not
2479 currently work with native FTP proxying, and many features have not
2480 even been tested for compatibility. Test well before deploying!
2482 Native FTP proxying differs substantially from proxying HTTP requests
2483 with ftp:// URIs because Squid works as an FTP server and receives
2484 actual FTP commands (rather than HTTP requests with FTP URLs).
2486 Native FTP commands accepted at ftp_port are internally converted or
2487 wrapped into HTTP-like messages. The same happens to Native FTP
2488 responses received from FTP origin servers. Those HTTP-like messages
2489 are shoveled through regular access control and adaptation layers
2490 between the FTP client and the FTP origin server. This allows Squid to
2491 examine, adapt, block, and log FTP exchanges. Squid reuses most HTTP
2492 mechanisms when shoveling wrapped FTP messages. For example,
2493 http_access and adaptation_access directives are used.
2497 intercept Same as http_port intercept. The FTP origin address is
2498 determined based on the intended destination of the
2499 intercepted connection.
2501 tproxy Support Linux TPROXY for spoofing outgoing
2502 connections using the client IP address.
2503 NP: disables authentication and maybe IPv6 on the port.
2505 By default (i.e., without an explicit mode option), Squid extracts the
2506 FTP origin address from the login@origin parameter of the FTP USER
2507 command. Many popular FTP clients support such native FTP proxying.
2511 name=token Specifies an internal name for the port. Defaults to
2512 the port address. Usable with myportname ACL.
2515 Enables tracking of FTP directories by injecting extra
2516 PWD commands and adjusting Request-URI (in wrapping
2517 HTTP requests) to reflect the current FTP server
2518 directory. Tracking is disabled by default.
2520 protocol=FTP Protocol to reconstruct accelerated and intercepted
2521 requests with. Defaults to FTP. No other accepted
2522 values have been tested with. An unsupported value
2523 results in a FATAL error. Accepted values are FTP,
2524 HTTP (or HTTP/1.1), and HTTPS (or HTTPS/1.1).
2526 Other http_port modes and options that are not specific to HTTP and
2527 HTTPS may also work.
2530 NAME: tcp_outgoing_tos tcp_outgoing_ds tcp_outgoing_dscp
2533 LOC: Ip::Qos::TheConfig.tosToServer
2535 Allows you to select a TOS/Diffserv value for packets outgoing
2536 on the server side, based on an ACL.
2538 tcp_outgoing_tos ds-field [!]aclname ...
2540 Example where normal_service_net uses the TOS value 0x00
2541 and good_service_net uses 0x20
2543 acl normal_service_net src 10.0.0.0/24
2544 acl good_service_net src 10.0.1.0/24
2545 tcp_outgoing_tos 0x00 normal_service_net
2546 tcp_outgoing_tos 0x20 good_service_net
2548 TOS/DSCP values really only have local significance - so you should
2549 know what you're specifying. For more information, see RFC2474,
2550 RFC2475, and RFC3260.
2552 The TOS/DSCP byte must be exactly that - a octet value 0 - 255, or
2553 "default" to use whatever default your host has.
2554 Note that only multiples of 4 are usable as the two rightmost bits have
2555 been redefined for use by ECN (RFC 3168 section 23.1).
2556 The squid parser will enforce this by masking away the ECN bits.
2558 Processing proceeds in the order specified, and stops at first fully
2561 Only fast ACLs are supported.
2564 NAME: clientside_tos
2567 LOC: Ip::Qos::TheConfig.tosToClient
2569 Allows you to select a TOS/DSCP value for packets being transmitted
2570 on the client-side, based on an ACL.
2572 clientside_tos ds-field [!]aclname ...
2574 Example where normal_service_net uses the TOS value 0x00
2575 and good_service_net uses 0x20
2577 acl normal_service_net src 10.0.0.0/24
2578 acl good_service_net src 10.0.1.0/24
2579 clientside_tos 0x00 normal_service_net
2580 clientside_tos 0x20 good_service_net
2582 Note: This feature is incompatible with qos_flows. Any TOS values set here
2583 will be overwritten by TOS values in qos_flows.
2585 The TOS/DSCP byte must be exactly that - a octet value 0 - 255, or
2586 "default" to use whatever default your host has.
2587 Note that only multiples of 4 are usable as the two rightmost bits have
2588 been redefined for use by ECN (RFC 3168 section 23.1).
2589 The squid parser will enforce this by masking away the ECN bits.
2591 This clause only supports fast acl types.
2592 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2595 NAME: tcp_outgoing_mark
2597 IFDEF: SO_MARK&&USE_LIBCAP
2599 LOC: Ip::Qos::TheConfig.nfmarkToServer
2601 Allows you to apply a Netfilter mark value to outgoing packets
2602 on the server side, based on an ACL.
2604 tcp_outgoing_mark mark-value [!]aclname ...
2606 Example where normal_service_net uses the mark value 0x00
2607 and good_service_net uses 0x20
2609 acl normal_service_net src 10.0.0.0/24
2610 acl good_service_net src 10.0.1.0/24
2611 tcp_outgoing_mark 0x00 normal_service_net
2612 tcp_outgoing_mark 0x20 good_service_net
2614 Only fast ACLs are supported.
2617 NAME: mark_client_packet clientside_mark
2619 IFDEF: SO_MARK&&USE_LIBCAP
2621 LOC: Ip::Qos::TheConfig.nfmarkToClient
2623 Allows you to apply a Netfilter MARK value to packets being transmitted
2624 on the client-side, based on an ACL.
2626 mark_client_packet mark-value [!]aclname ...
2628 Example where normal_service_net uses the MARK value 0x00
2629 and good_service_net uses 0x20
2631 acl normal_service_net src 10.0.0.0/24
2632 acl good_service_net src 10.0.1.0/24
2633 mark_client_packet 0x00 normal_service_net
2634 mark_client_packet 0x20 good_service_net
2636 Note: This feature is incompatible with qos_flows. Any mark values set here
2637 will be overwritten by mark values in qos_flows.
2639 This clause only supports fast acl types.
2640 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2643 NAME: mark_client_connection
2645 IFDEF: SO_MARK&&USE_LIBCAP
2647 LOC: Ip::Qos::TheConfig.nfConnmarkToClient
2649 Allows you to apply a Netfilter CONNMARK value to a connection
2650 on the client-side, based on an ACL.
2652 mark_client_connection mark-value[/mask] [!]aclname ...
2654 The mark-value and mask are unsigned integers (hex, octal, or decimal).
2655 The mask may be used to preserve marking previously set by other agents
2658 A matching rule replaces the CONNMARK value. If a mask is also
2659 specified, then the masked bits of the original value are zeroed, and
2660 the configured mark-value is ORed with that adjusted value.
2661 For example, applying a mark-value 0xAB/0xF to 0x5F CONNMARK, results
2662 in a 0xFB marking (rather than a 0xAB or 0x5B).
2664 This directive semantics is similar to iptables --set-mark rather than
2665 --set-xmark functionality.
2667 The directive does not interfere with qos_flows (which uses packet MARKs,
2670 Example where squid marks intercepted FTP connections:
2672 acl proto_ftp proto FTP
2673 mark_client_connection 0x200/0xff00 proto_ftp
2675 This clause only supports fast acl types.
2676 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2683 LOC: Ip::Qos::TheConfig
2685 Allows you to select a TOS/DSCP value to mark outgoing
2686 connections to the client, based on where the reply was sourced.
2687 For platforms using netfilter, allows you to set a netfilter mark
2688 value instead of, or in addition to, a TOS value.
2690 By default this functionality is disabled. To enable it with the default
2691 settings simply use "qos_flows mark" or "qos_flows tos". Default
2692 settings will result in the netfilter mark or TOS value being copied
2693 from the upstream connection to the client. Note that it is the connection
2694 CONNMARK value not the packet MARK value that is copied.
2696 It is not currently possible to copy the mark or TOS value from the
2697 client to the upstream connection request.
2699 TOS values really only have local significance - so you should
2700 know what you're specifying. For more information, see RFC2474,
2701 RFC2475, and RFC3260.
2703 The TOS/DSCP byte must be exactly that - a octet value 0 - 255.
2704 Note that only multiples of 4 are usable as the two rightmost bits have
2705 been redefined for use by ECN (RFC 3168 section 23.1).
2706 The squid parser will enforce this by masking away the ECN bits.
2708 Mark values can be any unsigned 32-bit integer value.
2710 This setting is configured by setting the following values:
2712 tos|mark Whether to set TOS or netfilter mark values
2714 local-hit=0xFF Value to mark local cache hits.
2716 sibling-hit=0xFF Value to mark hits from sibling peers.
2718 parent-hit=0xFF Value to mark hits from parent peers.
2720 miss=0xFF[/mask] Value to mark cache misses. Takes precedence
2721 over the preserve-miss feature (see below), unless
2722 mask is specified, in which case only the bits
2723 specified in the mask are written.
2725 The TOS variant of the following features are only possible on Linux
2726 and require your kernel to be patched with the TOS preserving ZPH
2727 patch, available from http://zph.bratcheda.org
2728 No patch is needed to preserve the netfilter mark, which will work
2729 with all variants of netfilter.
2731 disable-preserve-miss
2732 This option disables the preservation of the TOS or netfilter
2733 mark. By default, the existing TOS or netfilter mark value of
2734 the response coming from the remote server will be retained
2735 and masked with miss-mark.
2736 NOTE: in the case of a netfilter mark, the mark must be set on
2737 the connection (using the CONNMARK target) not on the packet
2741 Allows you to mask certain bits in the TOS or mark value
2742 received from the remote server, before copying the value to
2743 the TOS sent towards clients.
2744 Default for tos: 0xFF (TOS from server is not changed).
2745 Default for mark: 0xFFFFFFFF (mark from server is not changed).
2747 All of these features require the --enable-zph-qos compilation flag
2748 (enabled by default). Netfilter marking also requires the
2749 libnetfilter_conntrack libraries (--with-netfilter-conntrack) and
2750 libcap 2.09+ (--with-libcap).
2754 NAME: tcp_outgoing_address
2757 DEFAULT_DOC: Address selection is performed by the operating system.
2758 LOC: Config.accessList.outgoing_address
2760 Allows you to map requests to different outgoing IP addresses
2761 based on the username or source address of the user making
2764 tcp_outgoing_address ipaddr [[!]aclname] ...
2767 Forwarding clients with dedicated IPs for certain subnets.
2769 acl normal_service_net src 10.0.0.0/24
2770 acl good_service_net src 10.0.2.0/24
2772 tcp_outgoing_address 2001:db8::c001 good_service_net
2773 tcp_outgoing_address 10.1.0.2 good_service_net
2775 tcp_outgoing_address 2001:db8::beef normal_service_net
2776 tcp_outgoing_address 10.1.0.1 normal_service_net
2778 tcp_outgoing_address 2001:db8::1
2779 tcp_outgoing_address 10.1.0.3
2781 Processing proceeds in the order specified, and stops at first fully
2784 Squid will add an implicit IP version test to each line.
2785 Requests going to IPv4 websites will use the outgoing 10.1.0.* addresses.
2786 Requests going to IPv6 websites will use the outgoing 2001:db8:* addresses.
2789 NOTE: The use of this directive using client dependent ACLs is
2790 incompatible with the use of server side persistent connections. To
2791 ensure correct results it is best to set server_persistent_connections
2792 to off when using this directive in such configurations.
2794 NOTE: The use of this directive to set a local IP on outgoing TCP links
2795 is incompatible with using TPROXY to set client IP out outbound TCP links.
2796 When needing to contact peers use the no-tproxy cache_peer option and the
2797 client_dst_passthru directive re-enable normal forwarding such as this.
2799 This clause only supports fast acl types.
2800 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2803 NAME: host_verify_strict
2806 LOC: Config.onoff.hostStrictVerify
2808 Regardless of this option setting, when dealing with intercepted
2809 traffic, Squid always verifies that the destination IP address matches
2810 the Host header domain or IP (called 'authority form URL').
2812 This enforcement is performed to satisfy a MUST-level requirement in
2813 RFC 2616 section 14.23: "The Host field value MUST represent the naming
2814 authority of the origin server or gateway given by the original URL".
2817 Squid always responds with an HTTP 409 (Conflict) error
2818 page and logs a security warning if there is no match.
2820 Squid verifies that the destination IP address matches
2821 the Host header for forward-proxy and reverse-proxy traffic
2822 as well. For those traffic types, Squid also enables the
2823 following checks, comparing the corresponding Host header
2824 and Request-URI components:
2826 * The host names (domain or IP) must be identical,
2827 but valueless or missing Host header disables all checks.
2828 For the two host names to match, both must be either IP
2831 * Port numbers must be identical, but if a port is missing
2832 the scheme-default port is assumed.
2835 When set to OFF (the default):
2836 Squid allows suspicious requests to continue but logs a
2837 security warning and blocks caching of the response.
2839 * Forward-proxy traffic is not checked at all.
2841 * Reverse-proxy traffic is not checked at all.
2843 * Intercepted traffic which passes verification is handled
2844 according to client_dst_passthru.
2846 * Intercepted requests which fail verification are sent
2847 to the client original destination instead of DIRECT.
2848 This overrides 'client_dst_passthru off'.
2850 For now suspicious intercepted CONNECT requests are always
2851 responded to with an HTTP 409 (Conflict) error page.
2856 As described in CVE-2009-0801 when the Host: header alone is used
2857 to determine the destination of a request it becomes trivial for
2858 malicious scripts on remote websites to bypass browser same-origin
2859 security policy and sandboxing protections.
2861 The cause of this is that such applets are allowed to perform their
2862 own HTTP stack, in which case the same-origin policy of the browser
2863 sandbox only verifies that the applet tries to contact the same IP
2864 as from where it was loaded at the IP level. The Host: header may
2865 be different from the connected IP and approved origin.
2869 NAME: client_dst_passthru
2872 LOC: Config.onoff.client_dst_passthru
2874 With NAT or TPROXY intercepted traffic Squid may pass the request
2875 directly to the original client destination IP or seek a faster
2876 source using the HTTP Host header.
2878 Using Host to locate alternative servers can provide faster
2879 connectivity with a range of failure recovery options.
2880 But can also lead to connectivity trouble when the client and
2881 server are attempting stateful interactions unaware of the proxy.
2883 This option (on by default) prevents alternative DNS entries being
2884 located to send intercepted traffic DIRECT to an origin server.
2885 The clients original destination IP and port will be used instead.
2887 Regardless of this option setting, when dealing with intercepted
2888 traffic Squid will verify the Host: header and any traffic which
2889 fails Host verification will be treated as if this option were ON.
2891 see host_verify_strict for details on the verification process.
2896 -----------------------------------------------------------------------------
2899 NAME: tls_outgoing_options
2900 IFDEF: USE_GNUTLS||USE_OPENSSL
2901 TYPE: securePeerOptions
2902 DEFAULT: min-version=1.0
2903 LOC: Security::ProxyOutgoingConfig
2905 disable Do not support https:// URLs.
2907 cert=/path/to/client/certificate
2908 A client X.509 certificate to use when connecting.
2910 key=/path/to/client/private_key
2911 The private key corresponding to the cert= above.
2913 If key= is not specified cert= is assumed to
2914 reference a PEM file containing both the certificate
2917 cipher=... The list of valid TLS ciphers to use.
2920 The minimum TLS protocol version to permit.
2921 To control SSLv3 use the options= parameter.
2922 Supported Values: 1.0 (default), 1.1, 1.2, 1.3
2924 options=... Specify various TLS/SSL implementation options.
2926 OpenSSL options most important are:
2928 NO_SSLv3 Disallow the use of SSLv3
2931 Always create a new key when using
2932 temporary/ephemeral DH key exchanges
2935 Disable use of RFC5077 session tickets.
2936 Some servers may have problems
2937 understanding the TLS extension due
2938 to ambiguous specification in RFC4507.
2940 ALL Enable various bug workarounds
2941 suggested as "harmless" by OpenSSL
2942 Be warned that this reduces SSL/TLS
2943 strength to some attacks.
2945 See the OpenSSL SSL_CTX_set_options documentation
2946 for a more complete list.
2948 GnuTLS options most important are:
2951 Disable use of RFC5077 session tickets.
2952 Some servers may have problems
2953 understanding the TLS extension due
2954 to ambiguous specification in RFC4507.
2956 See the GnuTLS Priority Strings documentation
2957 for a more complete list.
2958 http://www.gnutls.org/manual/gnutls.html#Priority-Strings
2961 cafile= PEM file containing CA certificates to use when verifying
2962 the peer certificate. May be repeated to load multiple files.
2964 capath= A directory containing additional CA certificates to
2965 use when verifying the peer certificate.
2966 Requires OpenSSL or LibreSSL.
2968 crlfile=... A certificate revocation list file to use when
2969 verifying the peer certificate.
2971 flags=... Specify various flags modifying the TLS implementation:
2974 Accept certificates even if they fail to
2977 Don't verify the peer certificate
2978 matches the server name
2981 Whether to use the system Trusted CAs. Default is ON.
2983 domain= The peer name as advertised in its certificate.
2984 Used for verifying the correctness of the received peer
2985 certificate. If not specified the peer hostname will be
2991 -----------------------------------------------------------------------------
2994 NAME: ssl_unclean_shutdown
2998 LOC: Config.SSL.unclean_shutdown
3000 Some browsers (especially MSIE) bugs out on SSL shutdown
3007 LOC: Config.SSL.ssl_engine
3010 The OpenSSL engine to use. You will need to set this if you
3011 would like to use hardware SSL acceleration for example.
3014 NAME: sslproxy_session_ttl
3017 LOC: Config.SSL.session_ttl
3020 Sets the timeout value for SSL sessions
3023 NAME: sslproxy_session_cache_size
3026 LOC: Config.SSL.sessionCacheSize
3029 Sets the cache size to use for ssl session
3032 NAME: sslproxy_foreign_intermediate_certs
3035 LOC: Config.ssl_client.foreignIntermediateCertsPath
3038 Many origin servers fail to send their full server certificate
3039 chain for verification, assuming the client already has or can
3040 easily locate any missing intermediate certificates.
3042 Squid uses the certificates from the specified file to fill in
3043 these missing chains when trying to validate origin server
3046 The file is expected to contain zero or more PEM-encoded
3047 intermediate certificates. These certificates are not treated
3048 as trusted root certificates, and any self-signed certificate in
3049 this file will be ignored.
3052 NAME: sslproxy_cert_sign_hash
3055 LOC: Config.SSL.certSignHash
3058 Sets the hashing algorithm to use when signing generated certificates.
3059 Valid algorithm names depend on the OpenSSL library used. The following
3060 names are usually available: sha1, sha256, sha512, and md5. Please see
3061 your OpenSSL library manual for the available hashes. By default, Squids
3062 that support this option use sha256 hashes.
3064 Squid does not forcefully purge cached certificates that were generated
3065 with an algorithm other than the currently configured one. They remain
3066 in the cache, subject to the regular cache eviction policy, and become
3067 useful if the algorithm changes again.
3072 TYPE: sslproxy_ssl_bump
3073 LOC: Config.accessList.ssl_bump
3074 DEFAULT_DOC: Become a TCP tunnel without decrypting proxied traffic.
3077 This option is consulted when a CONNECT request is received on
3078 an http_port (or a new connection is intercepted at an
3079 https_port), provided that port was configured with an ssl-bump
3080 flag. The subsequent data on the connection is either treated as
3081 HTTPS and decrypted OR tunneled at TCP level without decryption,
3082 depending on the first matching bumping "action".
3084 ssl_bump <action> [!]acl ...
3086 The following bumping actions are currently supported:
3089 Become a TCP tunnel without decrypting proxied traffic.
3090 This is the default action.
3093 When used on step SslBump1, establishes a secure connection
3094 with the client first, then connect to the server.
3095 When used on step SslBump2 or SslBump3, establishes a secure
3096 connection with the server and, using a mimicked server
3097 certificate, with the client.
3100 Receive client (step SslBump1) or server (step SslBump2)
3101 certificate while preserving the possibility of splicing the
3102 connection. Peeking at the server certificate (during step 2)
3103 usually precludes bumping of the connection at step 3.
3106 Receive client (step SslBump1) or server (step SslBump2)
3107 certificate while preserving the possibility of bumping the
3108 connection. Staring at the server certificate (during step 2)
3109 usually precludes splicing of the connection at step 3.
3112 Close client and server connections.
3114 Backward compatibility actions available at step SslBump1:
3117 Bump the connection. Establish a secure connection with the
3118 client first, then connect to the server. This old mode does
3119 not allow Squid to mimic server SSL certificate and does not
3120 work with intercepted SSL connections.
3123 Bump the connection. Establish a secure connection with the
3124 server first, then establish a secure connection with the
3125 client, using a mimicked server certificate. Works with both
3126 CONNECT requests and intercepted SSL connections, but does
3127 not allow to make decisions based on SSL handshake info.
3130 Decide whether to bump or splice the connection based on
3131 client-to-squid and server-to-squid SSL hello messages.
3135 Same as the "splice" action.
3137 All ssl_bump rules are evaluated at each of the supported bumping
3138 steps. Rules with actions that are impossible at the current step are
3139 ignored. The first matching ssl_bump action wins and is applied at the
3140 end of the current step. If no rules match, the splice action is used.
3141 See the at_step ACL for a list of the supported SslBump steps.
3143 This clause supports both fast and slow acl types.
3144 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
3146 See also: http_port ssl-bump, https_port ssl-bump, and acl at_step.
3149 # Example: Bump all TLS connections except those originating from
3150 # localhost or those going to example.com.
3152 acl broken_sites ssl::server_name .example.com
3153 ssl_bump splice localhost
3154 ssl_bump splice broken_sites
3158 NAME: sslproxy_cert_error
3161 DEFAULT_DOC: Server certificate errors terminate the transaction.
3162 LOC: Config.ssl_client.cert_error
3165 Use this ACL to bypass server certificate validation errors.
3167 For example, the following lines will bypass all validation errors
3168 when talking to servers for example.com. All other
3169 validation errors will result in ERR_SECURE_CONNECT_FAIL error.
3171 acl BrokenButTrustedServers dstdomain example.com
3172 sslproxy_cert_error allow BrokenButTrustedServers
3173 sslproxy_cert_error deny all
3175 This clause only supports fast acl types.
3176 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
3177 Using slow acl types may result in server crashes
3179 Without this option, all server certificate validation errors
3180 terminate the transaction to protect Squid and the client.
3182 SQUID_X509_V_ERR_INFINITE_VALIDATION error cannot be bypassed
3183 but should not happen unless your OpenSSL library is buggy.
3186 Bypassing validation errors is dangerous because an
3187 error usually implies that the server cannot be trusted
3188 and the connection may be insecure.
3190 See also: sslproxy_flags and DONT_VERIFY_PEER.
3193 NAME: sslproxy_cert_sign
3196 POSTSCRIPTUM: signUntrusted ssl::certUntrusted
3197 POSTSCRIPTUM: signSelf ssl::certSelfSigned
3198 POSTSCRIPTUM: signTrusted all
3199 TYPE: sslproxy_cert_sign
3200 LOC: Config.ssl_client.cert_sign
3203 sslproxy_cert_sign <signing algorithm> acl ...
3205 The following certificate signing algorithms are supported:
3208 Sign using the configured CA certificate which is usually
3209 placed in and trusted by end-user browsers. This is the
3210 default for trusted origin server certificates.
3213 Sign to guarantee an X509_V_ERR_CERT_UNTRUSTED browser error.
3214 This is the default for untrusted origin server certificates
3215 that are not self-signed (see ssl::certUntrusted).
3218 Sign using a self-signed certificate with the right CN to
3219 generate a X509_V_ERR_DEPTH_ZERO_SELF_SIGNED_CERT error in the
3220 browser. This is the default for self-signed origin server
3221 certificates (see ssl::certSelfSigned).
3223 This clause only supports fast acl types.
3225 When sslproxy_cert_sign acl(s) match, Squid uses the corresponding
3226 signing algorithm to generate the certificate and ignores all
3227 subsequent sslproxy_cert_sign options (the first match wins). If no
3228 acl(s) match, the default signing algorithm is determined by errors
3229 detected when obtaining and validating the origin server certificate.
3231 WARNING: SQUID_X509_V_ERR_DOMAIN_MISMATCH and ssl:certDomainMismatch can
3232 be used with sslproxy_cert_adapt, but if and only if Squid is bumping a
3233 CONNECT request that carries a domain name. In all other cases (CONNECT
3234 to an IP address or an intercepted SSL connection), Squid cannot detect
3235 the domain mismatch at certificate generation time when
3236 bump-server-first is used.
3239 NAME: sslproxy_cert_adapt
3242 TYPE: sslproxy_cert_adapt
3243 LOC: Config.ssl_client.cert_adapt
3246 sslproxy_cert_adapt <adaptation algorithm> acl ...
3248 The following certificate adaptation algorithms are supported:
3251 Sets the "Not After" property to the "Not After" property of
3252 the CA certificate used to sign generated certificates.
3255 Sets the "Not Before" property to the "Not Before" property of
3256 the CA certificate used to sign generated certificates.
3258 setCommonName or setCommonName{CN}
3259 Sets Subject.CN property to the host name specified as a
3260 CN parameter or, if no explicit CN parameter was specified,
3261 extracted from the CONNECT request. It is a misconfiguration
3262 to use setCommonName without an explicit parameter for
3263 intercepted or tproxied SSL connections.
3265 This clause only supports fast acl types.
3267 Squid first groups sslproxy_cert_adapt options by adaptation algorithm.
3268 Within a group, when sslproxy_cert_adapt acl(s) match, Squid uses the
3269 corresponding adaptation algorithm to generate the certificate and
3270 ignores all subsequent sslproxy_cert_adapt options in that algorithm's
3271 group (i.e., the first match wins within each algorithm group). If no
3272 acl(s) match, the default mimicking action takes place.
3274 WARNING: SQUID_X509_V_ERR_DOMAIN_MISMATCH and ssl:certDomainMismatch can
3275 be used with sslproxy_cert_adapt, but if and only if Squid is bumping a
3276 CONNECT request that carries a domain name. In all other cases (CONNECT
3277 to an IP address or an intercepted SSL connection), Squid cannot detect
3278 the domain mismatch at certificate generation time when
3279 bump-server-first is used.
3282 NAME: sslpassword_program
3285 LOC: Config.Program.ssl_password
3288 Specify a program used for entering SSL key passphrases
3289 when using encrypted SSL certificate keys. If not specified
3290 keys must either be unencrypted, or Squid started with the -N
3291 option to allow it to query interactively for the passphrase.
3293 The key file name is given as argument to the program allowing
3294 selection of the right password if you have multiple encrypted
3299 OPTIONS RELATING TO EXTERNAL SSL_CRTD
3300 -----------------------------------------------------------------------------
3303 NAME: sslcrtd_program
3306 DEFAULT: @DEFAULT_SSL_CRTD@ -s @DEFAULT_SSL_DB_DIR@ -M 4MB
3307 LOC: Ssl::TheConfig.ssl_crtd
3309 Specify the location and options of the executable for certificate
3312 @DEFAULT_SSL_CRTD@ program can use a disk cache to improve response
3313 times on repeated requests. To enable caching, specify -s and -M
3314 parameters. If those parameters are not given, the program generates
3315 a new certificate on every request.
3317 For more information use:
3318 @DEFAULT_SSL_CRTD@ -h
3321 NAME: sslcrtd_children
3322 TYPE: HelperChildConfig
3324 DEFAULT: 32 startup=5 idle=1
3325 LOC: Ssl::TheConfig.ssl_crtdChildren
3327 Specifies the maximum number of certificate generation processes that
3328 Squid may spawn (numberofchildren) and several related options. Using
3329 too few of these helper processes (a.k.a. "helpers") creates request
3330 queues. Using too many helpers wastes your system resources. Squid
3331 does not support spawning more than 32 helpers.
3333 Usage: numberofchildren [option]...
3335 The startup= and idle= options allow some measure of skew in your
3340 Sets the minimum number of processes to spawn when Squid
3341 starts or reconfigures. When set to zero the first request will
3342 cause spawning of the first child process to handle it.
3344 Starting too few children temporary slows Squid under load while it
3345 tries to spawn enough additional processes to cope with traffic.
3349 Sets a minimum of how many processes Squid is to try and keep available
3350 at all times. When traffic begins to rise above what the existing
3351 processes can handle this many more will be spawned up to the maximum
3352 configured. A minimum setting of 1 is required.
3356 Sets the maximum number of queued requests. A request is queued when
3357 no existing child is idle and no new child can be started due to
3358 numberofchildren limit. If the queued requests exceed queue size for
3359 more than 3 minutes squid aborts its operation. The default value is
3360 set to 2*numberofchildren.
3362 You must have at least one ssl_crtd process.
3365 NAME: sslcrtvalidator_program
3369 LOC: Ssl::TheConfig.ssl_crt_validator
3371 Specify the location and options of the executable for ssl_crt_validator
3374 Usage: sslcrtvalidator_program [ttl=n] [cache=n] path ...
3377 ttl=n TTL in seconds for cached results. The default is 60 secs
3378 cache=n limit the result cache size. The default value is 2048
3381 NAME: sslcrtvalidator_children
3382 TYPE: HelperChildConfig
3384 DEFAULT: 32 startup=5 idle=1 concurrency=1
3385 LOC: Ssl::TheConfig.ssl_crt_validator_Children
3387 Specifies the maximum number of certificate validation processes that
3388 Squid may spawn (numberofchildren) and several related options. Using
3389 too few of these helper processes (a.k.a. "helpers") creates request
3390 queues. Using too many helpers wastes your system resources. Squid
3391 does not support spawning more than 32 helpers.
3393 Usage: numberofchildren [option]...
3395 The startup= and idle= options allow some measure of skew in your
3400 Sets the minimum number of processes to spawn when Squid
3401 starts or reconfigures. When set to zero the first request will
3402 cause spawning of the first child process to handle it.
3404 Starting too few children temporary slows Squid under load while it
3405 tries to spawn enough additional processes to cope with traffic.
3409 Sets a minimum of how many processes Squid is to try and keep available
3410 at all times. When traffic begins to rise above what the existing
3411 processes can handle this many more will be spawned up to the maximum
3412 configured. A minimum setting of 1 is required.
3416 The number of requests each certificate validator helper can handle in
3417 parallel. A value of 0 indicates the certficate validator does not
3418 support concurrency. Defaults to 1.
3420 When this directive is set to a value >= 1 then the protocol
3421 used to communicate with the helper is modified to include
3422 a request ID in front of the request/response. The request
3423 ID from the request must be echoed back with the response
3428 Sets the maximum number of queued requests. A request is queued when
3429 no existing child can accept it due to concurrency limit and no new
3430 child can be started due to numberofchildren limit. If the queued
3431 requests exceed queue size for more than 3 minutes squid aborts its
3432 operation. The default value is set to 2*numberofchildren.
3434 You must have at least one ssl_crt_validator process.
3438 OPTIONS WHICH AFFECT THE NEIGHBOR SELECTION ALGORITHM
3439 -----------------------------------------------------------------------------
3447 To specify other caches in a hierarchy, use the format:
3449 cache_peer hostname type http-port icp-port [options]
3454 # hostname type port port options
3455 # -------------------- -------- ----- ----- -----------
3456 cache_peer parent.foo.net parent 3128 3130 default
3457 cache_peer sib1.foo.net sibling 3128 3130 proxy-only
3458 cache_peer sib2.foo.net sibling 3128 3130 proxy-only
3459 cache_peer example.com parent 80 0 default
3460 cache_peer cdn.example.com sibling 3128 0
3462 type: either 'parent', 'sibling', or 'multicast'.
3464 proxy-port: The port number where the peer accept HTTP requests.
3465 For other Squid proxies this is usually 3128
3466 For web servers this is usually 80
3468 icp-port: Used for querying neighbor caches about objects.
3469 Set to 0 if the peer does not support ICP or HTCP.
3470 See ICP and HTCP options below for additional details.
3473 ==== ICP OPTIONS ====
3475 You MUST also set icp_port and icp_access explicitly when using these options.
3476 The defaults will prevent peer traffic using ICP.
3479 no-query Disable ICP queries to this neighbor.
3482 Indicates the named peer is a member of a multicast group.
3483 ICP queries will not be sent directly to the peer, but ICP
3484 replies will be accepted from it.
3486 closest-only Indicates that, for ICP_OP_MISS replies, we'll only forward
3487 CLOSEST_PARENT_MISSes and never FIRST_PARENT_MISSes.
3490 To only send ICP queries to this neighbor infrequently.
3491 This is used to keep the neighbor round trip time updated
3492 and is usually used in conjunction with weighted-round-robin.
3495 ==== HTCP OPTIONS ====
3497 You MUST also set htcp_port and htcp_access explicitly when using these options.
3498 The defaults will prevent peer traffic using HTCP.
3501 htcp Send HTCP, instead of ICP, queries to the neighbor.
3502 You probably also want to set the "icp-port" to 4827
3503 instead of 3130. This directive accepts a comma separated
3504 list of options described below.
3506 htcp=oldsquid Send HTCP to old Squid versions (2.5 or earlier).
3508 htcp=no-clr Send HTCP to the neighbor but without
3509 sending any CLR requests. This cannot be used with
3512 htcp=only-clr Send HTCP to the neighbor but ONLY CLR requests.
3513 This cannot be used with no-clr.
3516 Send HTCP to the neighbor including CLRs but only when
3517 they do not result from PURGE requests.
3520 Forward any HTCP CLR requests this proxy receives to the peer.
3523 ==== PEER SELECTION METHODS ====
3525 The default peer selection method is ICP, with the first responding peer
3526 being used as source. These options can be used for better load balancing.
3529 default This is a parent cache which can be used as a "last-resort"
3530 if a peer cannot be located by any of the peer-selection methods.
3531 If specified more than once, only the first is used.
3533 round-robin Load-Balance parents which should be used in a round-robin
3534 fashion in the absence of any ICP queries.
3535 weight=N can be used to add bias.
3537 weighted-round-robin
3538 Load-Balance parents which should be used in a round-robin
3539 fashion with the frequency of each parent being based on the
3540 round trip time. Closer parents are used more often.
3541 Usually used for background-ping parents.
3542 weight=N can be used to add bias.
3544 carp Load-Balance parents which should be used as a CARP array.
3545 The requests will be distributed among the parents based on the
3546 CARP load balancing hash function based on their weight.
3548 userhash Load-balance parents based on the client proxy_auth or ident username.
3550 sourcehash Load-balance parents based on the client source IP.
3553 To be used only for cache peers of type "multicast".
3554 ALL members of this multicast group have "sibling"
3555 relationship with it, not "parent". This is to a multicast
3556 group when the requested object would be fetched only from
3557 a "parent" cache, anyway. It's useful, e.g., when
3558 configuring a pool of redundant Squid proxies, being
3559 members of the same multicast group.
3562 ==== PEER SELECTION OPTIONS ====
3564 weight=N use to affect the selection of a peer during any weighted
3565 peer-selection mechanisms.
3566 The weight must be an integer; default is 1,
3567 larger weights are favored more.
3568 This option does not affect parent selection if a peering
3569 protocol is not in use.
3571 basetime=N Specify a base amount to be subtracted from round trip
3573 It is subtracted before division by weight in calculating
3574 which parent to fectch from. If the rtt is less than the
3575 base time the rtt is set to a minimal value.
3577 ttl=N Specify a TTL to use when sending multicast ICP queries
3579 Only useful when sending to a multicast group.
3580 Because we don't accept ICP replies from random
3581 hosts, you must configure other group members as
3582 peers with the 'multicast-responder' option.
3584 no-delay To prevent access to this neighbor from influencing the
3587 digest-url=URL Tell Squid to fetch the cache digest (if digests are
3588 enabled) for this host from the specified URL rather
3589 than the Squid default location.
3592 ==== CARP OPTIONS ====
3594 carp-key=key-specification
3595 use a different key than the full URL to hash against the peer.
3596 the key-specification is a comma-separated list of the keywords
3597 scheme, host, port, path, params
3598 Order is not important.
3600 ==== ACCELERATOR / REVERSE-PROXY OPTIONS ====
3602 originserver Causes this parent to be contacted as an origin server.
3603 Meant to be used in accelerator setups when the peer
3607 Set the Host header of requests forwarded to this peer.
3608 Useful in accelerator setups where the server (peer)
3609 expects a certain domain name but clients may request
3610 others. ie example.com or www.example.com
3612 no-digest Disable request of cache digests.
3615 Disables requesting ICMP RTT database (NetDB).
3618 ==== AUTHENTICATION OPTIONS ====
3621 If this is a personal/workgroup proxy and your parent
3622 requires proxy authentication.
3624 Note: The string can include URL escapes (i.e. %20 for
3625 spaces). This also means % must be written as %%.
3628 Send login details received from client to this peer.
3629 Both Proxy- and WWW-Authorization headers are passed
3630 without alteration to the peer.
3631 Authentication is not required by Squid for this to work.
3633 Note: This will pass any form of authentication but
3634 only Basic auth will work through a proxy unless the
3635 connection-auth options are also used.
3637 login=PASS Send login details received from client to this peer.
3638 Authentication is not required by this option.
3640 If there are no client-provided authentication headers
3641 to pass on, but username and password are available
3642 from an external ACL user= and password= result tags
3643 they may be sent instead.
3645 Note: To combine this with proxy_auth both proxies must
3646 share the same user database as HTTP only allows for
3647 a single login (one for proxy, one for origin server).
3648 Also be warned this will expose your users proxy
3649 password to the peer. USE WITH CAUTION
3652 Send the username to the upstream cache, but with a
3653 fixed password. This is meant to be used when the peer
3654 is in another administrative domain, but it is still
3655 needed to identify each user.
3656 The star can optionally be followed by some extra
3657 information which is added to the username. This can
3658 be used to identify this proxy to the peer, similar to
3659 the login=username:password option above.
3662 If this is a personal/workgroup proxy and your parent
3663 requires a secure proxy authentication.
3664 The first principal from the default keytab or defined by
3665 the environment variable KRB5_KTNAME will be used.
3667 WARNING: The connection may transmit requests from multiple
3668 clients. Negotiate often assumes end-to-end authentication
3669 and a single-client. Which is not strictly true here.
3671 login=NEGOTIATE:principal_name
3672 If this is a personal/workgroup proxy and your parent
3673 requires a secure proxy authentication.
3674 The principal principal_name from the default keytab or
3675 defined by the environment variable KRB5_KTNAME will be
3678 WARNING: The connection may transmit requests from multiple
3679 clients. Negotiate often assumes end-to-end authentication
3680 and a single-client. Which is not strictly true here.
3682 connection-auth=on|off
3683 Tell Squid that this peer does or not support Microsoft
3684 connection oriented authentication, and any such
3685 challenges received from there should be ignored.
3686 Default is auto to automatically determine the status
3690 Do not use a keytab to authenticate to a peer when
3691 login=NEGOTIATE is specified. Let the GSSAPI
3692 implementation determine which already existing
3693 credentials cache to use instead.
3696 ==== SSL / HTTPS / TLS OPTIONS ====
3698 tls Encrypt connections to this peer with TLS.
3700 sslcert=/path/to/ssl/certificate
3701 A client X.509 certificate to use when connecting to
3704 sslkey=/path/to/ssl/key
3705 The private key corresponding to sslcert above.
3707 If sslkey= is not specified sslcert= is assumed to
3708 reference a PEM file containing both the certificate
3711 sslcipher=... The list of valid SSL ciphers to use when connecting
3715 The minimum TLS protocol version to permit. To control
3716 SSLv3 use the tls-options= parameter.
3717 Supported Values: 1.0 (default), 1.1, 1.2
3719 tls-options=... Specify various TLS implementation options.
3721 OpenSSL options most important are:
3723 NO_SSLv3 Disallow the use of SSLv3
3726 Always create a new key when using
3727 temporary/ephemeral DH key exchanges
3730 Disable use of RFC5077 session tickets.
3731 Some servers may have problems
3732 understanding the TLS extension due
3733 to ambiguous specification in RFC4507.
3735 ALL Enable various bug workarounds
3736 suggested as "harmless" by OpenSSL
3737 Be warned that this reduces SSL/TLS
3738 strength to some attacks.
3740 See the OpenSSL SSL_CTX_set_options documentation for a
3743 GnuTLS options most important are:
3746 Disable use of RFC5077 session tickets.
3747 Some servers may have problems
3748 understanding the TLS extension due
3749 to ambiguous specification in RFC4507.
3751 See the GnuTLS Priority Strings documentation
3752 for a more complete list.
3753 http://www.gnutls.org/manual/gnutls.html#Priority-Strings
3755 tls-cafile= PEM file containing CA certificates to use when verifying
3756 the peer certificate. May be repeated to load multiple files.
3758 sslcapath=... A directory containing additional CA certificates to
3759 use when verifying the peer certificate.
3760 Requires OpenSSL or LibreSSL.
3762 sslcrlfile=... A certificate revocation list file to use when
3763 verifying the peer certificate.
3765 sslflags=... Specify various flags modifying the SSL implementation:
3768 Accept certificates even if they fail to
3772 Don't verify the peer certificate
3773 matches the server name
3775 ssldomain= The peer name as advertised in it's certificate.
3776 Used for verifying the correctness of the received peer
3777 certificate. If not specified the peer hostname will be
3780 front-end-https[=off|on|auto]
3781 Enable the "Front-End-Https: On" header needed when
3782 using Squid as a SSL frontend in front of Microsoft OWA.
3783 See MS KB document Q307347 for details on this header.
3784 If set to auto the header will only be added if the
3785 request is forwarded as a https:// URL.
3787 tls-default-ca[=off]
3788 Whether to use the system Trusted CAs. Default is ON.
3790 tls-no-npn Do not use the TLS NPN extension to advertise HTTP/1.1.
3792 ==== GENERAL OPTIONS ====
3795 A peer-specific connect timeout.
3796 Also see the peer_connect_timeout directive.
3798 connect-fail-limit=N
3799 How many times connecting to a peer must fail before
3800 it is marked as down. Standby connection failures
3801 count towards this limit. Default is 10.
3803 allow-miss Disable Squid's use of only-if-cached when forwarding
3804 requests to siblings. This is primarily useful when
3805 icp_hit_stale is used by the sibling. Excessive use
3806 of this option may result in forwarding loops. One way
3807 to prevent peering loops when using this option, is to
3808 deny cache peer usage on requests from a peer:
3810 cache_peer_access peerName deny fromPeer
3812 max-conn=N Limit the number of concurrent connections the Squid
3813 may open to this peer, including already opened idle
3814 and standby connections. There is no peer-specific
3815 connection limit by default.
3817 A peer exceeding the limit is not used for new
3818 requests unless a standby connection is available.
3820 max-conn currently works poorly with idle persistent
3821 connections: When a peer reaches its max-conn limit,
3822 and there are idle persistent connections to the peer,
3823 the peer may not be selected because the limiting code
3824 does not know whether Squid can reuse those idle
3827 standby=N Maintain a pool of N "hot standby" connections to an
3828 UP peer, available for requests when no idle
3829 persistent connection is available (or safe) to use.
3830 By default and with zero N, no such pool is maintained.
3831 N must not exceed the max-conn limit (if any).
3833 At start or after reconfiguration, Squid opens new TCP
3834 standby connections until there are N connections
3835 available and then replenishes the standby pool as
3836 opened connections are used up for requests. A used
3837 connection never goes back to the standby pool, but
3838 may go to the regular idle persistent connection pool
3839 shared by all peers and origin servers.
3841 Squid never opens multiple new standby connections
3842 concurrently. This one-at-a-time approach minimizes
3843 flooding-like effect on peers. Furthermore, just a few
3844 standby connections should be sufficient in most cases
3845 to supply most new requests with a ready-to-use
3848 Standby connections obey server_idle_pconn_timeout.
3849 For the feature to work as intended, the peer must be
3850 configured to accept and keep them open longer than
3851 the idle timeout at the connecting Squid, to minimize
3852 race conditions typical to idle used persistent
3853 connections. Default request_timeout and
3854 server_idle_pconn_timeout values ensure such a
3857 name=xxx Unique name for the peer.
3858 Required if you have multiple peers on the same host
3859 but different ports.
3860 This name can be used in cache_peer_access and similar
3861 directives to identify the peer.
3862 Can be used by outgoing access controls through the
3865 no-tproxy Do not use the client-spoof TPROXY support when forwarding
3866 requests to this peer. Use normal address selection instead.
3867 This overrides the spoof_client_ip ACL.
3869 proxy-only objects fetched from the peer will not be stored locally.
3873 NAME: cache_peer_access
3876 DEFAULT_DOC: No peer usage restrictions.
3879 Restricts usage of cache_peer proxies.
3882 cache_peer_access peer-name allow|deny [!]aclname ...
3884 For the required peer-name parameter, use either the value of the
3885 cache_peer name=value parameter or, if name=value is missing, the
3886 cache_peer hostname parameter.
3888 This directive narrows down the selection of peering candidates, but
3889 does not determine the order in which the selected candidates are
3890 contacted. That order is determined by the peer selection algorithms
3891 (see PEER SELECTION sections in the cache_peer documentation).
3893 If a deny rule matches, the corresponding peer will not be contacted
3894 for the current transaction -- Squid will not send ICP queries and
3895 will not forward HTTP requests to that peer. An allow match leaves
3896 the corresponding peer in the selection. The first match for a given
3897 peer wins for that peer.
3899 The relative order of cache_peer_access directives for the same peer
3900 matters. The relative order of any two cache_peer_access directives
3901 for different peers does not matter. To ease interpretation, it is a
3902 good idea to group cache_peer_access directives for the same peer
3905 A single cache_peer_access directive may be evaluated multiple times
3906 for a given transaction because individual peer selection algorithms
3907 may check it independently from each other. These redundant checks
3908 may be optimized away in future Squid versions.
3910 This clause only supports fast acl types.
3911 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
3915 NAME: neighbor_type_domain
3916 TYPE: hostdomaintype
3918 DEFAULT_DOC: The peer type from cache_peer directive is used for all requests to that peer.
3921 Modify the cache_peer neighbor type when passing requests
3922 about specific domains to the peer.
3925 neighbor_type_domain neighbor parent|sibling domain domain ...
3928 cache_peer foo.example.com parent 3128 3130
3929 neighbor_type_domain foo.example.com sibling .au .de
3931 The above configuration treats all requests to foo.example.com as a
3932 parent proxy unless the request is for a .au or .de ccTLD domain name.
3935 NAME: dead_peer_timeout
3939 LOC: Config.Timeout.deadPeer
3941 This controls how long Squid waits to declare a peer cache
3942 as "dead." If there are no ICP replies received in this
3943 amount of time, Squid will declare the peer dead and not
3944 expect to receive any further ICP replies. However, it
3945 continues to send ICP queries, and will mark the peer as
3946 alive upon receipt of the first subsequent ICP reply.
3948 This timeout also affects when Squid expects to receive ICP
3949 replies from peers. If more than 'dead_peer' seconds have
3950 passed since the last ICP reply was received, Squid will not
3951 expect to receive an ICP reply on the next query. Thus, if
3952 your time between requests is greater than this timeout, you
3953 will see a lot of requests sent DIRECT to origin servers
3954 instead of to your parents.
3957 NAME: forward_max_tries
3960 LOC: Config.forward_max_tries
3962 Limits the number of attempts to forward the request.
3964 For the purpose of this limit, Squid counts all high-level request
3965 forwarding attempts, including any same-destination retries after
3966 certain persistent connection failures and any attempts to use a
3967 different peer. However, low-level connection reopening attempts
3968 (enabled using connect_retries) are not counted.
3970 See also: forward_timeout and connect_retries.
3974 MEMORY CACHE OPTIONS
3975 -----------------------------------------------------------------------------
3982 LOC: Config.memMaxSize
3984 NOTE: THIS PARAMETER DOES NOT SPECIFY THE MAXIMUM PROCESS SIZE.
3985 IT ONLY PLACES A LIMIT ON HOW MUCH ADDITIONAL MEMORY SQUID WILL
3986 USE AS A MEMORY CACHE OF OBJECTS. SQUID USES MEMORY FOR OTHER
3987 THINGS AS WELL. SEE THE SQUID FAQ SECTION 8 FOR DETAILS.
3989 'cache_mem' specifies the ideal amount of memory to be used
3991 * In-Transit objects
3993 * Negative-Cached objects
3995 Data for these objects are stored in 4 KB blocks. This
3996 parameter specifies the ideal upper limit on the total size of
3997 4 KB blocks allocated. In-Transit objects take the highest
4000 In-transit objects have priority over the others. When
4001 additional space is needed for incoming data, negative-cached
4002 and hot objects will be released. In other words, the
4003 negative-cached and hot objects will fill up any unused space
4004 not needed for in-transit objects.
4006 If circumstances require, this limit will be exceeded.
4007 Specifically, if your incoming request rate requires more than
4008 'cache_mem' of memory to hold in-transit objects, Squid will
4009 exceed this limit to satisfy the new requests. When the load
4010 decreases, blocks will be freed until the high-water mark is
4011 reached. Thereafter, blocks will be used to store hot
4014 If shared memory caching is enabled, Squid does not use the shared
4015 cache space for in-transit objects, but they still consume as much
4016 local memory as they need. For more details about the shared memory
4017 cache, see memory_cache_shared.
4020 NAME: maximum_object_size_in_memory
4024 LOC: Config.Store.maxInMemObjSize
4026 Objects greater than this size will not be attempted to kept in
4027 the memory cache. This should be set high enough to keep objects
4028 accessed frequently in memory to improve performance whilst low
4029 enough to keep larger objects from hoarding cache_mem.
4032 NAME: memory_cache_shared
4035 LOC: Config.memShared
4037 DEFAULT_DOC: "on" where supported if doing memory caching with multiple SMP workers.
4039 Controls whether the memory cache is shared among SMP workers.
4041 The shared memory cache is meant to occupy cache_mem bytes and replace
4042 the non-shared memory cache, although some entities may still be
4043 cached locally by workers for now (e.g., internal and in-transit
4044 objects may be served from a local memory cache even if shared memory
4045 caching is enabled).
4047 By default, the memory cache is shared if and only if all of the
4048 following conditions are satisfied: Squid runs in SMP mode with
4049 multiple workers, cache_mem is positive, and Squid environment
4050 supports required IPC primitives (e.g., POSIX shared memory segments
4051 and GCC-style atomic operations).
4053 To avoid blocking locks, shared memory uses opportunistic algorithms
4054 that do not guarantee that every cachable entity that could have been
4055 shared among SMP workers will actually be shared.
4058 NAME: memory_cache_mode
4062 DEFAULT_DOC: Keep the most recently fetched objects in memory
4064 Controls which objects to keep in the memory cache (cache_mem)
4066 always Keep most recently fetched objects in memory (default)
4068 disk Only disk cache hits are kept in memory, which means
4069 an object must first be cached on disk and then hit
4070 a second time before cached in memory.
4072 network Only objects fetched from network is kept in memory
4075 NAME: memory_replacement_policy
4077 LOC: Config.memPolicy
4080 The memory replacement policy parameter determines which
4081 objects are purged from memory when memory space is needed.
4083 See cache_replacement_policy for details on algorithms.
4088 -----------------------------------------------------------------------------
4091 NAME: cache_replacement_policy
4093 LOC: Config.replPolicy
4096 The cache replacement policy parameter determines which
4097 objects are evicted (replaced) when disk space is needed.
4099 lru : Squid's original list based LRU policy
4100 heap GDSF : Greedy-Dual Size Frequency
4101 heap LFUDA: Least Frequently Used with Dynamic Aging
4102 heap LRU : LRU policy implemented using a heap
4104 Applies to any cache_dir lines listed below this directive.
4106 The LRU policies keeps recently referenced objects.
4108 The heap GDSF policy optimizes object hit rate by keeping smaller
4109 popular objects in cache so it has a better chance of getting a
4110 hit. It achieves a lower byte hit rate than LFUDA though since
4111 it evicts larger (possibly popular) objects.
4113 The heap LFUDA policy keeps popular objects in cache regardless of
4114 their size and thus optimizes byte hit rate at the expense of
4115 hit rate since one large, popular object will prevent many
4116 smaller, slightly less popular objects from being cached.
4118 Both policies utilize a dynamic aging mechanism that prevents
4119 cache pollution that can otherwise occur with frequency-based
4120 replacement policies.
4122 NOTE: if using the LFUDA replacement policy you should increase
4123 the value of maximum_object_size above its default of 4 MB to
4124 to maximize the potential byte hit rate improvement of LFUDA.
4126 For more information about the GDSF and LFUDA cache replacement
4127 policies see http://www.hpl.hp.com/techreports/1999/HPL-1999-69.html
4128 and http://fog.hpl.external.hp.com/techreports/98/HPL-98-173.html.
4131 NAME: minimum_object_size
4135 DEFAULT_DOC: no limit
4136 LOC: Config.Store.minObjectSize
4138 Objects smaller than this size will NOT be saved on disk. The
4139 value is specified in bytes, and the default is 0 KB, which
4140 means all responses can be stored.
4143 NAME: maximum_object_size
4147 LOC: Config.Store.maxObjectSize
4149 Set the default value for max-size parameter on any cache_dir.
4150 The value is specified in bytes, and the default is 4 MB.
4152 If you wish to get a high BYTES hit ratio, you should probably
4153 increase this (one 32 MB object hit counts for 3200 10KB
4156 If you wish to increase hit ratio more than you want to
4157 save bandwidth you should leave this low.
4159 NOTE: if using the LFUDA replacement policy you should increase
4160 this value to maximize the byte hit rate improvement of LFUDA!
4161 See cache_replacement_policy for a discussion of this policy.
4167 DEFAULT_DOC: No disk cache. Store cache ojects only in memory.
4168 LOC: Config.cacheSwap
4171 cache_dir Type Directory-Name Fs-specific-data [options]
4173 You can specify multiple cache_dir lines to spread the
4174 cache among different disk partitions.
4176 Type specifies the kind of storage system to use. Only "ufs"
4177 is built by default. To enable any of the other storage systems
4178 see the --enable-storeio configure option.
4180 'Directory' is a top-level directory where cache swap
4181 files will be stored. If you want to use an entire disk
4182 for caching, this can be the mount-point directory.
4183 The directory must exist and be writable by the Squid
4184 process. Squid will NOT create this directory for you.
4186 In SMP configurations, cache_dir must not precede the workers option
4187 and should use configuration macros or conditionals to give each
4188 worker interested in disk caching a dedicated cache directory.
4191 ==== The ufs store type ====
4193 "ufs" is the old well-known Squid storage format that has always
4197 cache_dir ufs Directory-Name Mbytes L1 L2 [options]
4199 'Mbytes' is the amount of disk space (MB) to use under this
4200 directory. The default is 100 MB. Change this to suit your
4201 configuration. Do NOT put the size of your disk drive here.
4202 Instead, if you want Squid to use the entire disk drive,
4203 subtract 20% and use that value.
4205 'L1' is the number of first-level subdirectories which
4206 will be created under the 'Directory'. The default is 16.
4208 'L2' is the number of second-level subdirectories which
4209 will be created under each first-level directory. The default
4213 ==== The aufs store type ====
4215 "aufs" uses the same storage format as "ufs", utilizing
4216 POSIX-threads to avoid blocking the main Squid process on
4217 disk-I/O. This was formerly known in Squid as async-io.
4220 cache_dir aufs Directory-Name Mbytes L1 L2 [options]
4222 see argument descriptions under ufs above
4225 ==== The diskd store type ====
4227 "diskd" uses the same storage format as "ufs", utilizing a
4228 separate process to avoid blocking the main Squid process on
4232 cache_dir diskd Directory-Name Mbytes L1 L2 [options] [Q1=n] [Q2=n]
4234 see argument descriptions under ufs above
4236 Q1 specifies the number of unacknowledged I/O requests when Squid
4237 stops opening new files. If this many messages are in the queues,
4238 Squid won't open new files. Default is 64
4240 Q2 specifies the number of unacknowledged messages when Squid
4241 starts blocking. If this many messages are in the queues,
4242 Squid blocks until it receives some replies. Default is 72
4244 When Q1 < Q2 (the default), the cache directory is optimized
4245 for lower response time at the expense of a decrease in hit
4246 ratio. If Q1 > Q2, the cache directory is optimized for
4247 higher hit ratio at the expense of an increase in response
4251 ==== The rock store type ====
4254 cache_dir rock Directory-Name Mbytes [options]
4256 The Rock Store type is a database-style storage. All cached
4257 entries are stored in a "database" file, using fixed-size slots.
4258 A single entry occupies one or more slots.
4260 If possible, Squid using Rock Store creates a dedicated kid
4261 process called "disker" to avoid blocking Squid worker(s) on disk
4262 I/O. One disker kid is created for each rock cache_dir. Diskers
4263 are created only when Squid, running in daemon mode, has support
4264 for the IpcIo disk I/O module.
4266 swap-timeout=msec: Squid will not start writing a miss to or
4267 reading a hit from disk if it estimates that the swap operation
4268 will take more than the specified number of milliseconds. By
4269 default and when set to zero, disables the disk I/O time limit
4270 enforcement. Ignored when using blocking I/O module because
4271 blocking synchronous I/O does not allow Squid to estimate the
4272 expected swap wait time.
4274 max-swap-rate=swaps/sec: Artificially limits disk access using
4275 the specified I/O rate limit. Swap out requests that
4276 would cause the average I/O rate to exceed the limit are
4277 delayed. Individual swap in requests (i.e., hits or reads) are
4278 not delayed, but they do contribute to measured swap rate and
4279 since they are placed in the same FIFO queue as swap out
4280 requests, they may wait longer if max-swap-rate is smaller.
4281 This is necessary on file systems that buffer "too
4282 many" writes and then start blocking Squid and other processes
4283 while committing those writes to disk. Usually used together
4284 with swap-timeout to avoid excessive delays and queue overflows
4285 when disk demand exceeds available disk "bandwidth". By default
4286 and when set to zero, disables the disk I/O rate limit
4287 enforcement. Currently supported by IpcIo module only.
4289 slot-size=bytes: The size of a database "record" used for
4290 storing cached responses. A cached response occupies at least
4291 one slot and all database I/O is done using individual slots so
4292 increasing this parameter leads to more disk space waste while
4293 decreasing it leads to more disk I/O overheads. Should be a
4294 multiple of your operating system I/O page size. Defaults to
4295 16KBytes. A housekeeping header is stored with each slot and
4296 smaller slot-sizes will be rejected. The header is smaller than
4300 ==== COMMON OPTIONS ====
4302 no-store no new objects should be stored to this cache_dir.
4304 min-size=n the minimum object size in bytes this cache_dir
4305 will accept. It's used to restrict a cache_dir
4306 to only store large objects (e.g. AUFS) while
4307 other stores are optimized for smaller objects
4311 max-size=n the maximum object size in bytes this cache_dir
4313 The value in maximum_object_size directive sets
4314 the default unless more specific details are
4315 available (ie a small store capacity).
4317 Note: To make optimal use of the max-size limits you should order
4318 the cache_dir lines with the smallest max-size value first.
4322 # Uncomment and adjust the following to add a disk cache directory.
4323 #cache_dir ufs @DEFAULT_SWAP_DIR@ 100 16 256
4327 NAME: store_dir_select_algorithm
4329 LOC: Config.store_dir_select_algorithm
4332 How Squid selects which cache_dir to use when the response
4333 object will fit into more than one.
4335 Regardless of which algorithm is used the cache_dir min-size
4336 and max-size parameters are obeyed. As such they can affect
4337 the selection algorithm by limiting the set of considered
4344 This algorithm is suited to caches with similar cache_dir
4345 sizes and disk speeds.
4347 The disk with the least I/O pending is selected.
4348 When there are multiple disks with the same I/O load ranking
4349 the cache_dir with most available capacity is selected.
4351 When a mix of cache_dir sizes are configured the faster disks
4352 have a naturally lower I/O loading and larger disks have more
4353 capacity. So space used to store objects and data throughput
4354 may be very unbalanced towards larger disks.
4359 This algorithm is suited to caches with unequal cache_dir
4362 Each cache_dir is selected in a rotation. The next suitable
4365 Available cache_dir capacity is only considered in relation
4366 to whether the object will fit and meets the min-size and
4367 max-size parameters.
4369 Disk I/O loading is only considered to prevent overload on slow
4370 disks. This algorithm does not spread objects by size, so any
4371 I/O loading per-disk may appear very unbalanced and volatile.
4373 If several cache_dirs use similar min-size, max-size, or other
4374 limits to to reject certain responses, then do not group such
4375 cache_dir lines together, to avoid round-robin selection bias
4376 towards the first cache_dir after the group. Instead, interleave
4377 cache_dir lines from different groups. For example:
4379 store_dir_select_algorithm round-robin
4380 cache_dir rock /hdd1 ... min-size=100000
4381 cache_dir rock /ssd1 ... max-size=99999
4382 cache_dir rock /hdd2 ... min-size=100000
4383 cache_dir rock /ssd2 ... max-size=99999
4384 cache_dir rock /hdd3 ... min-size=100000
4385 cache_dir rock /ssd3 ... max-size=99999
4388 NAME: max_open_disk_fds
4390 LOC: Config.max_open_disk_fds
4392 DEFAULT_DOC: no limit
4394 To avoid having disk as the I/O bottleneck Squid can optionally
4395 bypass the on-disk cache if more than this amount of disk file
4396 descriptors are open.
4398 A value of 0 indicates no limit.
4401 NAME: cache_swap_low
4402 COMMENT: (percent, 0-100)
4405 LOC: Config.Swap.lowWaterMark
4407 The low-water mark for AUFS/UFS/diskd cache object eviction by
4408 the cache_replacement_policy algorithm.
4410 Removal begins when the swap (disk) usage of a cache_dir is
4411 above this low-water mark and attempts to maintain utilization
4412 near the low-water mark.
4414 As swap utilization increases towards the high-water mark set
4415 by cache_swap_high object eviction becomes more agressive.
4417 The value difference in percentages between low- and high-water
4418 marks represent an eviction rate of 300 objects per second and
4419 the rate continues to scale in agressiveness by multiples of
4420 this above the high-water mark.
4422 Defaults are 90% and 95%. If you have a large cache, 5% could be
4423 hundreds of MB. If this is the case you may wish to set these
4424 numbers closer together.
4426 See also cache_swap_high and cache_replacement_policy
4429 NAME: cache_swap_high
4430 COMMENT: (percent, 0-100)
4433 LOC: Config.Swap.highWaterMark
4435 The high-water mark for AUFS/UFS/diskd cache object eviction by
4436 the cache_replacement_policy algorithm.
4438 Removal begins when the swap (disk) usage of a cache_dir is
4439 above the low-water mark set by cache_swap_low and attempts to
4440 maintain utilization near the low-water mark.
4442 As swap utilization increases towards this high-water mark object
4443 eviction becomes more agressive.
4445 The value difference in percentages between low- and high-water
4446 marks represent an eviction rate of 300 objects per second and
4447 the rate continues to scale in agressiveness by multiples of
4448 this above the high-water mark.
4450 Defaults are 90% and 95%. If you have a large cache, 5% could be
4451 hundreds of MB. If this is the case you may wish to set these
4452 numbers closer together.
4454 See also cache_swap_low and cache_replacement_policy
4459 -----------------------------------------------------------------------------
4466 DEFAULT_DOC: The format definitions squid, common, combined, referrer, useragent are built in.
4470 logformat <name> <format specification>
4472 Defines an access log format.
4474 The <format specification> is a string with embedded % format codes
4476 % format codes all follow the same basic structure where all
4477 components but the formatcode are optional and usually unnecessary,
4478 especially when dealing with common codes.
4480 % [encoding] [-] [[0]width] [{arg}] formatcode [{arg}]
4482 encoding escapes or otherwise protects "special" characters:
4484 " Quoted string encoding where quote(") and
4485 backslash(\) characters are \-escaped while
4486 CR, LF, and TAB characters are encoded as \r,
4487 \n, and \t two-character sequences.
4489 [ Custom Squid encoding where percent(%), square
4490 brackets([]), backslash(\) and characters with
4491 codes outside of [32,126] range are %-encoded.
4492 SP is not encoded. Used by log_mime_hdrs.
4494 # URL encoding (a.k.a. percent-encoding) where
4495 all URL unsafe and control characters (per RFC
4496 1738) are %-encoded.
4498 / Shell-like encoding where quote(") and
4499 backslash(\) characters are \-escaped while CR
4500 and LF characters are encoded as \r and \n
4501 two-character sequences. Values containing SP
4502 character(s) are surrounded by quotes(").
4504 ' Raw/as-is encoding with no escaping/quoting.
4506 Default encoding: When no explicit encoding is
4507 specified, each %code determines its own encoding.
4508 Most %codes use raw/as-is encoding, but some codes use
4509 a so called "pass-through URL encoding" where all URL
4510 unsafe and control characters (per RFC 1738) are
4511 %-encoded, but the percent character(%) is left as is.
4515 width minimum and/or maximum field width:
4516 [width_min][.width_max]
4517 When minimum starts with 0, the field is zero-padded.
4518 String values exceeding maximum width are truncated.
4520 {arg} argument such as header name etc. This field may be
4521 placed before or after the token, but not both at once.
4525 % a literal % character
4526 sn Unique sequence number per log line entry
4527 err_code The ID of an error response served by Squid or
4528 a similar internal error identifier.
4529 err_detail Additional err_code-dependent error information.
4530 note The annotation specified by the argument. Also
4531 logs the adaptation meta headers set by the
4532 adaptation_meta configuration parameter.
4533 If no argument given all annotations logged.
4534 The argument may include a separator to use with
4537 By default, multiple note values are separated with ","
4538 and multiple notes are separated with "\r\n".
4539 When logging named notes with %{name}note, the
4540 explicitly configured separator is used between note
4541 values. When logging all notes with %note, the
4542 explicitly configured separator is used between
4543 individual notes. There is currently no way to
4544 specify both value and notes separators when logging
4545 all notes with %note.
4546 master_xaction The master transaction identifier is an unsigned
4547 integer. These IDs are guaranteed to monotonically
4548 increase within a single worker process lifetime, with
4549 higher values corresponding to transactions that were
4550 accepted or initiated later. Due to current implementation
4551 deficiencies, some IDs are skipped (i.e. never logged).
4552 Concurrent workers and restarted workers use similar,
4553 overlapping sequences of master transaction IDs.
4555 Connection related format codes:
4557 >a Client source IP address
4559 >p Client source port
4560 >eui Client source EUI (MAC address, EUI-48 or EUI-64 identifier)
4561 >la Local IP address the client connected to
4562 >lp Local port number the client connected to
4563 >qos Client connection TOS/DSCP value set by Squid
4564 >nfmark Client connection netfilter packet MARK set by Squid
4566 la Local listening IP address the client connection was connected to.
4567 lp Local listening port number the client connection was connected to.
4569 <a Server IP address of the last server or peer connection
4570 <A Server FQDN or peer name
4571 <p Server port number of the last server or peer connection
4572 <la Local IP address of the last server or peer connection
4573 <lp Local port number of the last server or peer connection
4574 <qos Server connection TOS/DSCP value set by Squid
4575 <nfmark Server connection netfilter packet MARK set by Squid
4577 >handshake Raw client handshake
4578 Initial client bytes received by Squid on a newly
4579 accepted TCP connection or inside a just established
4580 CONNECT tunnel. Squid stops accumulating handshake
4581 bytes as soon as the handshake parser succeeds or
4582 fails (determining whether the client is using the
4585 For HTTP clients, the handshake is the request line.
4586 For TLS clients, the handshake consists of all TLS
4587 records up to and including the TLS record that
4588 contains the last byte of the first ClientHello
4589 message. For clients using an unsupported protocol,
4590 this field contains the bytes received by Squid at the
4591 time of the handshake parsing failure.
4593 See the on_unsupported_protocol directive for more
4594 information on Squid handshake traffic expectations.
4596 Current support is limited to these contexts:
4597 - http_port connections, but only when the
4598 on_unsupported_protocol directive is in use.
4599 - https_port connections (and CONNECT tunnels) that
4600 are subject to the ssl_bump peek or stare action.
4602 To protect binary handshake data, this field is always
4603 base64-encoded (RFC 4648 Section 4). If logformat
4604 field encoding is configured, that encoding is applied
4605 on top of base64. Otherwise, the computed base64 value
4608 Time related format codes:
4610 ts Seconds since epoch
4611 tu subsecond time (milliseconds)
4612 tl Local time. Optional strftime format argument
4613 default %d/%b/%Y:%H:%M:%S %z
4614 tg GMT time. Optional strftime format argument
4615 default %d/%b/%Y:%H:%M:%S %z
4616 tr Response time (milliseconds)
4617 dt Total time spent making DNS lookups (milliseconds)
4618 tS Approximate master transaction start time in
4619 <full seconds since epoch>.<fractional seconds> format.
4620 Currently, Squid considers the master transaction
4621 started when a complete HTTP request header initiating
4622 the transaction is received from the client. This is
4623 the same value that Squid uses to calculate transaction
4624 response time when logging %tr to access.log. Currently,
4625 Squid uses millisecond resolution for %tS values,
4626 similar to the default access.log "current time" field
4629 Access Control related format codes:
4631 et Tag returned by external acl
4632 ea Log string returned by external acl
4633 un User name (any available)
4634 ul User name from authentication
4635 ue User name from external acl helper
4636 ui User name from ident
4637 un A user name. Expands to the first available name
4638 from the following list of information sources:
4639 - authenticated user name, like %ul
4640 - user name supplied by an external ACL, like %ue
4641 - SSL client name, like %us
4642 - ident user name, like %ui
4643 credentials Client credentials. The exact meaning depends on
4644 the authentication scheme: For Basic authentication,
4645 it is the password; for Digest, the realm sent by the
4646 client; for NTLM and Negotiate, the client challenge
4647 or client credentials prefixed with "YR " or "KK ".
4649 HTTP related format codes:
4653 [http::]rm Request method (GET/POST etc)
4654 [http::]>rm Request method from client
4655 [http::]<rm Request method sent to server or peer
4657 [http::]ru Request URL received (or computed) and sanitized
4659 Logs request URI received from the client, a
4660 request adaptation service, or a request
4661 redirector (whichever was applied last).
4663 Computed URLs are URIs of internally generated
4664 requests and various "error:..." URIs.
4666 Honors strip_query_terms and uri_whitespace.
4668 This field is not encoded by default. Encoding
4669 this field using variants of %-encoding will
4670 clash with uri_whitespace modifications that
4671 also use %-encoding.
4673 [http::]>ru Request URL received from the client (or computed)
4675 Computed URLs are URIs of internally generated
4676 requests and various "error:..." URIs.
4678 Unlike %ru, this request URI is not affected
4679 by request adaptation, URL rewriting services,
4680 and strip_query_terms.
4682 Honors uri_whitespace.
4684 This field is using pass-through URL encoding
4685 by default. Encoding this field using other
4686 variants of %-encoding will clash with
4687 uri_whitespace modifications that also use
4690 [http::]<ru Request URL sent to server or peer
4691 [http::]>rs Request URL scheme from client
4692 [http::]<rs Request URL scheme sent to server or peer
4693 [http::]>rd Request URL domain from client
4694 [http::]<rd Request URL domain sent to server or peer
4695 [http::]>rP Request URL port from client
4696 [http::]<rP Request URL port sent to server or peer
4697 [http::]rp Request URL path excluding hostname
4698 [http::]>rp Request URL path excluding hostname from client
4699 [http::]<rp Request URL path excluding hostname sent to server or peer
4700 [http::]rv Request protocol version
4701 [http::]>rv Request protocol version from client
4702 [http::]<rv Request protocol version sent to server or peer
4704 [http::]>h Original received request header.
4705 Usually differs from the request header sent by
4706 Squid, although most fields are often preserved.
4707 Accepts optional header field name/value filter
4708 argument using name[:[separator]element] format.
4709 [http::]>ha Received request header after adaptation and
4710 redirection (pre-cache REQMOD vectoring point).
4711 Usually differs from the request header sent by
4712 Squid, although most fields are often preserved.
4713 Optional header name argument as for >h
4717 [http::]<Hs HTTP status code received from the next hop
4718 [http::]>Hs HTTP status code sent to the client
4720 [http::]<h Reply header. Optional header name argument
4723 [http::]mt MIME content type
4728 [http::]st Total size of request + reply traffic with client
4729 [http::]>st Total size of request received from client.
4730 Excluding chunked encoding bytes.
4731 [http::]<st Total size of reply sent to client (after adaptation)
4733 [http::]>sh Size of request headers received from client
4734 [http::]<sh Size of reply headers sent to client (after adaptation)
4736 [http::]<sH Reply high offset sent
4737 [http::]<sS Upstream object size
4739 [http::]<bs Number of HTTP-equivalent message body bytes
4740 received from the next hop, excluding chunked
4741 transfer encoding and control messages.
4742 Generated FTP/Gopher listings are treated as
4747 [http::]<pt Peer response time in milliseconds. The timer starts
4748 when the last request byte is sent to the next hop
4749 and stops when the last response byte is received.
4750 [http::]<tt Total time in milliseconds. The timer
4751 starts with the first connect request (or write I/O)
4752 sent to the first selected peer. The timer stops
4753 with the last I/O with the last peer.
4755 Squid handling related format codes:
4757 Ss Squid request status (TCP_MISS etc)
4758 Sh Squid hierarchy status (DEFAULT_PARENT etc)
4760 SSL-related format codes:
4762 ssl::bump_mode SslBump decision for the transaction:
4764 For CONNECT requests that initiated bumping of
4765 a connection and for any request received on
4766 an already bumped connection, Squid logs the
4767 corresponding SslBump mode ("splice", "bump",
4768 "peek", "stare", "terminate", "server-first"
4769 or "client-first"). See the ssl_bump option
4770 for more information about these modes.
4772 A "none" token is logged for requests that
4773 triggered "ssl_bump" ACL evaluation matching
4776 In all other cases, a single dash ("-") is
4779 ssl::>sni SSL client SNI sent to Squid.
4782 The Subject field of the received client
4783 SSL certificate or a dash ('-') if Squid has
4784 received an invalid/malformed certificate or
4785 no certificate at all. Consider encoding the
4786 logged value because Subject often has spaces.
4789 The Issuer field of the received client
4790 SSL certificate or a dash ('-') if Squid has
4791 received an invalid/malformed certificate or
4792 no certificate at all. Consider encoding the
4793 logged value because Issuer often has spaces.
4796 The Subject field of the received server
4797 TLS certificate or a dash ('-') if this is
4798 not available. Consider encoding the logged
4799 value because Subject often has spaces.
4802 The Issuer field of the received server
4803 TLS certificate or a dash ('-') if this is
4804 not available. Consider encoding the logged
4805 value because Issuer often has spaces.
4808 The received server x509 certificate in PEM
4809 format, including BEGIN and END lines (or a
4810 dash ('-') if the certificate is unavailable).
4812 WARNING: Large certificates will exceed the
4813 current 8KB access.log record limit, resulting
4814 in truncated records. Such truncation usually
4815 happens in the middle of a record field. The
4816 limit applies to all access logging modules.
4818 The logged certificate may have failed
4819 validation and may not be trusted by Squid.
4820 This field does not include any intermediate
4821 certificates that may have been received from
4822 the server or fetched during certificate
4825 Currently, Squid only collects server
4826 certificates during step3 of SslBump
4827 processing; connections that were not subject
4828 to ssl_bump rules or that did not match a peek
4829 or stare rule at step2 will not have the
4830 server certificate information.
4832 This field is using pass-through URL encoding
4836 The list of certificate validation errors
4837 detected by Squid (including OpenSSL and
4838 certificate validation helper components). The
4839 errors are listed in the discovery order. By
4840 default, the error codes are separated by ':'.
4841 Accepts an optional separator argument.
4843 %ssl::>negotiated_version The negotiated TLS version of the
4846 %ssl::<negotiated_version The negotiated TLS version of the
4847 last server or peer connection.
4849 %ssl::>received_hello_version The TLS version of the Hello
4850 message received from TLS client.
4852 %ssl::<received_hello_version The TLS version of the Hello
4853 message received from TLS server.
4855 %ssl::>received_supported_version The maximum TLS version
4856 supported by the TLS client.
4858 %ssl::<received_supported_version The maximum TLS version
4859 supported by the TLS server.
4861 %ssl::>negotiated_cipher The negotiated cipher of the
4864 %ssl::<negotiated_cipher The negotiated cipher of the
4865 last server or peer connection.
4867 If ICAP is enabled, the following code becomes available (as
4868 well as ICAP log codes documented with the icap_log option):
4870 icap::tt Total ICAP processing time for the HTTP
4871 transaction. The timer ticks when ICAP
4872 ACLs are checked and when ICAP
4873 transaction is in progress.
4875 If adaptation is enabled the following codes become available:
4877 adapt::<last_h The header of the last ICAP response or
4878 meta-information from the last eCAP
4879 transaction related to the HTTP transaction.
4880 Like <h, accepts an optional header name
4883 adapt::sum_trs Summed adaptation transaction response
4884 times recorded as a comma-separated list in
4885 the order of transaction start time. Each time
4886 value is recorded as an integer number,
4887 representing response time of one or more
4888 adaptation (ICAP or eCAP) transaction in
4889 milliseconds. When a failed transaction is
4890 being retried or repeated, its time is not
4891 logged individually but added to the
4892 replacement (next) transaction. See also:
4895 adapt::all_trs All adaptation transaction response times.
4896 Same as adaptation_strs but response times of
4897 individual transactions are never added
4898 together. Instead, all transaction response
4899 times are recorded individually.
4901 You can prefix adapt::*_trs format codes with adaptation
4902 service name in curly braces to record response time(s) specific
4903 to that service. For example: %{my_service}adapt::sum_trs
4905 Format codes related to the PROXY protocol:
4907 proxy_protocol::>h PROXY protocol header, including optional TLVs.
4909 Supports the same field and element reporting/extraction logic
4910 as %http::>h. For configuration and reporting purposes, Squid
4911 maps each PROXY TLV to an HTTP header field: the TLV type
4912 (configured as a decimal integer) is the field name, and the
4913 TLV value is the field value. All TLVs of "LOCAL" connections
4914 (in PROXY protocol terminology) are currently skipped/ignored.
4916 Squid also maps the following standard PROXY protocol header
4917 blocks to pseudo HTTP headers (their names use PROXY
4918 terminology and start with a colon, following HTTP tradition
4919 for pseudo headers): :command, :version, :src_addr, :dst_addr,
4920 :src_port, and :dst_port.
4922 Without optional parameters, this logformat code logs
4923 pseudo headers and TLVs.
4925 This format code uses pass-through URL encoding by default.
4928 # relay custom PROXY TLV #224 to adaptation services
4929 adaptation_meta Client-Foo "%proxy_protocol::>h{224}"
4933 The default formats available (which do not need re-defining) are:
4935 logformat squid %ts.%03tu %6tr %>a %Ss/%03>Hs %<st %rm %ru %[un %Sh/%<a %mt
4936 logformat common %>a %[ui %[un [%tl] "%rm %ru HTTP/%rv" %>Hs %<st %Ss:%Sh
4937 logformat combined %>a %[ui %[un [%tl] "%rm %ru HTTP/%rv" %>Hs %<st "%{Referer}>h" "%{User-Agent}>h" %Ss:%Sh
4938 logformat referrer %ts.%03tu %>a %{Referer}>h %ru
4939 logformat useragent %>a [%tl] "%{User-Agent}>h"
4941 NOTE: When the log_mime_hdrs directive is set to ON.
4942 The squid, common and combined formats have a safely encoded copy
4943 of the mime headers appended to each line within a pair of brackets.
4945 NOTE: The common and combined formats are not quite true to the Apache definition.
4946 The logs from Squid contain an extra status and hierarchy code appended.
4950 NAME: access_log cache_access_log
4952 LOC: Config.Log.accesslogs
4953 DEFAULT_IF_NONE: daemon:@DEFAULT_ACCESS_LOG@ squid
4955 Configures whether and how Squid logs HTTP and ICP transactions.
4956 If access logging is enabled, a single line is logged for every
4957 matching HTTP or ICP request. The recommended directive formats are:
4959 access_log <module>:<place> [option ...] [acl acl ...]
4960 access_log none [acl acl ...]
4962 The following directive format is accepted but may be deprecated:
4963 access_log <module>:<place> [<logformat name> [acl acl ...]]
4965 In most cases, the first ACL name must not contain the '=' character
4966 and should not be equal to an existing logformat name. You can always
4967 start with an 'all' ACL to work around those restrictions.
4969 Will log to the specified module:place using the specified format (which
4970 must be defined in a logformat directive) those entries which match
4971 ALL the acl's specified (which must be defined in acl clauses).
4972 If no acl is specified, all requests will be logged to this destination.
4974 ===== Available options for the recommended directive format =====
4976 logformat=name Names log line format (either built-in or
4977 defined by a logformat directive). Defaults
4980 buffer-size=64KB Defines approximate buffering limit for log
4981 records (see buffered_logs). Squid should not
4982 keep more than the specified size and, hence,
4983 should flush records before the buffer becomes
4984 full to avoid overflows under normal
4985 conditions (the exact flushing algorithm is
4986 module-dependent though). The on-error option
4987 controls overflow handling.
4989 on-error=die|drop Defines action on unrecoverable errors. The
4990 'drop' action ignores (i.e., does not log)
4991 affected log records. The default 'die' action
4992 kills the affected worker. The drop action
4993 support has not been tested for modules other
4996 rotate=N Specifies the number of log file rotations to
4997 make when you run 'squid -k rotate'. The default
4998 is to obey the logfile_rotate directive. Setting
4999 rotate=0 will disable the file name rotation,
5000 but the log files are still closed and re-opened.
5001 This will enable you to rename the logfiles
5002 yourself just before sending the rotate signal.
5003 Only supported by the stdio module.
5005 ===== Modules Currently available =====
5007 none Do not log any requests matching these ACL.
5008 Do not specify Place or logformat name.
5010 stdio Write each log line to disk immediately at the completion of
5012 Place: the filename and path to be written.
5014 daemon Very similar to stdio. But instead of writing to disk the log
5015 line is passed to a daemon helper for asychronous handling instead.
5016 Place: varies depending on the daemon.
5018 log_file_daemon Place: the file name and path to be written.
5020 syslog To log each request via syslog facility.
5021 Place: The syslog facility and priority level for these entries.
5022 Place Format: facility.priority
5024 where facility could be any of:
5025 authpriv, daemon, local0 ... local7 or user.
5027 And priority could be any of:
5028 err, warning, notice, info, debug.
5030 udp To send each log line as text data to a UDP receiver.
5031 Place: The destination host name or IP and port.
5032 Place Format: //host:port
5034 tcp To send each log line as text data to a TCP receiver.
5035 Lines may be accumulated before sending (see buffered_logs).
5036 Place: The destination host name or IP and port.
5037 Place Format: //host:port
5040 access_log daemon:@DEFAULT_ACCESS_LOG@ squid
5046 LOC: Config.Log.icaplogs
5049 ICAP log files record ICAP transaction summaries, one line per
5052 The icap_log option format is:
5053 icap_log <filepath> [<logformat name> [acl acl ...]]
5054 icap_log none [acl acl ...]]
5056 Please see access_log option documentation for details. The two
5057 kinds of logs share the overall configuration approach and many
5060 ICAP processing of a single HTTP message or transaction may
5061 require multiple ICAP transactions. In such cases, multiple
5062 ICAP transaction log lines will correspond to a single access
5065 ICAP log supports many access.log logformat %codes. In ICAP context,
5066 HTTP message-related %codes are applied to the HTTP message embedded
5067 in an ICAP message. Logformat "%http::>..." codes are used for HTTP
5068 messages embedded in ICAP requests while "%http::<..." codes are used
5069 for HTTP messages embedded in ICAP responses. For example:
5071 http::>h To-be-adapted HTTP message headers sent by Squid to
5072 the ICAP service. For REQMOD transactions, these are
5073 HTTP request headers. For RESPMOD, these are HTTP
5074 response headers, but Squid currently cannot log them
5075 (i.e., %http::>h will expand to "-" for RESPMOD).
5077 http::<h Adapted HTTP message headers sent by the ICAP
5078 service to Squid (i.e., HTTP request headers in regular
5079 REQMOD; HTTP response headers in RESPMOD and during
5080 request satisfaction in REQMOD).
5082 ICAP OPTIONS transactions do not embed HTTP messages.
5084 Several logformat codes below deal with ICAP message bodies. An ICAP
5085 message body, if any, typically includes a complete HTTP message
5086 (required HTTP headers plus optional HTTP message body). When
5087 computing HTTP message body size for these logformat codes, Squid
5088 either includes or excludes chunked encoding overheads; see
5089 code-specific documentation for details.
5091 For Secure ICAP services, all size-related information is currently
5092 computed before/after TLS encryption/decryption, as if TLS was not
5095 The following format codes are also available for ICAP logs:
5097 icap::<A ICAP server IP address. Similar to <A.
5099 icap::<service_name ICAP service name from the icap_service
5100 option in Squid configuration file.
5102 icap::ru ICAP Request-URI. Similar to ru.
5104 icap::rm ICAP request method (REQMOD, RESPMOD, or
5105 OPTIONS). Similar to existing rm.
5107 icap::>st The total size of the ICAP request sent to the ICAP
5108 server (ICAP headers + ICAP body), including chunking
5111 icap::<st The total size of the ICAP response received from the
5112 ICAP server (ICAP headers + ICAP body), including
5113 chunking metadata (if any).
5115 icap::<bs The size of the ICAP response body received from the
5116 ICAP server, excluding chunking metadata (if any).
5118 icap::tr Transaction response time (in
5119 milliseconds). The timer starts when
5120 the ICAP transaction is created and
5121 stops when the transaction is completed.
5124 icap::tio Transaction I/O time (in milliseconds). The
5125 timer starts when the first ICAP request
5126 byte is scheduled for sending. The timers
5127 stops when the last byte of the ICAP response
5130 icap::to Transaction outcome: ICAP_ERR* for all
5131 transaction errors, ICAP_OPT for OPTION
5132 transactions, ICAP_ECHO for 204
5133 responses, ICAP_MOD for message
5134 modification, and ICAP_SAT for request
5135 satisfaction. Similar to Ss.
5137 icap::Hs ICAP response status code. Similar to Hs.
5139 icap::>h ICAP request header(s). Similar to >h.
5141 icap::<h ICAP response header(s). Similar to <h.
5143 The default ICAP log format, which can be used without an explicit
5144 definition, is called icap_squid:
5146 logformat icap_squid %ts.%03tu %6icap::tr %>A %icap::to/%03icap::Hs %icap::<st %icap::rm %icap::ru %un -/%icap::<A -
5148 See also: logformat and %adapt::<last_h
5151 NAME: logfile_daemon
5153 DEFAULT: @DEFAULT_LOGFILED@
5154 LOC: Log::TheConfig.logfile_daemon
5156 Specify the path to the logfile-writing daemon. This daemon is
5157 used to write the access and store logs, if configured.
5159 Squid sends a number of commands to the log daemon:
5160 L<data>\n - logfile data
5165 r<n>\n - set rotate count to <n>
5166 b<n>\n - 1 = buffer output, 0 = don't buffer output
5168 No responses is expected.
5171 NAME: stats_collection
5173 LOC: Config.accessList.stats_collection
5175 DEFAULT_DOC: Allow logging for all transactions.
5176 COMMENT: allow|deny acl acl...
5178 This options allows you to control which requests gets accounted
5179 in performance counters.
5181 This clause only supports fast acl types.
5182 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
5185 NAME: cache_store_log
5188 LOC: Config.Log.store
5190 Logs the activities of the storage manager. Shows which
5191 objects are ejected from the cache, and which objects are
5192 saved and for how long.
5193 There are not really utilities to analyze this data, so you can safely
5194 disable it (the default).
5196 Store log uses modular logging outputs. See access_log for the list
5197 of modules supported.
5200 cache_store_log stdio:@DEFAULT_STORE_LOG@
5201 cache_store_log daemon:@DEFAULT_STORE_LOG@
5204 NAME: cache_swap_state cache_swap_log
5206 LOC: Config.Log.swap
5208 DEFAULT_DOC: Store the journal inside its cache_dir
5210 Location for the cache "swap.state" file. This index file holds
5211 the metadata of objects saved on disk. It is used to rebuild
5212 the cache during startup. Normally this file resides in each
5213 'cache_dir' directory, but you may specify an alternate
5214 pathname here. Note you must give a full filename, not just
5215 a directory. Since this is the index for the whole object
5216 list you CANNOT periodically rotate it!
5218 If %s can be used in the file name it will be replaced with a
5219 a representation of the cache_dir name where each / is replaced
5220 with '.'. This is needed to allow adding/removing cache_dir
5221 lines when cache_swap_log is being used.
5223 If have more than one 'cache_dir', and %s is not used in the name
5224 these swap logs will have names such as:
5230 The numbered extension (which is added automatically)
5231 corresponds to the order of the 'cache_dir' lines in this
5232 configuration file. If you change the order of the 'cache_dir'
5233 lines in this file, these index files will NOT correspond to
5234 the correct 'cache_dir' entry (unless you manually rename
5235 them). We recommend you do NOT use this option. It is
5236 better to keep these index files in each 'cache_dir' directory.
5239 NAME: logfile_rotate
5242 LOC: Config.Log.rotateNumber
5244 Specifies the default number of logfile rotations to make when you
5245 type 'squid -k rotate'. The default is 10, which will rotate
5246 with extensions 0 through 9. Setting logfile_rotate to 0 will
5247 disable the file name rotation, but the logfiles are still closed
5248 and re-opened. This will enable you to rename the logfiles
5249 yourself just before sending the rotate signal.
5251 Note, from Squid-3.1 this option is only a default for cache.log,
5252 that log can be rotated separately by using debug_options.
5254 Note, from Squid-4 this option is only a default for access.log
5255 recorded by stdio: module. Those logs can be rotated separately by
5256 using the rotate=N option on their access_log directive.
5258 Note, the 'squid -k rotate' command normally sends a USR1
5259 signal to the running squid process. In certain situations
5260 (e.g. on Linux with Async I/O), USR1 is used for other
5261 purposes, so -k rotate uses another signal. It is best to get
5262 in the habit of using 'squid -k rotate' instead of 'kill -USR1
5269 DEFAULT: @DEFAULT_MIME_TABLE@
5270 LOC: Config.mimeTablePathname
5272 Path to Squid's icon configuration file.
5274 You shouldn't need to change this, but the default file contains
5275 examples and formatting information if you do.
5281 LOC: Config.onoff.log_mime_hdrs
5284 The Cache can record both the request and the response MIME
5285 headers for each HTTP transaction. The headers are encoded
5286 safely and will appear as two bracketed fields at the end of
5287 the access log (for either the native or httpd-emulated log
5288 formats). To enable this logging set log_mime_hdrs to 'on'.
5293 DEFAULT: @DEFAULT_PID_FILE@
5294 LOC: Config.pidFilename
5296 A filename to write the process-id to. To disable, enter "none".
5299 NAME: client_netmask
5301 LOC: Config.Addrs.client_netmask
5303 DEFAULT_DOC: Log full client IP address
5305 A netmask for client addresses in logfiles and cachemgr output.
5306 Change this to protect the privacy of your cache clients.
5307 A netmask of 255.255.255.0 will log all IP's in that range with
5308 the last digit set to '0'.
5311 NAME: strip_query_terms
5313 LOC: Config.onoff.strip_query_terms
5316 By default, Squid strips query terms from requested URLs before
5317 logging. This protects your user's privacy and reduces log size.
5319 When investigating HIT/MISS or other caching behaviour you
5320 will need to disable this to see the full URL used by Squid.
5327 LOC: Config.onoff.buffered_logs
5329 Whether to write/send access_log records ASAP or accumulate them and
5330 then write/send them in larger chunks. Buffering may improve
5331 performance because it decreases the number of I/Os. However,
5332 buffering increases the delay before log records become available to
5333 the final recipient (e.g., a disk file or logging daemon) and,
5334 hence, increases the risk of log records loss.
5336 Note that even when buffered_logs are off, Squid may have to buffer
5337 records if it cannot write/send them immediately due to pending I/Os
5338 (e.g., the I/O writing the previous log record) or connectivity loss.
5340 Currently honored by 'daemon' and 'tcp' access_log modules only.
5343 NAME: netdb_filename
5345 DEFAULT: stdio:@DEFAULT_NETDB_FILE@
5346 LOC: Config.netdbFilename
5349 Where Squid stores it's netdb journal.
5350 When enabled this journal preserves netdb state between restarts.
5352 To disable, enter "none".
5356 OPTIONS FOR TROUBLESHOOTING
5357 -----------------------------------------------------------------------------
5362 DEFAULT_IF_NONE: @DEFAULT_CACHE_LOG@
5363 LOC: Debug::cache_log
5365 Squid administrative logging file.
5367 This is where general information about Squid behavior goes. You can
5368 increase the amount of data logged to this file and how often it is
5369 rotated with "debug_options"
5375 DEFAULT_DOC: Log all critical and important messages.
5376 LOC: Debug::debugOptions
5378 Logging options are set as section,level where each source file
5379 is assigned a unique section. Lower levels result in less
5380 output, Full debugging (level 9) can result in a very large
5381 log file, so be careful.
5383 The magic word "ALL" sets debugging levels for all sections.
5384 The default is to run with "ALL,1" to record important warnings.
5386 The rotate=N option can be used to keep more or less of these logs
5387 than would otherwise be kept by logfile_rotate.
5388 For most uses a single log should be enough to monitor current
5389 events affecting Squid.
5394 LOC: Config.coredump_dir
5395 DEFAULT_IF_NONE: none
5396 DEFAULT_DOC: Use the directory from where Squid was started.
5398 By default Squid leaves core files in the directory from where
5399 it was started. If you set 'coredump_dir' to a directory
5400 that exists, Squid will chdir() to that directory at startup
5401 and coredump files will be left there.
5405 # Leave coredumps in the first cache dir
5406 coredump_dir @DEFAULT_SWAP_DIR@
5412 OPTIONS FOR FTP GATEWAYING
5413 -----------------------------------------------------------------------------
5419 LOC: Config.Ftp.anon_user
5421 If you want the anonymous login password to be more informative
5422 (and enable the use of picky FTP servers), set this to something
5423 reasonable for your domain, like wwwuser@somewhere.net
5425 The reason why this is domainless by default is the
5426 request can be made on the behalf of a user in any domain,
5427 depending on how the cache is used.
5428 Some FTP server also validate the email address is valid
5429 (for example perl.com).
5435 LOC: Config.Ftp.passive
5437 If your firewall does not allow Squid to use passive
5438 connections, turn off this option.
5440 Use of ftp_epsv_all option requires this to be ON.
5446 LOC: Config.Ftp.epsv_all
5448 FTP Protocol extensions permit the use of a special "EPSV ALL" command.
5450 NATs may be able to put the connection on a "fast path" through the
5451 translator, as the EPRT command will never be used and therefore,
5452 translation of the data portion of the segments will never be needed.
5454 When a client only expects to do two-way FTP transfers this may be
5456 If squid finds that it must do a three-way FTP transfer after issuing
5457 an EPSV ALL command, the FTP session will fail.
5459 If you have any doubts about this option do not use it.
5460 Squid will nicely attempt all other connection methods.
5462 Requires ftp_passive to be ON (default) for any effect.
5468 LOC: Config.accessList.ftp_epsv
5470 FTP Protocol extensions permit the use of a special "EPSV" command.
5472 NATs may be able to put the connection on a "fast path" through the
5473 translator using EPSV, as the EPRT command will never be used
5474 and therefore, translation of the data portion of the segments
5475 will never be needed.
5477 EPSV is often required to interoperate with FTP servers on IPv6
5478 networks. On the other hand, it may break some IPv4 servers.
5480 By default, EPSV may try EPSV with any FTP server. To fine tune
5481 that decision, you may restrict EPSV to certain clients or servers
5484 ftp_epsv allow|deny al1 acl2 ...
5486 WARNING: Disabling EPSV may cause problems with external NAT and IPv6.
5488 Only fast ACLs are supported.
5489 Requires ftp_passive to be ON (default) for any effect.
5495 LOC: Config.Ftp.eprt
5497 FTP Protocol extensions permit the use of a special "EPRT" command.
5499 This extension provides a protocol neutral alternative to the
5500 IPv4-only PORT command. When supported it enables active FTP data
5501 channels over IPv6 and efficient NAT handling.
5503 Turning this OFF will prevent EPRT being attempted and will skip
5504 straight to using PORT for IPv4 servers.
5506 Some devices are known to not handle this extension correctly and
5507 may result in crashes. Devices which suport EPRT enough to fail
5508 cleanly will result in Squid attempting PORT anyway. This directive
5509 should only be disabled when EPRT results in device failures.
5511 WARNING: Doing so will convert Squid back to the old behavior with all
5512 the related problems with external NAT devices/layers and IPv4-only FTP.
5515 NAME: ftp_sanitycheck
5518 LOC: Config.Ftp.sanitycheck
5520 For security and data integrity reasons Squid by default performs
5521 sanity checks of the addresses of FTP data connections ensure the
5522 data connection is to the requested server. If you need to allow
5523 FTP connections to servers using another IP address for the data
5524 connection turn this off.
5527 NAME: ftp_telnet_protocol
5530 LOC: Config.Ftp.telnet
5532 The FTP protocol is officially defined to use the telnet protocol
5533 as transport channel for the control connection. However, many
5534 implementations are broken and does not respect this aspect of
5537 If you have trouble accessing files with ASCII code 255 in the
5538 path or similar problems involving this ASCII code you can
5539 try setting this directive to off. If that helps, report to the
5540 operator of the FTP server in question that their FTP server
5541 is broken and does not follow the FTP standard.
5545 OPTIONS FOR EXTERNAL SUPPORT PROGRAMS
5546 -----------------------------------------------------------------------------
5551 DEFAULT: @DEFAULT_DISKD@
5552 LOC: Config.Program.diskd
5554 Specify the location of the diskd executable.
5555 Note this is only useful if you have compiled in
5556 diskd as one of the store io modules.
5559 NAME: unlinkd_program
5562 DEFAULT: @DEFAULT_UNLINKD@
5563 LOC: Config.Program.unlinkd
5565 Specify the location of the executable for file deletion process.
5568 NAME: pinger_program
5571 DEFAULT: @DEFAULT_PINGER@
5574 Specify the location of the executable for the pinger process.
5583 Control whether the pinger is active at run-time.
5584 Enables turning ICMP pinger on and off with a simple
5585 squid -k reconfigure.
5590 OPTIONS FOR URL REWRITING
5591 -----------------------------------------------------------------------------
5594 NAME: url_rewrite_program redirect_program
5596 LOC: Config.Program.redirect
5599 Specify the location of the executable URL rewriter to use.
5600 Since they can perform almost any function there isn't one included.
5602 For each requested URL, the rewriter will receive on line with the format
5604 [channel-ID <SP>] URL [<SP> extras]<NL>
5606 See url_rewrite_extras on how to send "extras" with optional values to
5608 After processing the request the helper must reply using the following format:
5610 [channel-ID <SP>] result [<SP> kv-pairs]
5612 The result code can be:
5614 OK status=30N url="..."
5615 Redirect the URL to the one supplied in 'url='.
5616 'status=' is optional and contains the status code to send
5617 the client in Squids HTTP response. It must be one of the
5618 HTTP redirect status codes: 301, 302, 303, 307, 308.
5619 When no status is given Squid will use 302.
5621 OK rewrite-url="..."
5622 Rewrite the URL to the one supplied in 'rewrite-url='.
5623 The new URL is fetched directly by Squid and returned to
5624 the client as the response to its request.
5627 When neither of url= and rewrite-url= are sent Squid does
5631 Do not change the URL.
5634 An internal error occurred in the helper, preventing
5635 a result being identified. The 'message=' key name is
5636 reserved for delivering a log message.
5639 In addition to the above kv-pairs Squid also understands the following
5640 optional kv-pairs received from URL rewriters:
5642 Associates a TAG with the client TCP connection.
5643 The TAG is treated as a regular annotation but persists across
5644 future requests on the client connection rather than just the
5645 current request. A helper may update the TAG during subsequent
5646 requests be returning a new kv-pair.
5648 When using the concurrency= option the protocol is changed by
5649 introducing a query channel tag in front of the request/response.
5650 The query channel tag is a number between 0 and concurrency-1.
5651 This value must be echoed back unchanged to Squid as the first part
5652 of the response relating to its request.
5654 WARNING: URL re-writing ability should be avoided whenever possible.
5655 Use the URL redirect form of response instead.
5657 Re-write creates a difference in the state held by the client
5658 and server. Possibly causing confusion when the server response
5659 contains snippets of its view state. Embeded URLs, response
5660 and content Location headers, etc. are not re-written by this
5663 By default, a URL rewriter is not used.
5666 NAME: url_rewrite_children redirect_children
5667 TYPE: HelperChildConfig
5668 DEFAULT: 20 startup=0 idle=1 concurrency=0
5669 LOC: Config.redirectChildren
5671 Specifies the maximum number of redirector processes that Squid may
5672 spawn (numberofchildren) and several related options. Using too few of
5673 these helper processes (a.k.a. "helpers") creates request queues.
5674 Using too many helpers wastes your system resources.
5676 Usage: numberofchildren [option]...
5678 The startup= and idle= options allow some measure of skew in your
5683 Sets a minimum of how many processes are to be spawned when Squid
5684 starts or reconfigures. When set to zero the first request will
5685 cause spawning of the first child process to handle it.
5687 Starting too few will cause an initial slowdown in traffic as Squid
5688 attempts to simultaneously spawn enough processes to cope.
5692 Sets a minimum of how many processes Squid is to try and keep available
5693 at all times. When traffic begins to rise above what the existing
5694 processes can handle this many more will be spawned up to the maximum
5695 configured. A minimum setting of 1 is required.
5699 The number of requests each redirector helper can handle in
5700 parallel. Defaults to 0 which indicates the redirector
5701 is a old-style single threaded redirector.
5703 When this directive is set to a value >= 1 then the protocol
5704 used to communicate with the helper is modified to include
5705 an ID in front of the request/response. The ID from the request
5706 must be echoed back with the response to that request.
5710 Sets the maximum number of queued requests. A request is queued when
5711 no existing child can accept it due to concurrency limit and no new
5712 child can be started due to numberofchildren limit. The default
5713 maximum is zero if url_rewrite_bypass is enabled and
5714 2*numberofchildren otherwise. If the queued requests exceed queue size
5715 and redirector_bypass configuration option is set, then redirector is
5716 bypassed. Otherwise, Squid is allowed to temporarily exceed the
5717 configured maximum, marking the affected helper as "overloaded". If
5718 the helper overload lasts more than 3 minutes, the action prescribed
5719 by the on-persistent-overload option applies.
5721 on-persistent-overload=action
5723 Specifies Squid reaction to a new helper request arriving when the helper
5724 has been overloaded for more that 3 minutes already. The number of queued
5725 requests determines whether the helper is overloaded (see the queue-size
5728 Two actions are supported:
5730 die Squid worker quits. This is the default behavior.
5732 ERR Squid treats the helper request as if it was
5733 immediately submitted, and the helper immediately
5734 replied with an ERR response. This action has no effect
5735 on the already queued and in-progress helper requests.
5738 NAME: url_rewrite_host_header redirect_rewrites_host_header
5741 LOC: Config.onoff.redir_rewrites_host
5743 To preserve same-origin security policies in browsers and
5744 prevent Host: header forgery by redirectors Squid rewrites
5745 any Host: header in redirected requests.
5747 If you are running an accelerator this may not be a wanted
5748 effect of a redirector. This directive enables you disable
5749 Host: alteration in reverse-proxy traffic.
5751 WARNING: Entries are cached on the result of the URL rewriting
5752 process, so be careful if you have domain-virtual hosts.
5754 WARNING: Squid and other software verifies the URL and Host
5755 are matching, so be careful not to relay through other proxies
5756 or inspecting firewalls with this disabled.
5759 NAME: url_rewrite_access redirector_access
5762 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
5763 LOC: Config.accessList.redirector
5765 If defined, this access list specifies which requests are
5766 sent to the redirector processes.
5768 This clause supports both fast and slow acl types.
5769 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
5772 NAME: url_rewrite_bypass redirector_bypass
5774 LOC: Config.onoff.redirector_bypass
5777 When this is 'on', a request will not go through the
5778 redirector if all the helpers are busy. If this is 'off' and the
5779 redirector queue grows too large, the action is prescribed by the
5780 on-persistent-overload option. You should only enable this if the
5781 redirectors are not critical to your caching system. If you use
5782 redirectors for access control, and you enable this option,
5783 users may have access to pages they should not
5784 be allowed to request.
5786 Enabling this option sets the default url_rewrite_children queue-size
5790 NAME: url_rewrite_extras
5791 TYPE: TokenOrQuotedString
5792 LOC: Config.redirector_extras
5793 DEFAULT: "%>a/%>A %un %>rm myip=%la myport=%lp"
5795 Specifies a string to be append to request line format for the
5796 rewriter helper. "Quoted" format values may contain spaces and
5797 logformat %macros. In theory, any logformat %macro can be used.
5798 In practice, a %macro expands as a dash (-) if the helper request is
5799 sent before the required macro information is available to Squid.
5802 NAME: url_rewrite_timeout
5803 TYPE: UrlHelperTimeout
5804 LOC: Config.onUrlRewriteTimeout
5806 DEFAULT_DOC: Squid waits for the helper response forever
5808 Squid times active requests to redirector. The timeout value and Squid
5809 reaction to a timed out request are configurable using the following
5812 url_rewrite_timeout timeout time-units on_timeout=<action> [response=<quoted-response>]
5814 supported timeout actions:
5815 fail Squid return a ERR_GATEWAY_FAILURE error page
5817 bypass Do not re-write the URL
5819 retry Send the lookup to the helper again
5821 use_configured_response
5822 Use the <quoted-response> as helper response
5826 OPTIONS FOR STORE ID
5827 -----------------------------------------------------------------------------
5830 NAME: store_id_program storeurl_rewrite_program
5832 LOC: Config.Program.store_id
5835 Specify the location of the executable StoreID helper to use.
5836 Since they can perform almost any function there isn't one included.
5838 For each requested URL, the helper will receive one line with the format
5840 [channel-ID <SP>] URL [<SP> extras]<NL>
5843 After processing the request the helper must reply using the following format:
5845 [channel-ID <SP>] result [<SP> kv-pairs]
5847 The result code can be:
5850 Use the StoreID supplied in 'store-id='.
5853 The default is to use HTTP request URL as the store ID.
5856 An internal error occurred in the helper, preventing
5857 a result being identified.
5859 In addition to the above kv-pairs Squid also understands the following
5860 optional kv-pairs received from URL rewriters:
5862 Associates a TAG with the client TCP connection.
5863 Please see url_rewrite_program related documentation for this
5866 Helper programs should be prepared to receive and possibly ignore
5867 additional whitespace-separated tokens on each input line.
5869 When using the concurrency= option the protocol is changed by
5870 introducing a query channel tag in front of the request/response.
5871 The query channel tag is a number between 0 and concurrency-1.
5872 This value must be echoed back unchanged to Squid as the first part
5873 of the response relating to its request.
5875 NOTE: when using StoreID refresh_pattern will apply to the StoreID
5876 returned from the helper and not the URL.
5878 WARNING: Wrong StoreID value returned by a careless helper may result
5879 in the wrong cached response returned to the user.
5881 By default, a StoreID helper is not used.
5884 NAME: store_id_extras
5885 TYPE: TokenOrQuotedString
5886 LOC: Config.storeId_extras
5887 DEFAULT: "%>a/%>A %un %>rm myip=%la myport=%lp"
5889 Specifies a string to be append to request line format for the
5890 StoreId helper. "Quoted" format values may contain spaces and
5891 logformat %macros. In theory, any logformat %macro can be used.
5892 In practice, a %macro expands as a dash (-) if the helper request is
5893 sent before the required macro information is available to Squid.
5896 NAME: store_id_children storeurl_rewrite_children
5897 TYPE: HelperChildConfig
5898 DEFAULT: 20 startup=0 idle=1 concurrency=0
5899 LOC: Config.storeIdChildren
5901 Specifies the maximum number of StoreID helper processes that Squid
5902 may spawn (numberofchildren) and several related options. Using
5903 too few of these helper processes (a.k.a. "helpers") creates request
5904 queues. Using too many helpers wastes your system resources.
5906 Usage: numberofchildren [option]...
5908 The startup= and idle= options allow some measure of skew in your
5913 Sets a minimum of how many processes are to be spawned when Squid
5914 starts or reconfigures. When set to zero the first request will
5915 cause spawning of the first child process to handle it.
5917 Starting too few will cause an initial slowdown in traffic as Squid
5918 attempts to simultaneously spawn enough processes to cope.
5922 Sets a minimum of how many processes Squid is to try and keep available
5923 at all times. When traffic begins to rise above what the existing
5924 processes can handle this many more will be spawned up to the maximum
5925 configured. A minimum setting of 1 is required.
5929 The number of requests each storeID helper can handle in
5930 parallel. Defaults to 0 which indicates the helper
5931 is a old-style single threaded program.
5933 When this directive is set to a value >= 1 then the protocol
5934 used to communicate with the helper is modified to include
5935 an ID in front of the request/response. The ID from the request
5936 must be echoed back with the response to that request.
5940 Sets the maximum number of queued requests to N. A request is queued
5941 when no existing child can accept it due to concurrency limit and no
5942 new child can be started due to numberofchildren limit. The default
5943 maximum is 2*numberofchildren. If the queued requests exceed queue
5944 size and redirector_bypass configuration option is set, then
5945 redirector is bypassed. Otherwise, Squid is allowed to temporarily
5946 exceed the configured maximum, marking the affected helper as
5947 "overloaded". If the helper overload lasts more than 3 minutes, the
5948 action prescribed by the on-persistent-overload option applies.
5950 on-persistent-overload=action
5952 Specifies Squid reaction to a new helper request arriving when the helper
5953 has been overloaded for more that 3 minutes already. The number of queued
5954 requests determines whether the helper is overloaded (see the queue-size
5957 Two actions are supported:
5959 die Squid worker quits. This is the default behavior.
5961 ERR Squid treats the helper request as if it was
5962 immediately submitted, and the helper immediately
5963 replied with an ERR response. This action has no effect
5964 on the already queued and in-progress helper requests.
5967 NAME: store_id_access storeurl_rewrite_access
5970 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
5971 LOC: Config.accessList.store_id
5973 If defined, this access list specifies which requests are
5974 sent to the StoreID processes. By default all requests
5977 This clause supports both fast and slow acl types.
5978 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
5981 NAME: store_id_bypass storeurl_rewrite_bypass
5983 LOC: Config.onoff.store_id_bypass
5986 When this is 'on', a request will not go through the
5987 helper if all helpers are busy. If this is 'off' and the helper
5988 queue grows too large, the action is prescribed by the
5989 on-persistent-overload option. You should only enable this if the
5990 helpers are not critical to your caching system. If you use
5991 helpers for critical caching components, and you enable this
5992 option, users may not get objects from cache.
5993 This options sets default queue-size option of the store_id_children
5998 OPTIONS FOR TUNING THE CACHE
5999 -----------------------------------------------------------------------------
6002 NAME: cache no_cache
6005 DEFAULT_DOC: By default, this directive is unused and has no effect.
6006 LOC: Config.accessList.noCache
6008 Requests denied by this directive will not be served from the cache
6009 and their responses will not be stored in the cache. This directive
6010 has no effect on other transactions and on already cached responses.
6012 This clause supports both fast and slow acl types.
6013 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6015 This and the two other similar caching directives listed below are
6016 checked at different transaction processing stages, have different
6017 access to response information, affect different cache operations,
6018 and differ in slow ACLs support:
6020 * cache: Checked before Squid makes a hit/miss determination.
6021 No access to reply information!
6022 Denies both serving a hit and storing a miss.
6023 Supports both fast and slow ACLs.
6024 * send_hit: Checked after a hit was detected.
6025 Has access to reply (hit) information.
6026 Denies serving a hit only.
6027 Supports fast ACLs only.
6028 * store_miss: Checked before storing a cachable miss.
6029 Has access to reply (miss) information.
6030 Denies storing a miss only.
6031 Supports fast ACLs only.
6033 If you are not sure which of the three directives to use, apply the
6034 following decision logic:
6036 * If your ACL(s) are of slow type _and_ need response info, redesign.
6037 Squid does not support that particular combination at this time.
6039 * If your directive ACL(s) are of slow type, use "cache"; and/or
6040 * if your directive ACL(s) need no response info, use "cache".
6042 * If you do not want the response cached, use store_miss; and/or
6043 * if you do not want a hit on a cached response, use send_hit.
6049 DEFAULT_DOC: By default, this directive is unused and has no effect.
6050 LOC: Config.accessList.sendHit
6052 Responses denied by this directive will not be served from the cache
6053 (but may still be cached, see store_miss). This directive has no
6054 effect on the responses it allows and on the cached objects.
6056 Please see the "cache" directive for a summary of differences among
6057 store_miss, send_hit, and cache directives.
6059 Unlike the "cache" directive, send_hit only supports fast acl
6060 types. See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6064 # apply custom Store ID mapping to some URLs
6065 acl MapMe dstdomain .c.example.com
6066 store_id_program ...
6067 store_id_access allow MapMe
6069 # but prevent caching of special responses
6070 # such as 302 redirects that cause StoreID loops
6071 acl Ordinary http_status 200-299
6072 store_miss deny MapMe !Ordinary
6074 # and do not serve any previously stored special responses
6075 # from the cache (in case they were already cached before
6076 # the above store_miss rule was in effect).
6077 send_hit deny MapMe !Ordinary
6083 DEFAULT_DOC: By default, this directive is unused and has no effect.
6084 LOC: Config.accessList.storeMiss
6086 Responses denied by this directive will not be cached (but may still
6087 be served from the cache, see send_hit). This directive has no
6088 effect on the responses it allows and on the already cached responses.
6090 Please see the "cache" directive for a summary of differences among
6091 store_miss, send_hit, and cache directives. See the
6092 send_hit directive for a usage example.
6094 Unlike the "cache" directive, store_miss only supports fast acl
6095 types. See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6101 LOC: Config.maxStale
6104 This option puts an upper limit on how stale content Squid
6105 will serve from the cache if cache validation fails.
6106 Can be overriden by the refresh_pattern max-stale option.
6109 NAME: refresh_pattern
6110 TYPE: refreshpattern
6114 usage: refresh_pattern [-i] regex min percent max [options]
6116 By default, regular expressions are CASE-SENSITIVE. To make
6117 them case-insensitive, use the -i option.
6119 'Min' is the time (in minutes) an object without an explicit
6120 expiry time should be considered fresh. The recommended
6121 value is 0, any higher values may cause dynamic applications
6122 to be erroneously cached unless the application designer
6123 has taken the appropriate actions.
6125 'Percent' is a percentage of the objects age (time since last
6126 modification age) an object without explicit expiry time
6127 will be considered fresh.
6129 'Max' is an upper limit on how long objects without an explicit
6130 expiry time will be considered fresh. The value is also used
6131 to form Cache-Control: max-age header for a request sent from
6132 Squid to origin/parent.
6134 options: override-expire
6144 override-expire enforces min age even if the server
6145 sent an explicit expiry time (e.g., with the
6146 Expires: header or Cache-Control: max-age). Doing this
6147 VIOLATES the HTTP standard. Enabling this feature
6148 could make you liable for problems which it causes.
6150 Note: override-expire does not enforce staleness - it only extends
6151 freshness / min. If the server returns a Expires time which
6152 is longer than your max time, Squid will still consider
6153 the object fresh for that period of time.
6155 override-lastmod enforces min age even on objects
6156 that were modified recently.
6158 reload-into-ims changes a client no-cache or ``reload''
6159 request for a cached entry into a conditional request using
6160 If-Modified-Since and/or If-None-Match headers, provided the
6161 cached entry has a Last-Modified and/or a strong ETag header.
6162 Doing this VIOLATES the HTTP standard. Enabling this feature
6163 could make you liable for problems which it causes.
6165 ignore-reload ignores a client no-cache or ``reload''
6166 header. Doing this VIOLATES the HTTP standard. Enabling
6167 this feature could make you liable for problems which
6170 ignore-no-store ignores any ``Cache-control: no-store''
6171 headers received from a server. Doing this VIOLATES
6172 the HTTP standard. Enabling this feature could make you
6173 liable for problems which it causes.
6175 ignore-private ignores any ``Cache-control: private''
6176 headers received from a server. Doing this VIOLATES
6177 the HTTP standard. Enabling this feature could make you
6178 liable for problems which it causes.
6180 refresh-ims causes squid to contact the origin server
6181 when a client issues an If-Modified-Since request. This
6182 ensures that the client will receive an updated version
6183 if one is available.
6185 store-stale stores responses even if they don't have explicit
6186 freshness or a validator (i.e., Last-Modified or an ETag)
6187 present, or if they're already stale. By default, Squid will
6188 not cache such responses because they usually can't be
6189 reused. Note that such responses will be stale by default.
6191 max-stale=NN provide a maximum staleness factor. Squid won't
6192 serve objects more stale than this even if it failed to
6193 validate the object. Default: use the max_stale global limit.
6195 Basically a cached object is:
6197 FRESH if expire > now, else STALE
6199 FRESH if lm-factor < percent, else STALE
6203 The refresh_pattern lines are checked in the order listed here.
6204 The first entry which matches is used. If none of the entries
6205 match the default will be used.
6207 Note, you must uncomment all the default lines if you want
6208 to change one. The default setting is only active if none is
6214 # Add any of your own refresh_pattern entries above these.
6216 refresh_pattern ^ftp: 1440 20% 10080
6217 refresh_pattern ^gopher: 1440 0% 1440
6218 refresh_pattern -i (/cgi-bin/|\?) 0 0% 0
6219 refresh_pattern . 0 20% 4320
6223 NAME: quick_abort_min
6227 LOC: Config.quickAbort.min
6230 NAME: quick_abort_max
6234 LOC: Config.quickAbort.max
6237 NAME: quick_abort_pct
6241 LOC: Config.quickAbort.pct
6243 The cache by default continues downloading aborted requests
6244 which are almost completed (less than 16 KB remaining). This
6245 may be undesirable on slow (e.g. SLIP) links and/or very busy
6246 caches. Impatient users may tie up file descriptors and
6247 bandwidth by repeatedly requesting and immediately aborting
6250 When the user aborts a request, Squid will check the
6251 quick_abort values to the amount of data transferred until
6254 If the transfer has less than 'quick_abort_min' KB remaining,
6255 it will finish the retrieval.
6257 If the transfer has more than 'quick_abort_max' KB remaining,
6258 it will abort the retrieval.
6260 If more than 'quick_abort_pct' of the transfer has completed,
6261 it will finish the retrieval.
6263 If you do not want any retrieval to continue after the client
6264 has aborted, set both 'quick_abort_min' and 'quick_abort_max'
6267 If you want retrievals to always continue if they are being
6268 cached set 'quick_abort_min' to '-1 KB'.
6271 NAME: read_ahead_gap
6272 COMMENT: buffer-size
6274 LOC: Config.readAheadGap
6277 The amount of data the cache will buffer ahead of what has been
6278 sent to the client when retrieving an object from another server.
6282 IFDEF: USE_HTTP_VIOLATIONS
6285 LOC: Config.negativeTtl
6288 Set the Default Time-to-Live (TTL) for failed requests.
6289 Certain types of failures (such as "connection refused" and
6290 "404 Not Found") are able to be negatively-cached for a short time.
6291 Modern web servers should provide Expires: header, however if they
6292 do not this can provide a minimum TTL.
6293 The default is not to cache errors with unknown expiry details.
6295 Note that this is different from negative caching of DNS lookups.
6297 WARNING: Doing this VIOLATES the HTTP standard. Enabling
6298 this feature could make you liable for problems which it
6302 NAME: positive_dns_ttl
6305 LOC: Config.positiveDnsTtl
6308 Upper limit on how long Squid will cache positive DNS responses.
6309 Default is 6 hours (360 minutes). This directive must be set
6310 larger than negative_dns_ttl.
6313 NAME: negative_dns_ttl
6316 LOC: Config.negativeDnsTtl
6319 Time-to-Live (TTL) for negative caching of failed DNS lookups.
6320 This also sets the lower cache limit on positive lookups.
6321 Minimum value is 1 second, and it is not recommendable to go
6322 much below 10 seconds.
6325 NAME: range_offset_limit
6326 COMMENT: size [acl acl...]
6328 LOC: Config.rangeOffsetLimit
6331 usage: (size) [units] [[!]aclname]
6333 Sets an upper limit on how far (number of bytes) into the file
6334 a Range request may be to cause Squid to prefetch the whole file.
6335 If beyond this limit, Squid forwards the Range request as it is and
6336 the result is NOT cached.
6338 This is to stop a far ahead range request (lets say start at 17MB)
6339 from making Squid fetch the whole object up to that point before
6340 sending anything to the client.
6342 Multiple range_offset_limit lines may be specified, and they will
6343 be searched from top to bottom on each request until a match is found.
6344 The first match found will be used. If no line matches a request, the
6345 default limit of 0 bytes will be used.
6347 'size' is the limit specified as a number of units.
6349 'units' specifies whether to use bytes, KB, MB, etc.
6350 If no units are specified bytes are assumed.
6352 A size of 0 causes Squid to never fetch more than the
6353 client requested. (default)
6355 A size of 'none' causes Squid to always fetch the object from the
6356 beginning so it may cache the result. (2.0 style)
6358 'aclname' is the name of a defined ACL.
6360 NP: Using 'none' as the byte value here will override any quick_abort settings
6361 that may otherwise apply to the range request. The range request will
6362 be fully fetched from start to finish regardless of the client
6363 actions. This affects bandwidth usage.
6366 NAME: minimum_expiry_time
6369 LOC: Config.minimum_expiry_time
6372 The minimum caching time according to (Expires - Date)
6373 headers Squid honors if the object can't be revalidated.
6374 The default is 60 seconds.
6376 In reverse proxy environments it might be desirable to honor
6377 shorter object lifetimes. It is most likely better to make
6378 your server return a meaningful Last-Modified header however.
6380 In ESI environments where page fragments often have short
6381 lifetimes, this will often be best set to 0.
6384 NAME: store_avg_object_size
6388 LOC: Config.Store.avgObjectSize
6390 Average object size, used to estimate number of objects your
6391 cache can hold. The default is 13 KB.
6393 This is used to pre-seed the cache index memory allocation to
6394 reduce expensive reallocate operations while handling clients
6395 traffic. Too-large values may result in memory allocation during
6396 peak traffic, too-small values will result in wasted memory.
6398 Check the cache manager 'info' report metrics for the real
6399 object sizes seen by your Squid before tuning this.
6402 NAME: store_objects_per_bucket
6405 LOC: Config.Store.objectsPerBucket
6407 Target number of objects per bucket in the store hash table.
6408 Lowering this value increases the total number of buckets and
6409 also the storage maintenance rate. The default is 20.
6414 -----------------------------------------------------------------------------
6417 NAME: request_header_max_size
6421 LOC: Config.maxRequestHeaderSize
6423 This specifies the maximum size for HTTP headers in a request.
6424 Request headers are usually relatively small (about 512 bytes).
6425 Placing a limit on the request header size will catch certain
6426 bugs (for example with persistent connections) and possibly
6427 buffer-overflow or denial-of-service attacks.
6430 NAME: reply_header_max_size
6434 LOC: Config.maxReplyHeaderSize
6436 This specifies the maximum size for HTTP headers in a reply.
6437 Reply headers are usually relatively small (about 512 bytes).
6438 Placing a limit on the reply header size will catch certain
6439 bugs (for example with persistent connections) and possibly
6440 buffer-overflow or denial-of-service attacks.
6443 NAME: request_body_max_size
6447 DEFAULT_DOC: No limit.
6448 LOC: Config.maxRequestBodySize
6450 This specifies the maximum size for an HTTP request body.
6451 In other words, the maximum size of a PUT/POST request.
6452 A user who attempts to send a request with a body larger
6453 than this limit receives an "Invalid Request" error message.
6454 If you set this parameter to a zero (the default), there will
6455 be no limit imposed.
6457 See also client_request_buffer_max_size for an alternative
6458 limitation on client uploads which can be configured.
6461 NAME: client_request_buffer_max_size
6465 LOC: Config.maxRequestBufferSize
6467 This specifies the maximum buffer size of a client request.
6468 It prevents squid eating too much memory when somebody uploads
6473 IFDEF: USE_HTTP_VIOLATIONS
6476 DEFAULT_DOC: Obey RFC 2616.
6477 LOC: Config.accessList.brokenPosts
6479 A list of ACL elements which, if matched, causes Squid to send
6480 an extra CRLF pair after the body of a PUT/POST request.
6482 Some HTTP servers has broken implementations of PUT/POST,
6483 and rely on an extra CRLF pair sent by some WWW clients.
6485 Quote from RFC2616 section 4.1 on this matter:
6487 Note: certain buggy HTTP/1.0 client implementations generate an
6488 extra CRLF's after a POST request. To restate what is explicitly
6489 forbidden by the BNF, an HTTP/1.1 client must not preface or follow
6490 a request with an extra CRLF.
6492 This clause only supports fast acl types.
6493 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6496 acl buggy_server url_regex ^http://....
6497 broken_posts allow buggy_server
6500 NAME: adaptation_uses_indirect_client icap_uses_indirect_client
6503 IFDEF: FOLLOW_X_FORWARDED_FOR&&USE_ADAPTATION
6505 LOC: Adaptation::Config::use_indirect_client
6507 Controls whether the indirect client IP address (instead of the direct
6508 client IP address) is passed to adaptation services.
6510 See also: follow_x_forwarded_for adaptation_send_client_ip
6514 IFDEF: USE_HTTP_VIOLATIONS
6518 LOC: Config.onoff.via
6520 If set (default), Squid will include a Via header in requests and
6521 replies as required by RFC2616.
6524 NAME: vary_ignore_expire
6527 LOC: Config.onoff.vary_ignore_expire
6530 Many HTTP servers supporting Vary gives such objects
6531 immediate expiry time with no cache-control header
6532 when requested by a HTTP/1.0 client. This option
6533 enables Squid to ignore such expiry times until
6534 HTTP/1.1 is fully implemented.
6536 WARNING: If turned on this may eventually cause some
6537 varying objects not intended for caching to get cached.
6540 NAME: request_entities
6542 LOC: Config.onoff.request_entities
6545 Squid defaults to deny GET and HEAD requests with request entities,
6546 as the meaning of such requests are undefined in the HTTP standard
6547 even if not explicitly forbidden.
6549 Set this directive to on if you have clients which insists
6550 on sending request entities in GET or HEAD requests. But be warned
6551 that there is server software (both proxies and web servers) which
6552 can fail to properly process this kind of request which may make you
6553 vulnerable to cache pollution attacks if enabled.
6556 NAME: request_header_access
6557 IFDEF: USE_HTTP_VIOLATIONS
6558 TYPE: http_header_access
6559 LOC: Config.request_header_access
6561 DEFAULT_DOC: No limits.
6563 Usage: request_header_access header_name allow|deny [!]aclname ...
6565 WARNING: Doing this VIOLATES the HTTP standard. Enabling
6566 this feature could make you liable for problems which it
6569 This option replaces the old 'anonymize_headers' and the
6570 older 'http_anonymizer' option with something that is much
6571 more configurable. A list of ACLs for each header name allows
6572 removal of specific header fields under specific conditions.
6574 This option only applies to outgoing HTTP request headers (i.e.,
6575 headers sent by Squid to the next HTTP hop such as a cache peer
6576 or an origin server). The option has no effect during cache hit
6577 detection. The equivalent adaptation vectoring point in ICAP
6578 terminology is post-cache REQMOD.
6580 The option is applied to individual outgoing request header
6581 fields. For each request header field F, Squid uses the first
6582 qualifying sets of request_header_access rules:
6584 1. Rules with header_name equal to F's name.
6585 2. Rules with header_name 'Other', provided F's name is not
6586 on the hard-coded list of commonly used HTTP header names.
6587 3. Rules with header_name 'All'.
6589 Within that qualifying rule set, rule ACLs are checked as usual.
6590 If ACLs of an "allow" rule match, the header field is allowed to
6591 go through as is. If ACLs of a "deny" rule match, the header is
6592 removed and request_header_replace is then checked to identify
6593 if the removed header has a replacement. If no rules within the
6594 set have matching ACLs, the header field is left as is.
6596 For example, to achieve the same behavior as the old
6597 'http_anonymizer standard' option, you should use:
6599 request_header_access From deny all
6600 request_header_access Referer deny all
6601 request_header_access User-Agent deny all
6603 Or, to reproduce the old 'http_anonymizer paranoid' feature
6606 request_header_access Authorization allow all
6607 request_header_access Proxy-Authorization allow all
6608 request_header_access Cache-Control allow all
6609 request_header_access Content-Length allow all
6610 request_header_access Content-Type allow all
6611 request_header_access Date allow all
6612 request_header_access Host allow all
6613 request_header_access If-Modified-Since allow all
6614 request_header_access Pragma allow all
6615 request_header_access Accept allow all
6616 request_header_access Accept-Charset allow all
6617 request_header_access Accept-Encoding allow all
6618 request_header_access Accept-Language allow all
6619 request_header_access Connection allow all
6620 request_header_access All deny all
6622 HTTP reply headers are controlled with the reply_header_access directive.
6624 By default, all headers are allowed (no anonymizing is performed).
6627 NAME: reply_header_access
6628 IFDEF: USE_HTTP_VIOLATIONS
6629 TYPE: http_header_access
6630 LOC: Config.reply_header_access
6632 DEFAULT_DOC: No limits.
6634 Usage: reply_header_access header_name allow|deny [!]aclname ...
6636 WARNING: Doing this VIOLATES the HTTP standard. Enabling
6637 this feature could make you liable for problems which it
6640 This option only applies to reply headers, i.e., from the
6641 server to the client.
6643 This is the same as request_header_access, but in the other
6644 direction. Please see request_header_access for detailed
6647 For example, to achieve the same behavior as the old
6648 'http_anonymizer standard' option, you should use:
6650 reply_header_access Server deny all
6651 reply_header_access WWW-Authenticate deny all
6652 reply_header_access Link deny all
6654 Or, to reproduce the old 'http_anonymizer paranoid' feature
6657 reply_header_access Allow allow all
6658 reply_header_access WWW-Authenticate allow all
6659 reply_header_access Proxy-Authenticate allow all
6660 reply_header_access Cache-Control allow all
6661 reply_header_access Content-Encoding allow all
6662 reply_header_access Content-Length allow all
6663 reply_header_access Content-Type allow all
6664 reply_header_access Date allow all
6665 reply_header_access Expires allow all
6666 reply_header_access Last-Modified allow all
6667 reply_header_access Location allow all
6668 reply_header_access Pragma allow all
6669 reply_header_access Content-Language allow all
6670 reply_header_access Retry-After allow all
6671 reply_header_access Title allow all
6672 reply_header_access Content-Disposition allow all
6673 reply_header_access Connection allow all
6674 reply_header_access All deny all
6676 HTTP request headers are controlled with the request_header_access directive.
6678 By default, all headers are allowed (no anonymizing is
6682 NAME: request_header_replace header_replace
6683 IFDEF: USE_HTTP_VIOLATIONS
6684 TYPE: http_header_replace
6685 LOC: Config.request_header_access
6688 Usage: request_header_replace header_name message
6689 Example: request_header_replace User-Agent Nutscrape/1.0 (CP/M; 8-bit)
6691 This option allows you to change the contents of headers
6692 denied with request_header_access above, by replacing them
6693 with some fixed string.
6695 This only applies to request headers, not reply headers.
6697 By default, headers are removed if denied.
6700 NAME: reply_header_replace
6701 IFDEF: USE_HTTP_VIOLATIONS
6702 TYPE: http_header_replace
6703 LOC: Config.reply_header_access
6706 Usage: reply_header_replace header_name message
6707 Example: reply_header_replace Server Foo/1.0
6709 This option allows you to change the contents of headers
6710 denied with reply_header_access above, by replacing them
6711 with some fixed string.
6713 This only applies to reply headers, not request headers.
6715 By default, headers are removed if denied.
6718 NAME: request_header_add
6719 TYPE: HeaderWithAclList
6720 LOC: Config.request_header_add
6723 Usage: request_header_add field-name field-value [ acl ... ]
6724 Example: request_header_add X-Client-CA "CA=%ssl::>cert_issuer" all
6726 This option adds header fields to outgoing HTTP requests (i.e.,
6727 request headers sent by Squid to the next HTTP hop such as a
6728 cache peer or an origin server). The option has no effect during
6729 cache hit detection. The equivalent adaptation vectoring point
6730 in ICAP terminology is post-cache REQMOD.
6732 Field-name is a token specifying an HTTP header name. If a
6733 standard HTTP header name is used, Squid does not check whether
6734 the new header conflicts with any existing headers or violates
6735 HTTP rules. If the request to be modified already contains a
6736 field with the same name, the old field is preserved but the
6737 header field values are not merged.
6739 Field-value is either a token or a quoted string. If quoted
6740 string format is used, then the surrounding quotes are removed
6741 while escape sequences and %macros are processed.
6743 One or more Squid ACLs may be specified to restrict header
6744 injection to matching requests. As always in squid.conf, all
6745 ACLs in the ACL list must be satisfied for the insertion to
6746 happen. The request_header_add supports fast ACLs only.
6748 See also: reply_header_add.
6751 NAME: reply_header_add
6752 TYPE: HeaderWithAclList
6753 LOC: Config.reply_header_add
6756 Usage: reply_header_add field-name field-value [ acl ... ]
6757 Example: reply_header_add X-Client-CA "CA=%ssl::>cert_issuer" all
6759 This option adds header fields to outgoing HTTP responses (i.e., response
6760 headers delivered by Squid to the client). This option has no effect on
6761 cache hit detection. The equivalent adaptation vectoring point in
6762 ICAP terminology is post-cache RESPMOD. This option does not apply to
6763 successful CONNECT replies.
6765 Field-name is a token specifying an HTTP header name. If a
6766 standard HTTP header name is used, Squid does not check whether
6767 the new header conflicts with any existing headers or violates
6768 HTTP rules. If the response to be modified already contains a
6769 field with the same name, the old field is preserved but the
6770 header field values are not merged.
6772 Field-value is either a token or a quoted string. If quoted
6773 string format is used, then the surrounding quotes are removed
6774 while escape sequences and %macros are processed.
6776 One or more Squid ACLs may be specified to restrict header
6777 injection to matching responses. As always in squid.conf, all
6778 ACLs in the ACL list must be satisfied for the insertion to
6779 happen. The reply_header_add option supports fast ACLs only.
6781 See also: request_header_add.
6789 This option used to log custom information about the master
6790 transaction. For example, an admin may configure Squid to log
6791 which "user group" the transaction belongs to, where "user group"
6792 will be determined based on a set of ACLs and not [just]
6793 authentication information.
6794 Values of key/value pairs can be logged using %{key}note macros:
6796 note key value acl ...
6797 logformat myFormat ... %{key}note ...
6799 This clause only supports fast acl types.
6800 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6803 NAME: relaxed_header_parser
6804 COMMENT: on|off|warn
6806 LOC: Config.onoff.relaxed_header_parser
6809 In the default "on" setting Squid accepts certain forms
6810 of non-compliant HTTP messages where it is unambiguous
6811 what the sending application intended even if the message
6812 is not correctly formatted. The messages is then normalized
6813 to the correct form when forwarded by Squid.
6815 If set to "warn" then a warning will be emitted in cache.log
6816 each time such HTTP error is encountered.
6818 If set to "off" then such HTTP errors will cause the request
6819 or response to be rejected.
6822 NAME: collapsed_forwarding
6825 LOC: Config.onoff.collapsed_forwarding
6828 This option controls whether Squid is allowed to merge multiple
6829 potentially cachable requests for the same URI before Squid knows
6830 whether the response is going to be cachable.
6832 When enabled, instead of forwarding each concurrent request for
6833 the same URL, Squid just sends the first of them. The other, so
6834 called "collapsed" requests, wait for the response to the first
6835 request and, if it happens to be cachable, use that response.
6836 Here, "concurrent requests" means "received after the first
6837 request headers were parsed and before the corresponding response
6838 headers were parsed".
6840 This feature is disabled by default: enabling collapsed
6841 forwarding needlessly delays forwarding requests that look
6842 cachable (when they are collapsed) but then need to be forwarded
6843 individually anyway because they end up being for uncachable
6844 content. However, in some cases, such as acceleration of highly
6845 cachable content with periodic or grouped expiration times, the
6846 gains from collapsing [large volumes of simultaneous refresh
6847 requests] outweigh losses from such delays.
6849 Squid collapses two kinds of requests: regular client requests
6850 received on one of the listening ports and internal "cache
6851 revalidation" requests which are triggered by those regular
6852 requests hitting a stale cached object. Revalidation collapsing
6853 is currently disabled for Squid instances containing SMP-aware
6854 disk or memory caches and for Vary-controlled cached objects.
6857 NAME: collapsed_forwarding_access
6860 DEFAULT_DOC: Requests may be collapsed if collapsed_forwarding is on.
6861 LOC: Config.accessList.collapsedForwardingAccess
6863 Use this directive to restrict collapsed forwarding to a subset of
6864 eligible requests. The directive is checked for regular HTTP
6865 requests, internal revalidation requests, and HTCP/ICP requests.
6867 collapsed_forwarding_access allow|deny [!]aclname ...
6869 This directive cannot force collapsing. It has no effect on
6870 collapsing unless collapsed_forwarding is 'on', and all other
6871 collapsing preconditions are satisfied.
6873 * A denied request will not collapse, and future transactions will
6874 not collapse on it (even if they are allowed to collapse).
6876 * An allowed request may collapse, or future transactions may
6877 collapse on it (provided they are allowed to collapse).
6879 This directive is evaluated before receiving HTTP response headers
6880 and without access to Squid-to-peer connection (if any).
6882 Only fast ACLs are supported.
6884 See also: collapsed_forwarding.
6887 NAME: shared_transient_entries_limit collapsed_forwarding_shared_entries_limit
6888 COMMENT: (number of entries)
6890 LOC: Config.shared_transient_entries_limit
6893 This directive limits the size of a table used for sharing current
6894 transaction information among SMP workers. A table entry stores meta
6895 information about a single cache entry being delivered to Squid
6896 client(s) by one or more SMP workers. A single table entry consumes
6897 less than 128 shared memory bytes.
6899 The limit should be significantly larger than the number of
6900 concurrent non-collapsed cachable responses leaving Squid. For a
6901 cache that handles less than 5000 concurrent requests, the default
6902 setting of 16384 should be plenty.
6904 Using excessively large values wastes shared memory. Limiting the
6905 table size too much results in hash collisions, leading to lower hit
6906 ratio and missed SMP request collapsing opportunities: Transactions
6907 left without a table entry cannot cache their responses and are
6908 invisible to other concurrent requests for the same resource.
6910 A zero limit is allowed but unsupported. A positive small limit
6911 lowers hit ratio, but zero limit disables a lot of essential
6912 synchronization among SMP workers, leading to HTTP violations (e.g.,
6913 stale hit responses). It also disables shared collapsed forwarding:
6914 A worker becomes unable to collapse its requests on transactions in
6915 other workers, resulting in more trips to the origin server and more
6921 -----------------------------------------------------------------------------
6924 NAME: forward_timeout
6927 LOC: Config.Timeout.forward
6930 This parameter specifies how long Squid should at most attempt in
6931 finding a forwarding path for the request before giving up.
6934 NAME: connect_timeout
6937 LOC: Config.Timeout.connect
6940 This parameter specifies how long to wait for the TCP connect to
6941 the requested server or peer to complete before Squid should
6942 attempt to find another path where to forward the request.
6945 NAME: peer_connect_timeout
6948 LOC: Config.Timeout.peer_connect
6951 This parameter specifies how long to wait for a pending TCP
6952 connection to a peer cache. The default is 30 seconds. You
6953 may also set different timeout values for individual neighbors
6954 with the 'connect-timeout' option on a 'cache_peer' line.
6960 LOC: Config.Timeout.read
6963 Applied on peer server connections.
6965 After each successful read(), the timeout will be extended by this
6966 amount. If no data is read again after this amount of time,
6967 the request is aborted and logged with ERR_READ_TIMEOUT.
6969 The default is 15 minutes.
6975 LOC: Config.Timeout.write
6978 This timeout is tracked for all connections that have data
6979 available for writing and are waiting for the socket to become
6980 ready. After each successful write, the timeout is extended by
6981 the configured amount. If Squid has data to write but the
6982 connection is not ready for the configured duration, the
6983 transaction associated with the connection is terminated. The
6984 default is 15 minutes.
6987 NAME: request_timeout
6989 LOC: Config.Timeout.request
6992 How long to wait for complete HTTP request headers after initial
6993 connection establishment.
6996 NAME: request_start_timeout
6998 LOC: Config.Timeout.request_start_timeout
7001 How long to wait for the first request byte after initial
7002 connection establishment.
7005 NAME: client_idle_pconn_timeout persistent_request_timeout
7007 LOC: Config.Timeout.clientIdlePconn
7010 How long to wait for the next HTTP request on a persistent
7011 client connection after the previous request completes.
7014 NAME: ftp_client_idle_timeout
7016 LOC: Config.Timeout.ftpClientIdle
7019 How long to wait for an FTP request on a connection to Squid ftp_port.
7020 Many FTP clients do not deal with idle connection closures well,
7021 necessitating a longer default timeout than client_idle_pconn_timeout
7022 used for incoming HTTP requests.
7025 NAME: client_lifetime
7028 LOC: Config.Timeout.lifetime
7031 The maximum amount of time a client (browser) is allowed to
7032 remain connected to the cache process. This protects the Cache
7033 from having a lot of sockets (and hence file descriptors) tied up
7034 in a CLOSE_WAIT state from remote clients that go away without
7035 properly shutting down (either because of a network failure or
7036 because of a poor client implementation). The default is one
7039 NOTE: The default value is intended to be much larger than any
7040 client would ever need to be connected to your cache. You
7041 should probably change client_lifetime only as a last resort.
7042 If you seem to have many client connections tying up
7043 filedescriptors, we recommend first tuning the read_timeout,
7044 request_timeout, persistent_request_timeout and quick_abort values.
7047 NAME: pconn_lifetime
7050 LOC: Config.Timeout.pconnLifetime
7053 Desired maximum lifetime of a persistent connection.
7054 When set, Squid will close a now-idle persistent connection that
7055 exceeded configured lifetime instead of moving the connection into
7056 the idle connection pool (or equivalent). No effect on ongoing/active
7057 transactions. Connection lifetime is the time period from the
7058 connection acceptance or opening time until "now".
7060 This limit is useful in environments with long-lived connections
7061 where Squid configuration or environmental factors change during a
7062 single connection lifetime. If unrestricted, some connections may
7063 last for hours and even days, ignoring those changes that should
7064 have affected their behavior or their existence.
7066 Currently, a new lifetime value supplied via Squid reconfiguration
7067 has no effect on already idle connections unless they become busy.
7069 When set to '0' this limit is not used.
7072 NAME: half_closed_clients
7074 LOC: Config.onoff.half_closed_clients
7077 Some clients may shutdown the sending side of their TCP
7078 connections, while leaving their receiving sides open. Sometimes,
7079 Squid can not tell the difference between a half-closed and a
7080 fully-closed TCP connection.
7082 By default, Squid will immediately close client connections when
7083 read(2) returns "no more data to read."
7085 Change this option to 'on' and Squid will keep open connections
7086 until a read(2) or write(2) on the socket returns an error.
7087 This may show some benefits for reverse proxies. But if not
7088 it is recommended to leave OFF.
7091 NAME: server_idle_pconn_timeout pconn_timeout
7093 LOC: Config.Timeout.serverIdlePconn
7096 Timeout for idle persistent connections to servers and other
7103 LOC: Ident::TheConfig.timeout
7106 Maximum time to wait for IDENT lookups to complete.
7108 If this is too high, and you enabled IDENT lookups from untrusted
7109 users, you might be susceptible to denial-of-service by having
7110 many ident requests going at once.
7113 NAME: shutdown_lifetime
7116 LOC: Config.shutdownLifetime
7119 When SIGTERM or SIGHUP is received, the cache is put into
7120 "shutdown pending" mode until all active sockets are closed.
7121 This value is the lifetime to set for all open descriptors
7122 during shutdown mode. Any active clients after this many
7123 seconds will receive a 'timeout' message.
7127 ADMINISTRATIVE PARAMETERS
7128 -----------------------------------------------------------------------------
7134 LOC: Config.adminEmail
7136 Email-address of local cache manager who will receive
7137 mail if the cache dies. The default is "webmaster".
7143 LOC: Config.EmailFrom
7145 From: email-address for mail sent when the cache dies.
7146 The default is to use 'squid@unique_hostname'.
7148 See also: unique_hostname directive.
7154 LOC: Config.EmailProgram
7156 Email program used to send mail if the cache dies.
7157 The default is "mail". The specified program must comply
7158 with the standard Unix mail syntax:
7159 mail-program recipient < mailfile
7161 Optional command line options can be specified.
7164 NAME: cache_effective_user
7166 DEFAULT: @DEFAULT_CACHE_EFFECTIVE_USER@
7167 LOC: Config.effectiveUser
7169 If you start Squid as root, it will change its effective/real
7170 UID/GID to the user specified below. The default is to change
7171 to UID of @DEFAULT_CACHE_EFFECTIVE_USER@.
7172 see also; cache_effective_group
7175 NAME: cache_effective_group
7178 DEFAULT_DOC: Use system group memberships of the cache_effective_user account
7179 LOC: Config.effectiveGroup
7181 Squid sets the GID to the effective user's default group ID
7182 (taken from the password file) and supplementary group list
7183 from the groups membership.
7185 If you want Squid to run with a specific GID regardless of
7186 the group memberships of the effective user then set this
7187 to the group (or GID) you want Squid to run as. When set
7188 all other group privileges of the effective user are ignored
7189 and only this GID is effective. If Squid is not started as
7190 root the user starting Squid MUST be member of the specified
7193 This option is not recommended by the Squid Team.
7194 Our preference is for administrators to configure a secure
7195 user account for squid with UID/GID matching system policies.
7198 NAME: httpd_suppress_version_string
7202 LOC: Config.onoff.httpd_suppress_version_string
7204 Suppress Squid version string info in HTTP headers and HTML error pages.
7207 NAME: visible_hostname
7209 LOC: Config.visibleHostname
7211 DEFAULT_DOC: Automatically detect the system host name
7213 If you want to present a special hostname in error messages, etc,
7214 define this. Otherwise, the return value of gethostname()
7215 will be used. If you have multiple caches in a cluster and
7216 get errors about IP-forwarding you must set them to have individual
7217 names with this setting.
7220 NAME: unique_hostname
7222 LOC: Config.uniqueHostname
7224 DEFAULT_DOC: Copy the value from visible_hostname
7226 If you want to have multiple machines with the same
7227 'visible_hostname' you must give each machine a different
7228 'unique_hostname' so forwarding loops can be detected.
7231 NAME: hostname_aliases
7233 LOC: Config.hostnameAliases
7236 A list of other DNS names your cache has.
7244 Minimum umask which should be enforced while the proxy
7245 is running, in addition to the umask set at startup.
7247 For a traditional octal representation of umasks, start
7252 OPTIONS FOR THE CACHE REGISTRATION SERVICE
7253 -----------------------------------------------------------------------------
7255 This section contains parameters for the (optional) cache
7256 announcement service. This service is provided to help
7257 cache administrators locate one another in order to join or
7258 create cache hierarchies.
7260 An 'announcement' message is sent (via UDP) to the registration
7261 service by Squid. By default, the announcement message is NOT
7262 SENT unless you enable it with 'announce_period' below.
7264 The announcement message includes your hostname, plus the
7265 following information from this configuration file:
7271 All current information is processed regularly and made
7272 available on the Web at http://www.ircache.net/Cache/Tracker/.
7275 NAME: announce_period
7277 LOC: Config.Announce.period
7279 DEFAULT_DOC: Announcement messages disabled.
7281 This is how frequently to send cache announcements.
7283 To enable announcing your cache, just set an announce period.
7286 announce_period 1 day
7291 DEFAULT: tracker.ircache.net
7292 LOC: Config.Announce.host
7294 Set the hostname where announce registration messages will be sent.
7296 See also announce_port and announce_file
7302 LOC: Config.Announce.file
7304 The contents of this file will be included in the announce
7305 registration messages.
7311 LOC: Config.Announce.port
7313 Set the port where announce registration messages will be sent.
7315 See also announce_host and announce_file
7319 HTTPD-ACCELERATOR OPTIONS
7320 -----------------------------------------------------------------------------
7323 NAME: httpd_accel_surrogate_id
7326 DEFAULT_DOC: visible_hostname is used if no specific ID is set.
7327 LOC: Config.Accel.surrogate_id
7329 Surrogates (http://www.esi.org/architecture_spec_1.0.html)
7330 need an identification token to allow control targeting. Because
7331 a farm of surrogates may all perform the same tasks, they may share
7332 an identification token.
7334 When the surrogate is a reverse-proxy, this ID is also
7335 used as cdn-id for CDN-Loop detection (RFC 8586).
7338 NAME: http_accel_surrogate_remote
7342 LOC: Config.onoff.surrogate_is_remote
7344 Remote surrogates (such as those in a CDN) honour the header
7345 "Surrogate-Control: no-store-remote".
7347 Set this to on to have squid behave as a remote surrogate.
7351 IFDEF: USE_SQUID_ESI
7352 COMMENT: libxml2|expat
7354 LOC: ESIParser::Type
7356 DEFAULT_DOC: Selects libxml2 if available at ./configure time or libexpat otherwise.
7358 Selects the XML parsing library to use when interpreting responses with
7361 To disable ESI handling completely, ./configure Squid with --disable-esi.
7365 DELAY POOL PARAMETERS
7366 -----------------------------------------------------------------------------
7370 TYPE: delay_pool_count
7372 IFDEF: USE_DELAY_POOLS
7375 This represents the number of delay pools to be used. For example,
7376 if you have one class 2 delay pool and one class 3 delays pool, you
7377 have a total of 2 delay pools.
7379 See also delay_parameters, delay_class, delay_access for pool
7380 configuration details.
7384 TYPE: delay_pool_class
7386 IFDEF: USE_DELAY_POOLS
7389 This defines the class of each delay pool. There must be exactly one
7390 delay_class line for each delay pool. For example, to define two
7391 delay pools, one of class 2 and one of class 3, the settings above
7395 delay_pools 4 # 4 delay pools
7396 delay_class 1 2 # pool 1 is a class 2 pool
7397 delay_class 2 3 # pool 2 is a class 3 pool
7398 delay_class 3 4 # pool 3 is a class 4 pool
7399 delay_class 4 5 # pool 4 is a class 5 pool
7401 The delay pool classes are:
7403 class 1 Everything is limited by a single aggregate
7406 class 2 Everything is limited by a single aggregate
7407 bucket as well as an "individual" bucket chosen
7408 from bits 25 through 32 of the IPv4 address.
7410 class 3 Everything is limited by a single aggregate
7411 bucket as well as a "network" bucket chosen
7412 from bits 17 through 24 of the IP address and a
7413 "individual" bucket chosen from bits 17 through
7414 32 of the IPv4 address.
7416 class 4 Everything in a class 3 delay pool, with an
7417 additional limit on a per user basis. This
7418 only takes effect if the username is established
7419 in advance - by forcing authentication in your
7422 class 5 Requests are grouped according their tag (see
7423 external_acl's tag= reply).
7426 Each pool also requires a delay_parameters directive to configure the pool size
7427 and speed limits used whenever the pool is applied to a request. Along with
7428 a set of delay_access directives to determine when it is used.
7430 NOTE: If an IP address is a.b.c.d
7431 -> bits 25 through 32 are "d"
7432 -> bits 17 through 24 are "c"
7433 -> bits 17 through 32 are "c * 256 + d"
7435 NOTE-2: Due to the use of bitmasks in class 2,3,4 pools they only apply to
7436 IPv4 traffic. Class 1 and 5 pools may be used with IPv6 traffic.
7438 This clause only supports fast acl types.
7439 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
7441 See also delay_parameters and delay_access.
7445 TYPE: delay_pool_access
7447 DEFAULT_DOC: Deny using the pool, unless allow rules exist in squid.conf for the pool.
7448 IFDEF: USE_DELAY_POOLS
7451 This is used to determine which delay pool a request falls into.
7453 delay_access is sorted per pool and the matching starts with pool 1,
7454 then pool 2, ..., and finally pool N. The first delay pool where the
7455 request is allowed is selected for the request. If it does not allow
7456 the request to any pool then the request is not delayed (default).
7458 For example, if you want some_big_clients in delay
7459 pool 1 and lotsa_little_clients in delay pool 2:
7461 delay_access 1 allow some_big_clients
7462 delay_access 1 deny all
7463 delay_access 2 allow lotsa_little_clients
7464 delay_access 2 deny all
7465 delay_access 3 allow authenticated_clients
7467 See also delay_parameters and delay_class.
7471 NAME: delay_parameters
7472 TYPE: delay_pool_rates
7474 IFDEF: USE_DELAY_POOLS
7477 This defines the parameters for a delay pool. Each delay pool has
7478 a number of "buckets" associated with it, as explained in the
7479 description of delay_class.
7481 For a class 1 delay pool, the syntax is:
7483 delay_parameters pool aggregate
7485 For a class 2 delay pool:
7487 delay_parameters pool aggregate individual
7489 For a class 3 delay pool:
7491 delay_parameters pool aggregate network individual
7493 For a class 4 delay pool:
7495 delay_parameters pool aggregate network individual user
7497 For a class 5 delay pool:
7499 delay_parameters pool tagrate
7501 The option variables are:
7503 pool a pool number - ie, a number between 1 and the
7504 number specified in delay_pools as used in
7507 aggregate the speed limit parameters for the aggregate bucket
7510 individual the speed limit parameters for the individual
7511 buckets (class 2, 3).
7513 network the speed limit parameters for the network buckets
7516 user the speed limit parameters for the user buckets
7519 tagrate the speed limit parameters for the tag buckets
7522 A pair of delay parameters is written restore/maximum, where restore is
7523 the number of bytes (not bits - modem and network speeds are usually
7524 quoted in bits) per second placed into the bucket, and maximum is the
7525 maximum number of bytes which can be in the bucket at any time.
7527 There must be one delay_parameters line for each delay pool.
7530 For example, if delay pool number 1 is a class 2 delay pool as in the
7531 above example, and is being used to strictly limit each host to 64Kbit/sec
7532 (plus overheads), with no overall limit, the line is:
7534 delay_parameters 1 none 8000/8000
7536 Note that 8 x 8K Byte/sec -> 64K bit/sec.
7538 Note that the word 'none' is used to represent no limit.
7541 And, if delay pool number 2 is a class 3 delay pool as in the above
7542 example, and you want to limit it to a total of 256Kbit/sec (strict limit)
7543 with each 8-bit network permitted 64Kbit/sec (strict limit) and each
7544 individual host permitted 4800bit/sec with a bucket maximum size of 64Kbits
7545 to permit a decent web page to be downloaded at a decent speed
7546 (if the network is not being limited due to overuse) but slow down
7547 large downloads more significantly:
7549 delay_parameters 2 32000/32000 8000/8000 600/8000
7551 Note that 8 x 32K Byte/sec -> 256K bit/sec.
7552 8 x 8K Byte/sec -> 64K bit/sec.
7553 8 x 600 Byte/sec -> 4800 bit/sec.
7556 Finally, for a class 4 delay pool as in the example - each user will
7557 be limited to 128Kbits/sec no matter how many workstations they are logged into.:
7559 delay_parameters 4 32000/32000 8000/8000 600/64000 16000/16000
7562 See also delay_class and delay_access.
7566 NAME: delay_initial_bucket_level
7567 COMMENT: (percent, 0-100)
7570 IFDEF: USE_DELAY_POOLS
7571 LOC: Config.Delay.initial
7573 The initial bucket percentage is used to determine how much is put
7574 in each bucket when squid starts, is reconfigured, or first notices
7575 a host accessing it (in class 2 and class 3, individual hosts and
7576 networks only have buckets associated with them once they have been
7581 CLIENT DELAY POOL PARAMETERS
7582 -----------------------------------------------------------------------------
7585 NAME: client_delay_pools
7586 TYPE: client_delay_pool_count
7588 IFDEF: USE_DELAY_POOLS
7589 LOC: Config.ClientDelay
7591 This option specifies the number of client delay pools used. It must
7592 preceed other client_delay_* options.
7595 client_delay_pools 2
7597 See also client_delay_parameters and client_delay_access.
7600 NAME: client_delay_initial_bucket_level
7601 COMMENT: (percent, 0-no_limit)
7604 IFDEF: USE_DELAY_POOLS
7605 LOC: Config.ClientDelay.initial
7607 This option determines the initial bucket size as a percentage of
7608 max_bucket_size from client_delay_parameters. Buckets are created
7609 at the time of the "first" connection from the matching IP. Idle
7610 buckets are periodically deleted up.
7612 You can specify more than 100 percent but note that such "oversized"
7613 buckets are not refilled until their size goes down to max_bucket_size
7614 from client_delay_parameters.
7617 client_delay_initial_bucket_level 50
7620 NAME: client_delay_parameters
7621 TYPE: client_delay_pool_rates
7623 IFDEF: USE_DELAY_POOLS
7624 LOC: Config.ClientDelay
7627 This option configures client-side bandwidth limits using the
7630 client_delay_parameters pool speed_limit max_bucket_size
7632 pool is an integer ID used for client_delay_access matching.
7634 speed_limit is bytes added to the bucket per second.
7636 max_bucket_size is the maximum size of a bucket, enforced after any
7637 speed_limit additions.
7639 Please see the delay_parameters option for more information and
7643 client_delay_parameters 1 1024 2048
7644 client_delay_parameters 2 51200 16384
7646 See also client_delay_access.
7650 NAME: client_delay_access
7651 TYPE: client_delay_pool_access
7653 DEFAULT_DOC: Deny use of the pool, unless allow rules exist in squid.conf for the pool.
7654 IFDEF: USE_DELAY_POOLS
7655 LOC: Config.ClientDelay
7657 This option determines the client-side delay pool for the
7660 client_delay_access pool_ID allow|deny acl_name
7662 All client_delay_access options are checked in their pool ID
7663 order, starting with pool 1. The first checked pool with allowed
7664 request is selected for the request. If no ACL matches or there
7665 are no client_delay_access options, the request bandwidth is not
7668 The ACL-selected pool is then used to find the
7669 client_delay_parameters for the request. Client-side pools are
7670 not used to aggregate clients. Clients are always aggregated
7671 based on their source IP addresses (one bucket per source IP).
7673 This clause only supports fast acl types.
7674 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
7675 Additionally, only the client TCP connection details are available.
7676 ACLs testing HTTP properties will not work.
7678 Please see delay_access for more examples.
7681 client_delay_access 1 allow low_rate_network
7682 client_delay_access 2 allow vips_network
7685 See also client_delay_parameters and client_delay_pools.
7688 NAME: response_delay_pool
7689 TYPE: response_delay_pool_parameters
7691 IFDEF: USE_DELAY_POOLS
7692 LOC: Config.MessageDelay
7694 This option configures client response bandwidth limits using the
7697 response_delay_pool name [option=value] ...
7699 name the response delay pool name
7703 individual-restore The speed limit of an individual
7704 bucket(bytes/s). To be used in conjunction
7705 with 'individual-maximum'.
7707 individual-maximum The maximum number of bytes which can
7708 be placed into the individual bucket. To be used
7709 in conjunction with 'individual-restore'.
7711 aggregate-restore The speed limit for the aggregate
7712 bucket(bytes/s). To be used in conjunction with
7713 'aggregate-maximum'.
7715 aggregate-maximum The maximum number of bytes which can
7716 be placed into the aggregate bucket. To be used
7717 in conjunction with 'aggregate-restore'.
7719 initial-bucket-level The initial bucket size as a percentage
7720 of individual-maximum.
7722 Individual and(or) aggregate bucket options may not be specified,
7723 meaning no individual and(or) aggregate speed limitation.
7724 See also response_delay_pool_access and delay_parameters for
7725 terminology details.
7728 NAME: response_delay_pool_access
7729 TYPE: response_delay_pool_access
7731 DEFAULT_DOC: Deny use of the pool, unless allow rules exist in squid.conf for the pool.
7732 IFDEF: USE_DELAY_POOLS
7733 LOC: Config.MessageDelay
7735 Determines whether a specific named response delay pool is used
7736 for the transaction. The syntax for this directive is:
7738 response_delay_pool_access pool_name allow|deny acl_name
7740 All response_delay_pool_access options are checked in the order
7741 they appear in this configuration file. The first rule with a
7742 matching ACL wins. If (and only if) an "allow" rule won, Squid
7743 assigns the response to the corresponding named delay pool.
7747 WCCPv1 AND WCCPv2 CONFIGURATION OPTIONS
7748 -----------------------------------------------------------------------------
7753 LOC: Config.Wccp.router
7755 DEFAULT_DOC: WCCP disabled.
7758 Use this option to define your WCCP ``home'' router for
7761 wccp_router supports a single WCCP(v1) router
7763 wccp2_router supports multiple WCCPv2 routers
7765 only one of the two may be used at the same time and defines
7766 which version of WCCP to use.
7770 TYPE: IpAddress_list
7771 LOC: Config.Wccp2.router
7773 DEFAULT_DOC: WCCPv2 disabled.
7776 Use this option to define your WCCP ``home'' router for
7779 wccp_router supports a single WCCP(v1) router
7781 wccp2_router supports multiple WCCPv2 routers
7783 only one of the two may be used at the same time and defines
7784 which version of WCCP to use.
7789 LOC: Config.Wccp.version
7793 This directive is only relevant if you need to set up WCCP(v1)
7794 to some very old and end-of-life Cisco routers. In all other
7795 setups it must be left unset or at the default setting.
7796 It defines an internal version in the WCCP(v1) protocol,
7797 with version 4 being the officially documented protocol.
7799 According to some users, Cisco IOS 11.2 and earlier only
7800 support WCCP version 3. If you're using that or an earlier
7801 version of IOS, you may need to change this value to 3, otherwise
7802 do not specify this parameter.
7805 NAME: wccp2_rebuild_wait
7807 LOC: Config.Wccp2.rebuildwait
7811 If this is enabled Squid will wait for the cache dir rebuild to finish
7812 before sending the first wccp2 HereIAm packet
7815 NAME: wccp2_forwarding_method
7817 LOC: Config.Wccp2.forwarding_method
7821 WCCP2 allows the setting of forwarding methods between the
7822 router/switch and the cache. Valid values are as follows:
7824 gre - GRE encapsulation (forward the packet in a GRE/WCCP tunnel)
7825 l2 - L2 redirect (forward the packet using Layer 2/MAC rewriting)
7827 Currently (as of IOS 12.4) cisco routers only support GRE.
7828 Cisco switches only support the L2 redirect assignment method.
7831 NAME: wccp2_return_method
7833 LOC: Config.Wccp2.return_method
7837 WCCP2 allows the setting of return methods between the
7838 router/switch and the cache for packets that the cache
7839 decides not to handle. Valid values are as follows:
7841 gre - GRE encapsulation (forward the packet in a GRE/WCCP tunnel)
7842 l2 - L2 redirect (forward the packet using Layer 2/MAC rewriting)
7844 Currently (as of IOS 12.4) cisco routers only support GRE.
7845 Cisco switches only support the L2 redirect assignment.
7847 If the "ip wccp redirect exclude in" command has been
7848 enabled on the cache interface, then it is still safe for
7849 the proxy server to use a l2 redirect method even if this
7850 option is set to GRE.
7853 NAME: wccp2_assignment_method
7855 LOC: Config.Wccp2.assignment_method
7859 WCCP2 allows the setting of methods to assign the WCCP hash
7860 Valid values are as follows:
7862 hash - Hash assignment
7863 mask - Mask assignment
7865 As a general rule, cisco routers support the hash assignment method
7866 and cisco switches support the mask assignment method.
7871 LOC: Config.Wccp2.info
7872 DEFAULT_IF_NONE: standard 0
7873 DEFAULT_DOC: Use the 'web-cache' standard service.
7876 WCCP2 allows for multiple traffic services. There are two
7877 types: "standard" and "dynamic". The standard type defines
7878 one service id - http (id 0). The dynamic service ids can be from
7879 51 to 255 inclusive. In order to use a dynamic service id
7880 one must define the type of traffic to be redirected; this is done
7881 using the wccp2_service_info option.
7883 The "standard" type does not require a wccp2_service_info option,
7884 just specifying the service id will suffice.
7886 MD5 service authentication can be enabled by adding
7887 "password=<password>" to the end of this service declaration.
7891 wccp2_service standard 0 # for the 'web-cache' standard service
7892 wccp2_service dynamic 80 # a dynamic service type which will be
7893 # fleshed out with subsequent options.
7894 wccp2_service standard 0 password=foo
7897 NAME: wccp2_service_info
7898 TYPE: wccp2_service_info
7899 LOC: Config.Wccp2.info
7903 Dynamic WCCPv2 services require further information to define the
7904 traffic you wish to have diverted.
7908 wccp2_service_info <id> protocol=<protocol> flags=<flag>,<flag>..
7909 priority=<priority> ports=<port>,<port>..
7911 The relevant WCCPv2 flags:
7912 + src_ip_hash, dst_ip_hash
7913 + source_port_hash, dst_port_hash
7914 + src_ip_alt_hash, dst_ip_alt_hash
7915 + src_port_alt_hash, dst_port_alt_hash
7918 The port list can be one to eight entries.
7922 wccp2_service_info 80 protocol=tcp flags=src_ip_hash,ports_source
7923 priority=240 ports=80
7925 Note: the service id must have been defined by a previous
7926 'wccp2_service dynamic <id>' entry.
7931 LOC: Config.Wccp2.weight
7935 Each cache server gets assigned a set of the destination
7936 hash proportional to their weight.
7941 LOC: Config.Wccp.address
7943 DEFAULT_DOC: Address selected by the operating system.
7946 Use this option if you require WCCPv2 to use a specific
7949 The default behavior is to not bind to any specific address.
7954 LOC: Config.Wccp2.address
7956 DEFAULT_DOC: Address selected by the operating system.
7959 Use this option if you require WCCP to use a specific
7962 The default behavior is to not bind to any specific address.
7966 PERSISTENT CONNECTION HANDLING
7967 -----------------------------------------------------------------------------
7969 Also see "pconn_timeout" in the TIMEOUTS section
7972 NAME: client_persistent_connections
7974 LOC: Config.onoff.client_pconns
7977 Persistent connection support for clients.
7978 Squid uses persistent connections (when allowed). You can use
7979 this option to disable persistent connections with clients.
7982 NAME: server_persistent_connections
7984 LOC: Config.onoff.server_pconns
7987 Persistent connection support for servers.
7988 Squid uses persistent connections (when allowed). You can use
7989 this option to disable persistent connections with servers.
7992 NAME: persistent_connection_after_error
7994 LOC: Config.onoff.error_pconns
7997 With this directive the use of persistent connections after
7998 HTTP errors can be disabled. Useful if you have clients
7999 who fail to handle errors on persistent connections proper.
8002 NAME: detect_broken_pconn
8004 LOC: Config.onoff.detect_broken_server_pconns
8007 Some servers have been found to incorrectly signal the use
8008 of HTTP/1.0 persistent connections even on replies not
8009 compatible, causing significant delays. This server problem
8010 has mostly been seen on redirects.
8012 By enabling this directive Squid attempts to detect such
8013 broken replies and automatically assume the reply is finished
8014 after 10 seconds timeout.
8018 CACHE DIGEST OPTIONS
8019 -----------------------------------------------------------------------------
8022 NAME: digest_generation
8023 IFDEF: USE_CACHE_DIGESTS
8025 LOC: Config.onoff.digest_generation
8028 This controls whether the server will generate a Cache Digest
8029 of its contents. By default, Cache Digest generation is
8030 enabled if Squid is compiled with --enable-cache-digests defined.
8033 NAME: digest_bits_per_entry
8034 IFDEF: USE_CACHE_DIGESTS
8036 LOC: Config.digest.bits_per_entry
8039 This is the number of bits of the server's Cache Digest which
8040 will be associated with the Digest entry for a given HTTP
8041 Method and URL (public key) combination. The default is 5.
8044 NAME: digest_rebuild_period
8045 IFDEF: USE_CACHE_DIGESTS
8048 LOC: Config.digest.rebuild_period
8051 This is the wait time between Cache Digest rebuilds.
8054 NAME: digest_rewrite_period
8056 IFDEF: USE_CACHE_DIGESTS
8058 LOC: Config.digest.rewrite_period
8061 This is the wait time between Cache Digest writes to
8065 NAME: digest_swapout_chunk_size
8068 IFDEF: USE_CACHE_DIGESTS
8069 LOC: Config.digest.swapout_chunk_size
8072 This is the number of bytes of the Cache Digest to write to
8073 disk at a time. It defaults to 4096 bytes (4KB), the Squid
8077 NAME: digest_rebuild_chunk_percentage
8078 COMMENT: (percent, 0-100)
8079 IFDEF: USE_CACHE_DIGESTS
8081 LOC: Config.digest.rebuild_chunk_percentage
8084 This is the percentage of the Cache Digest to be scanned at a
8085 time. By default it is set to 10% of the Cache Digest.
8090 -----------------------------------------------------------------------------
8095 LOC: Config.Port.snmp
8097 DEFAULT_DOC: SNMP disabled.
8100 The port number where Squid listens for SNMP requests. To enable
8101 SNMP support set this to a suitable port number. Port number
8102 3401 is often used for the Squid SNMP agent. By default it's
8103 set to "0" (disabled)
8111 LOC: Config.accessList.snmp
8113 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
8116 Allowing or denying access to the SNMP port.
8118 All access to the agent is denied by default.
8121 snmp_access allow|deny [!]aclname ...
8123 This clause only supports fast acl types.
8124 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
8127 snmp_access allow snmppublic localhost
8128 snmp_access deny all
8131 NAME: snmp_incoming_address
8133 LOC: Config.Addrs.snmp_incoming
8135 DEFAULT_DOC: Accept SNMP packets from all machine interfaces.
8138 Just like 'udp_incoming_address', but for the SNMP port.
8140 snmp_incoming_address is used for the SNMP socket receiving
8141 messages from SNMP agents.
8143 The default snmp_incoming_address is to listen on all
8144 available network interfaces.
8147 NAME: snmp_outgoing_address
8149 LOC: Config.Addrs.snmp_outgoing
8151 DEFAULT_DOC: Use snmp_incoming_address or an address selected by the operating system.
8154 Just like 'udp_outgoing_address', but for the SNMP port.
8156 snmp_outgoing_address is used for SNMP packets returned to SNMP
8159 If snmp_outgoing_address is not set it will use the same socket
8160 as snmp_incoming_address. Only change this if you want to have
8161 SNMP replies sent using another address than where this Squid
8162 listens for SNMP queries.
8164 NOTE, snmp_incoming_address and snmp_outgoing_address can not have
8165 the same value since they both use the same port.
8170 -----------------------------------------------------------------------------
8173 NAME: icp_port udp_port
8176 DEFAULT_DOC: ICP disabled.
8177 LOC: Config.Port.icp
8179 The port number where Squid sends and receives ICP queries to
8180 and from neighbor caches. The standard UDP port for ICP is 3130.
8183 icp_port @DEFAULT_ICP_PORT@
8190 DEFAULT_DOC: HTCP disabled.
8191 LOC: Config.Port.htcp
8193 The port number where Squid sends and receives HTCP queries to
8194 and from neighbor caches. To turn it on you want to set it to
8201 NAME: log_icp_queries
8205 LOC: Config.onoff.log_udp
8207 If set, ICP queries are logged to access.log. You may wish
8208 do disable this if your ICP load is VERY high to speed things
8209 up or to simplify log analysis.
8212 NAME: udp_incoming_address
8214 LOC:Config.Addrs.udp_incoming
8216 DEFAULT_DOC: Accept packets from all machine interfaces.
8218 udp_incoming_address is used for UDP packets received from other
8221 The default behavior is to not bind to any specific address.
8223 Only change this if you want to have all UDP queries received on
8224 a specific interface/address.
8226 NOTE: udp_incoming_address is used by the ICP, HTCP, and DNS
8227 modules. Altering it will affect all of them in the same manner.
8229 see also; udp_outgoing_address
8231 NOTE, udp_incoming_address and udp_outgoing_address can not
8232 have the same value since they both use the same port.
8235 NAME: udp_outgoing_address
8237 LOC: Config.Addrs.udp_outgoing
8239 DEFAULT_DOC: Use udp_incoming_address or an address selected by the operating system.
8241 udp_outgoing_address is used for UDP packets sent out to other
8244 The default behavior is to not bind to any specific address.
8246 Instead it will use the same socket as udp_incoming_address.
8247 Only change this if you want to have UDP queries sent using another
8248 address than where this Squid listens for UDP queries from other
8251 NOTE: udp_outgoing_address is used by the ICP, HTCP, and DNS
8252 modules. Altering it will affect all of them in the same manner.
8254 see also; udp_incoming_address
8256 NOTE, udp_incoming_address and udp_outgoing_address can not
8257 have the same value since they both use the same port.
8264 LOC: Config.onoff.icp_hit_stale
8266 If you want to return ICP_HIT for stale cache objects, set this
8267 option to 'on'. If you have sibling relationships with caches
8268 in other administrative domains, this should be 'off'. If you only
8269 have sibling relationships with caches under your control,
8270 it is probably okay to set this to 'on'.
8271 If set to 'on', your siblings should use the option "allow-miss"
8272 on their cache_peer lines for connecting to you.
8275 NAME: minimum_direct_hops
8278 LOC: Config.minDirectHops
8280 If using the ICMP pinging stuff, do direct fetches for sites
8281 which are no more than this many hops away.
8284 NAME: minimum_direct_rtt
8288 LOC: Config.minDirectRtt
8290 If using the ICMP pinging stuff, do direct fetches for sites
8291 which are no more than this many rtt milliseconds away.
8297 LOC: Config.Netdb.low
8299 The low water mark for the ICMP measurement database.
8301 Note: high watermark controlled by netdb_high directive.
8303 These watermarks are counts, not percents. The defaults are
8304 (low) 900 and (high) 1000. When the high water mark is
8305 reached, database entries will be deleted until the low
8312 LOC: Config.Netdb.high
8314 The high water mark for the ICMP measurement database.
8316 Note: low watermark controlled by netdb_low directive.
8318 These watermarks are counts, not percents. The defaults are
8319 (low) 900 and (high) 1000. When the high water mark is
8320 reached, database entries will be deleted until the low
8324 NAME: netdb_ping_period
8326 LOC: Config.Netdb.period
8329 The minimum period for measuring a site. There will be at
8330 least this much delay between successive pings to the same
8331 network. The default is five minutes.
8338 LOC: Config.onoff.query_icmp
8340 If you want to ask your peers to include ICMP data in their ICP
8341 replies, enable this option.
8343 If your peer has configured Squid (during compilation) with
8344 '--enable-icmp' that peer will send ICMP pings to origin server
8345 sites of the URLs it receives. If you enable this option the
8346 ICP replies from that peer will include the ICMP data (if available).
8347 Then, when choosing a parent cache, Squid will choose the parent with
8348 the minimal RTT to the origin server. When this happens, the
8349 hierarchy field of the access.log will be
8350 "CLOSEST_PARENT_MISS". This option is off by default.
8353 NAME: test_reachability
8357 LOC: Config.onoff.test_reachability
8359 When this is 'on', ICP MISS replies will be ICP_MISS_NOFETCH
8360 instead of ICP_MISS if the target host is NOT in the ICMP
8361 database, or has a zero RTT.
8364 NAME: icp_query_timeout
8367 DEFAULT_DOC: Dynamic detection.
8369 LOC: Config.Timeout.icp_query
8371 Normally Squid will automatically determine an optimal ICP
8372 query timeout value based on the round-trip-time of recent ICP
8373 queries. If you want to override the value determined by
8374 Squid, set this 'icp_query_timeout' to a non-zero value. This
8375 value is specified in MILLISECONDS, so, to use a 2-second
8376 timeout (the old default), you would write:
8378 icp_query_timeout 2000
8381 NAME: maximum_icp_query_timeout
8385 LOC: Config.Timeout.icp_query_max
8387 Normally the ICP query timeout is determined dynamically. But
8388 sometimes it can lead to very large values (say 5 seconds).
8389 Use this option to put an upper limit on the dynamic timeout
8390 value. Do NOT use this option to always use a fixed (instead
8391 of a dynamic) timeout value. To set a fixed timeout see the
8392 'icp_query_timeout' directive.
8395 NAME: minimum_icp_query_timeout
8399 LOC: Config.Timeout.icp_query_min
8401 Normally the ICP query timeout is determined dynamically. But
8402 sometimes it can lead to very small timeouts, even lower than
8403 the normal latency variance on your link due to traffic.
8404 Use this option to put an lower limit on the dynamic timeout
8405 value. Do NOT use this option to always use a fixed (instead
8406 of a dynamic) timeout value. To set a fixed timeout see the
8407 'icp_query_timeout' directive.
8410 NAME: background_ping_rate
8414 LOC: Config.backgroundPingRate
8416 Controls how often the ICP pings are sent to siblings that
8417 have background-ping set.
8421 MULTICAST ICP OPTIONS
8422 -----------------------------------------------------------------------------
8427 LOC: Config.mcast_group_list
8430 This tag specifies a list of multicast groups which your server
8431 should join to receive multicasted ICP queries.
8433 NOTE! Be very careful what you put here! Be sure you
8434 understand the difference between an ICP _query_ and an ICP
8435 _reply_. This option is to be set only if you want to RECEIVE
8436 multicast queries. Do NOT set this option to SEND multicast
8437 ICP (use cache_peer for that). ICP replies are always sent via
8438 unicast, so this option does not affect whether or not you will
8439 receive replies from multicast group members.
8441 You must be very careful to NOT use a multicast address which
8442 is already in use by another group of caches.
8444 If you are unsure about multicast, please read the Multicast
8445 chapter in the Squid FAQ (http://www.squid-cache.org/FAQ/).
8447 Usage: mcast_groups 239.128.16.128 224.0.1.20
8449 By default, Squid doesn't listen on any multicast groups.
8452 NAME: mcast_miss_addr
8453 IFDEF: MULTICAST_MISS_STREAM
8455 LOC: Config.mcast_miss.addr
8457 DEFAULT_DOC: disabled.
8459 If you enable this option, every "cache miss" URL will
8460 be sent out on the specified multicast address.
8462 Do not enable this option unless you are are absolutely
8463 certain you understand what you are doing.
8466 NAME: mcast_miss_ttl
8467 IFDEF: MULTICAST_MISS_STREAM
8469 LOC: Config.mcast_miss.ttl
8472 This is the time-to-live value for packets multicasted
8473 when multicasting off cache miss URLs is enabled. By
8474 default this is set to 'site scope', i.e. 16.
8477 NAME: mcast_miss_port
8478 IFDEF: MULTICAST_MISS_STREAM
8480 LOC: Config.mcast_miss.port
8483 This is the port number to be used in conjunction with
8487 NAME: mcast_miss_encode_key
8488 IFDEF: MULTICAST_MISS_STREAM
8490 LOC: Config.mcast_miss.encode_key
8491 DEFAULT: XXXXXXXXXXXXXXXX
8493 The URLs that are sent in the multicast miss stream are
8494 encrypted. This is the encryption key.
8497 NAME: mcast_icp_query_timeout
8501 LOC: Config.Timeout.mcast_icp_query
8503 For multicast peers, Squid regularly sends out ICP "probes" to
8504 count how many other peers are listening on the given multicast
8505 address. This value specifies how long Squid should wait to
8506 count all the replies. The default is 2000 msec, or 2
8511 INTERNAL ICON OPTIONS
8512 -----------------------------------------------------------------------------
8515 NAME: icon_directory
8517 LOC: Config.icons.directory
8518 DEFAULT: @DEFAULT_ICON_DIR@
8520 Where the icons are stored. These are normally kept in
8524 NAME: global_internal_static
8526 LOC: Config.onoff.global_internal_static
8529 This directive controls is Squid should intercept all requests for
8530 /squid-internal-static/ no matter which host the URL is requesting
8531 (default on setting), or if nothing special should be done for
8532 such URLs (off setting). The purpose of this directive is to make
8533 icons etc work better in complex cache hierarchies where it may
8534 not always be possible for all corners in the cache mesh to reach
8535 the server generating a directory listing.
8538 NAME: short_icon_urls
8540 LOC: Config.icons.use_short_names
8543 If this is enabled Squid will use short URLs for icons.
8544 If disabled it will revert to the old behavior of including
8545 it's own name and port in the URL.
8547 If you run a complex cache hierarchy with a mix of Squid and
8548 other proxies you may need to disable this directive.
8553 -----------------------------------------------------------------------------
8556 NAME: error_directory
8558 LOC: Config.errorDirectory
8560 DEFAULT_DOC: Send error pages in the clients preferred language
8562 If you wish to create your own versions of the default
8563 error files to customize them to suit your company copy
8564 the error/template files to another directory and point
8567 WARNING: This option will disable multi-language support
8568 on error pages if used.
8570 The squid developers are interested in making squid available in
8571 a wide variety of languages. If you are making translations for a
8572 language that Squid does not currently provide please consider
8573 contributing your translation back to the project.
8574 http://wiki.squid-cache.org/Translations
8576 The squid developers working on translations are happy to supply drop-in
8577 translated error files in exchange for any new language contributions.
8580 NAME: error_default_language
8581 IFDEF: USE_ERR_LOCALES
8583 LOC: Config.errorDefaultLanguage
8585 DEFAULT_DOC: Generate English language pages.
8587 Set the default language which squid will send error pages in
8588 if no existing translation matches the clients language
8591 If unset (default) generic English will be used.
8593 The squid developers are interested in making squid available in
8594 a wide variety of languages. If you are interested in making
8595 translations for any language see the squid wiki for details.
8596 http://wiki.squid-cache.org/Translations
8599 NAME: error_log_languages
8600 IFDEF: USE_ERR_LOCALES
8602 LOC: Config.errorLogMissingLanguages
8605 Log to cache.log what languages users are attempting to
8606 auto-negotiate for translations.
8608 Successful negotiations are not logged. Only failures
8609 have meaning to indicate that Squid may need an upgrade
8610 of its error page translations.
8613 NAME: err_page_stylesheet
8615 LOC: Config.errorStylesheet
8616 DEFAULT: @DEFAULT_CONFIG_DIR@/errorpage.css
8618 CSS Stylesheet to pattern the display of Squid default error pages.
8620 For information on CSS see http://www.w3.org/Style/CSS/
8625 LOC: Config.errHtmlText
8628 HTML text to include in error messages. Make this a "mailto"
8629 URL to your admin address, or maybe just a link to your
8630 organizations Web page.
8632 To include this in your error messages, you must rewrite
8633 the error template files (found in the "errors" directory).
8634 Wherever you want the 'err_html_text' line to appear,
8635 insert a %L tag in the error template file.
8638 NAME: email_err_data
8641 LOC: Config.onoff.emailErrData
8644 If enabled, information about the occurred error will be
8645 included in the mailto links of the ERR pages (if %W is set)
8646 so that the email body contains the data.
8647 Syntax is <A HREF="mailto:%w%W">%w</A>
8652 LOC: Config.denyInfoList
8655 Usage: deny_info err_page_name acl
8656 or deny_info http://... acl
8657 or deny_info TCP_RESET acl
8659 This can be used to return a ERR_ page for requests which
8660 do not pass the 'http_access' rules. Squid remembers the last
8661 acl it evaluated in http_access, and if a 'deny_info' line exists
8662 for that ACL Squid returns a corresponding error page.
8664 The acl is typically the last acl on the http_access deny line which
8665 denied access. The exceptions to this rule are:
8666 - When Squid needs to request authentication credentials. It's then
8667 the first authentication related acl encountered
8668 - When none of the http_access lines matches. It's then the last
8669 acl processed on the last http_access line.
8670 - When the decision to deny access was made by an adaptation service,
8671 the acl name is the corresponding eCAP or ICAP service_name.
8673 NP: If providing your own custom error pages with error_directory
8674 you may also specify them by your custom file name:
8675 Example: deny_info ERR_CUSTOM_ACCESS_DENIED bad_guys
8677 By defaut Squid will send "403 Forbidden". A different 4xx or 5xx
8678 may be specified by prefixing the file name with the code and a colon.
8679 e.g. 404:ERR_CUSTOM_ACCESS_DENIED
8681 Alternatively you can tell Squid to reset the TCP connection
8682 by specifying TCP_RESET.
8684 Or you can specify an error URL or URL pattern. The browsers will
8685 get redirected to the specified URL after formatting tags have
8686 been replaced. Redirect will be done with 302 or 307 according to
8687 HTTP/1.1 specs. A different 3xx code may be specified by prefixing
8688 the URL. e.g. 303:http://example.com/
8691 %a - username (if available. Password NOT included)
8692 %A - Local listening IP address the client connection was connected to
8695 %E - Error description
8697 %H - Request domain name
8698 %i - Client IP Address
8700 %O - Unescaped message result from external ACL helper
8701 %o - Message result from external ACL helper
8702 %p - Request Port number
8703 %P - Request Protocol name
8704 %R - Request URL path
8705 %T - Timestamp in RFC 1123 format
8706 %U - Full canonical URL from client
8707 (HTTPS URLs terminate with *)
8708 %u - Full canonical URL from client
8709 %w - Admin email from squid.conf
8711 %% - Literal percent (%) code
8716 OPTIONS INFLUENCING REQUEST FORWARDING
8717 -----------------------------------------------------------------------------
8720 NAME: nonhierarchical_direct
8722 LOC: Config.onoff.nonhierarchical_direct
8725 By default, Squid will send any non-hierarchical requests
8726 (not cacheable request type) direct to origin servers.
8728 When this is set to "off", Squid will prefer to send these
8729 requests to parents.
8731 Note that in most configurations, by turning this off you will only
8732 add latency to these request without any improvement in global hit
8735 This option only sets a preference. If the parent is unavailable a
8736 direct connection to the origin server may still be attempted. To
8737 completely prevent direct connections use never_direct.
8742 LOC: Config.onoff.prefer_direct
8745 Normally Squid tries to use parents for most requests. If you for some
8746 reason like it to first try going direct and only use a parent if
8747 going direct fails set this to on.
8749 By combining nonhierarchical_direct off and prefer_direct on you
8750 can set up Squid to use a parent as a backup path if going direct
8753 Note: If you want Squid to use parents for all requests see
8754 the never_direct directive. prefer_direct only modifies how Squid
8755 acts on cacheable requests.
8758 NAME: cache_miss_revalidate
8762 LOC: Config.onoff.cache_miss_revalidate
8764 RFC 7232 defines a conditional request mechanism to prevent
8765 response objects being unnecessarily transferred over the network.
8766 If that mechanism is used by the client and a cache MISS occurs
8767 it can prevent new cache entries being created.
8769 This option determines whether Squid on cache MISS will pass the
8770 client revalidation request to the server or tries to fetch new
8771 content for caching. It can be useful while the cache is mostly
8772 empty to more quickly have the cache populated by generating
8773 non-conditional GETs.
8775 When set to 'on' (default), Squid will pass all client If-* headers
8776 to the server. This permits server responses without a cacheable
8777 payload to be delivered and on MISS no new cache entry is created.
8779 When set to 'off' and if the request is cacheable, Squid will
8780 remove the clients If-Modified-Since and If-None-Match headers from
8781 the request sent to the server. This requests a 200 status response
8782 from the server to create a new cache entry with.
8787 LOC: Config.accessList.AlwaysDirect
8789 DEFAULT_DOC: Prevent any cache_peer being used for this request.
8791 Usage: always_direct allow|deny [!]aclname ...
8793 Here you can use ACL elements to specify requests which should
8794 ALWAYS be forwarded by Squid to the origin servers without using
8795 any peers. For example, to always directly forward requests for
8796 local servers ignoring any parents or siblings you may have use
8799 acl local-servers dstdomain my.domain.net
8800 always_direct allow local-servers
8802 To always forward FTP requests directly, use
8805 always_direct allow FTP
8807 NOTE: There is a similar, but opposite option named
8808 'never_direct'. You need to be aware that "always_direct deny
8809 foo" is NOT the same thing as "never_direct allow foo". You
8810 may need to use a deny rule to exclude a more-specific case of
8811 some other rule. Example:
8813 acl local-external dstdomain external.foo.net
8814 acl local-servers dstdomain .foo.net
8815 always_direct deny local-external
8816 always_direct allow local-servers
8818 NOTE: If your goal is to make the client forward the request
8819 directly to the origin server bypassing Squid then this needs
8820 to be done in the client configuration. Squid configuration
8821 can only tell Squid how Squid should fetch the object.
8823 NOTE: This directive is not related to caching. The replies
8824 is cached as usual even if you use always_direct. To not cache
8825 the replies see the 'cache' directive.
8827 This clause supports both fast and slow acl types.
8828 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
8833 LOC: Config.accessList.NeverDirect
8835 DEFAULT_DOC: Allow DNS results to be used for this request.
8837 Usage: never_direct allow|deny [!]aclname ...
8839 never_direct is the opposite of always_direct. Please read
8840 the description for always_direct if you have not already.
8842 With 'never_direct' you can use ACL elements to specify
8843 requests which should NEVER be forwarded directly to origin
8844 servers. For example, to force the use of a proxy for all
8845 requests, except those in your local domain use something like:
8847 acl local-servers dstdomain .foo.net
8848 never_direct deny local-servers
8849 never_direct allow all
8851 or if Squid is inside a firewall and there are local intranet
8852 servers inside the firewall use something like:
8854 acl local-intranet dstdomain .foo.net
8855 acl local-external dstdomain external.foo.net
8856 always_direct deny local-external
8857 always_direct allow local-intranet
8858 never_direct allow all
8860 This clause supports both fast and slow acl types.
8861 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
8865 ADVANCED NETWORKING OPTIONS
8866 -----------------------------------------------------------------------------
8869 NAME: incoming_udp_average incoming_icp_average
8872 LOC: Config.comm_incoming.udp.average
8874 Heavy voodoo here. I can't even believe you are reading this.
8875 Are you crazy? Don't even think about adjusting these unless
8876 you understand the algorithms in comm_select.c first!
8879 NAME: incoming_tcp_average incoming_http_average
8882 LOC: Config.comm_incoming.tcp.average
8884 Heavy voodoo here. I can't even believe you are reading this.
8885 Are you crazy? Don't even think about adjusting these unless
8886 you understand the algorithms in comm_select.c first!
8889 NAME: incoming_dns_average
8892 LOC: Config.comm_incoming.dns.average
8894 Heavy voodoo here. I can't even believe you are reading this.
8895 Are you crazy? Don't even think about adjusting these unless
8896 you understand the algorithms in comm_select.c first!
8899 NAME: min_udp_poll_cnt min_icp_poll_cnt
8902 LOC: Config.comm_incoming.udp.min_poll
8904 Heavy voodoo here. I can't even believe you are reading this.
8905 Are you crazy? Don't even think about adjusting these unless
8906 you understand the algorithms in comm_select.c first!
8909 NAME: min_dns_poll_cnt
8912 LOC: Config.comm_incoming.dns.min_poll
8914 Heavy voodoo here. I can't even believe you are reading this.
8915 Are you crazy? Don't even think about adjusting these unless
8916 you understand the algorithms in comm_select.c first!
8919 NAME: min_tcp_poll_cnt min_http_poll_cnt
8922 LOC: Config.comm_incoming.tcp.min_poll
8924 Heavy voodoo here. I can't even believe you are reading this.
8925 Are you crazy? Don't even think about adjusting these unless
8926 you understand the algorithms in comm_select.c first!
8932 LOC: Config.accept_filter
8936 The name of an accept(2) filter to install on Squid's
8937 listen socket(s). This feature is perhaps specific to
8938 FreeBSD and requires support in the kernel.
8940 The 'httpready' filter delays delivering new connections
8941 to Squid until a full HTTP request has been received.
8942 See the accf_http(9) man page for details.
8944 The 'dataready' filter delays delivering new connections
8945 to Squid until there is some data to process.
8946 See the accf_dataready(9) man page for details.
8950 The 'data' filter delays delivering of new connections
8951 to Squid until there is some data to process by TCP_ACCEPT_DEFER.
8952 You may optionally specify a number of seconds to wait by
8953 'data=N' where N is the number of seconds. Defaults to 30
8954 if not specified. See the tcp(7) man page for details.
8957 accept_filter httpready
8962 NAME: client_ip_max_connections
8964 LOC: Config.client_ip_max_connections
8966 DEFAULT_DOC: No limit.
8968 Set an absolute limit on the number of connections a single
8969 client IP can use. Any more than this and Squid will begin to drop
8970 new connections from the client until it closes some links.
8972 Note that this is a global limit. It affects all HTTP, HTCP, Gopher and FTP
8973 connections from the client. For finer control use the ACL access controls.
8975 Requires client_db to be enabled (the default).
8977 WARNING: This may noticably slow down traffic received via external proxies
8978 or NAT devices and cause them to rebound error messages back to their clients.
8981 NAME: tcp_recv_bufsize
8985 DEFAULT_DOC: Use operating system TCP defaults.
8986 LOC: Config.tcpRcvBufsz
8988 Size of receive buffer to set for TCP sockets. Probably just
8989 as easy to change your kernel's default.
8990 Omit from squid.conf to use the default buffer size.
8995 -----------------------------------------------------------------------------
9002 LOC: Adaptation::Icap::TheConfig.onoff
9005 If you want to enable the ICAP module support, set this to on.
9008 NAME: icap_connect_timeout
9011 LOC: Adaptation::Icap::TheConfig.connect_timeout_raw
9014 This parameter specifies how long to wait for the TCP connect to
9015 the requested ICAP server to complete before giving up and either
9016 terminating the HTTP transaction or bypassing the failure.
9018 The default for optional services is peer_connect_timeout.
9019 The default for essential services is connect_timeout.
9020 If this option is explicitly set, its value applies to all services.
9023 NAME: icap_io_timeout
9027 DEFAULT_DOC: Use read_timeout.
9028 LOC: Adaptation::Icap::TheConfig.io_timeout_raw
9031 This parameter specifies how long to wait for an I/O activity on
9032 an established, active ICAP connection before giving up and
9033 either terminating the HTTP transaction or bypassing the
9037 NAME: icap_service_failure_limit
9038 COMMENT: limit [in memory-depth time-units]
9039 TYPE: icap_service_failure_limit
9041 LOC: Adaptation::Icap::TheConfig
9044 The limit specifies the number of failures that Squid tolerates
9045 when establishing a new TCP connection with an ICAP service. If
9046 the number of failures exceeds the limit, the ICAP service is
9047 not used for new ICAP requests until it is time to refresh its
9050 A negative value disables the limit. Without the limit, an ICAP
9051 service will not be considered down due to connectivity failures
9052 between ICAP OPTIONS requests.
9054 Squid forgets ICAP service failures older than the specified
9055 value of memory-depth. The memory fading algorithm
9056 is approximate because Squid does not remember individual
9057 errors but groups them instead, splitting the option
9058 value into ten time slots of equal length.
9060 When memory-depth is 0 and by default this option has no
9061 effect on service failure expiration.
9063 Squid always forgets failures when updating service settings
9064 using an ICAP OPTIONS transaction, regardless of this option
9068 # suspend service usage after 10 failures in 5 seconds:
9069 icap_service_failure_limit 10 in 5 seconds
9072 NAME: icap_service_revival_delay
9075 LOC: Adaptation::Icap::TheConfig.service_revival_delay
9078 The delay specifies the number of seconds to wait after an ICAP
9079 OPTIONS request failure before requesting the options again. The
9080 failed ICAP service is considered "down" until fresh OPTIONS are
9083 The actual delay cannot be smaller than the hardcoded minimum
9084 delay of 30 seconds.
9087 NAME: icap_preview_enable
9091 LOC: Adaptation::Icap::TheConfig.preview_enable
9094 The ICAP Preview feature allows the ICAP server to handle the
9095 HTTP message by looking only at the beginning of the message body
9096 or even without receiving the body at all. In some environments,
9097 previews greatly speedup ICAP processing.
9099 During an ICAP OPTIONS transaction, the server may tell Squid what
9100 HTTP messages should be previewed and how big the preview should be.
9101 Squid will not use Preview if the server did not request one.
9103 To disable ICAP Preview for all ICAP services, regardless of
9104 individual ICAP server OPTIONS responses, set this option to "off".
9106 icap_preview_enable off
9109 NAME: icap_preview_size
9112 LOC: Adaptation::Icap::TheConfig.preview_size
9114 DEFAULT_DOC: No preview sent.
9116 The default size of preview data to be sent to the ICAP server.
9117 This value might be overwritten on a per server basis by OPTIONS requests.
9120 NAME: icap_206_enable
9124 LOC: Adaptation::Icap::TheConfig.allow206_enable
9127 206 (Partial Content) responses is an ICAP extension that allows the
9128 ICAP agents to optionally combine adapted and original HTTP message
9129 content. The decision to combine is postponed until the end of the
9130 ICAP response. Squid supports Partial Content extension by default.
9132 Activation of the Partial Content extension is negotiated with each
9133 ICAP service during OPTIONS exchange. Most ICAP servers should handle
9134 negotation correctly even if they do not support the extension, but
9135 some might fail. To disable Partial Content support for all ICAP
9136 services and to avoid any negotiation, set this option to "off".
9142 NAME: icap_default_options_ttl
9145 LOC: Adaptation::Icap::TheConfig.default_options_ttl
9148 The default TTL value for ICAP OPTIONS responses that don't have
9149 an Options-TTL header.
9152 NAME: icap_persistent_connections
9156 LOC: Adaptation::Icap::TheConfig.reuse_connections
9159 Whether or not Squid should use persistent connections to
9163 NAME: adaptation_send_client_ip icap_send_client_ip
9165 IFDEF: USE_ADAPTATION
9167 LOC: Adaptation::Config::send_client_ip
9170 If enabled, Squid shares HTTP client IP information with adaptation
9171 services. For ICAP, Squid adds the X-Client-IP header to ICAP requests.
9172 For eCAP, Squid sets the libecap::metaClientIp transaction option.
9174 See also: adaptation_uses_indirect_client
9177 NAME: adaptation_send_username icap_send_client_username
9179 IFDEF: USE_ADAPTATION
9181 LOC: Adaptation::Config::send_username
9184 This sends authenticated HTTP client username (if available) to
9185 the adaptation service.
9187 For ICAP, the username value is encoded based on the
9188 icap_client_username_encode option and is sent using the header
9189 specified by the icap_client_username_header option.
9192 NAME: icap_client_username_header
9195 LOC: Adaptation::Icap::TheConfig.client_username_header
9196 DEFAULT: X-Client-Username
9198 ICAP request header name to use for adaptation_send_username.
9201 NAME: icap_client_username_encode
9205 LOC: Adaptation::Icap::TheConfig.client_username_encode
9208 Whether to base64 encode the authenticated client username.
9212 TYPE: icap_service_type
9214 LOC: Adaptation::Icap::TheConfig
9217 Defines a single ICAP service using the following format:
9219 icap_service id vectoring_point uri [option ...]
9222 an opaque identifier or name which is used to direct traffic to
9223 this specific service. Must be unique among all adaptation
9224 services in squid.conf.
9226 vectoring_point: reqmod_precache|reqmod_postcache|respmod_precache|respmod_postcache
9227 This specifies at which point of transaction processing the
9228 ICAP service should be activated. *_postcache vectoring points
9229 are not yet supported.
9231 uri: icap://servername:port/servicepath
9232 ICAP server and service location.
9233 icaps://servername:port/servicepath
9234 The "icap:" URI scheme is used for traditional ICAP server and
9235 service location (default port is 1344, connections are not
9236 encrypted). The "icaps:" URI scheme is for Secure ICAP
9237 services that use SSL/TLS-encrypted ICAP connections (by
9238 default, on port 11344).
9240 ICAP does not allow a single service to handle both REQMOD and RESPMOD
9241 transactions. Squid does not enforce that requirement. You can specify
9242 services with the same service_url and different vectoring_points. You
9243 can even specify multiple identical services as long as their
9244 service_names differ.
9246 To activate a service, use the adaptation_access directive. To group
9247 services, use adaptation_service_chain and adaptation_service_set.
9249 Service options are separated by white space. ICAP services support
9250 the following name=value options:
9253 If set to 'on' or '1', the ICAP service is treated as
9254 optional. If the service cannot be reached or malfunctions,
9255 Squid will try to ignore any errors and process the message as
9256 if the service was not enabled. No all ICAP errors can be
9257 bypassed. If set to 0, the ICAP service is treated as
9258 essential and all ICAP errors will result in an error page
9259 returned to the HTTP client.
9261 Bypass is off by default: services are treated as essential.
9264 If set to 'on' or '1', the ICAP service is allowed to
9265 dynamically change the current message adaptation plan by
9266 returning a chain of services to be used next. The services
9267 are specified using the X-Next-Services ICAP response header
9268 value, formatted as a comma-separated list of service names.
9269 Each named service should be configured in squid.conf. Other
9270 services are ignored. An empty X-Next-Services value results
9271 in an empty plan which ends the current adaptation.
9273 Dynamic adaptation plan may cross or cover multiple supported
9274 vectoring points in their natural processing order.
9276 Routing is not allowed by default: the ICAP X-Next-Services
9277 response header is ignored.
9280 Only has effect on split-stack systems. The default on those systems
9281 is to use IPv4-only connections. When set to 'on' this option will
9282 make Squid use IPv6-only connections to contact this ICAP service.
9284 on-overload=block|bypass|wait|force
9285 If the service Max-Connections limit has been reached, do
9286 one of the following for each new ICAP transaction:
9287 * block: send an HTTP error response to the client
9288 * bypass: ignore the "over-connected" ICAP service
9289 * wait: wait (in a FIFO queue) for an ICAP connection slot
9290 * force: proceed, ignoring the Max-Connections limit
9292 In SMP mode with N workers, each worker assumes the service
9293 connection limit is Max-Connections/N, even though not all
9294 workers may use a given service.
9296 The default value is "bypass" if service is bypassable,
9297 otherwise it is set to "wait".
9301 Use the given number as the Max-Connections limit, regardless
9302 of the Max-Connections value given by the service, if any.
9304 connection-encryption=on|off
9305 Determines the ICAP service effect on the connections_encrypted
9308 The default is "on" for Secure ICAP services (i.e., those
9309 with the icaps:// service URIs scheme) and "off" for plain ICAP
9312 Does not affect ICAP connections (e.g., does not turn Secure
9315 ==== ICAPS / TLS OPTIONS ====
9317 These options are used for Secure ICAP (icaps://....) services only.
9319 tls-cert=/path/to/ssl/certificate
9320 A client X.509 certificate to use when connecting to
9323 tls-key=/path/to/ssl/key
9324 The private key corresponding to the previous
9327 If tls-key= is not specified tls-cert= is assumed to
9328 reference a PEM file containing both the certificate
9331 tls-cipher=... The list of valid TLS/SSL ciphers to use when connecting
9332 to this icap server.
9335 The minimum TLS protocol version to permit. To control
9336 SSLv3 use the tls-options= parameter.
9337 Supported Values: 1.0 (default), 1.1, 1.2
9339 tls-options=... Specify various OpenSSL library options:
9341 NO_SSLv3 Disallow the use of SSLv3
9344 Always create a new key when using
9345 temporary/ephemeral DH key exchanges
9347 ALL Enable various bug workarounds
9348 suggested as "harmless" by OpenSSL
9349 Be warned that this reduces SSL/TLS
9350 strength to some attacks.
9352 See the OpenSSL SSL_CTX_set_options documentation for a
9353 more complete list. Options relevant only to SSLv2 are
9356 tls-cafile= PEM file containing CA certificates to use when verifying
9357 the icap server certificate.
9358 Use to specify intermediate CA certificate(s) if not sent
9359 by the server. Or the full CA chain for the server when
9360 using the tls-default-ca=off flag.
9361 May be repeated to load multiple files.
9363 tls-capath=... A directory containing additional CA certificates to
9364 use when verifying the icap server certificate.
9365 Requires OpenSSL or LibreSSL.
9367 tls-crlfile=... A certificate revocation list file to use when
9368 verifying the icap server certificate.
9370 tls-flags=... Specify various flags modifying the Squid TLS implementation:
9373 Accept certificates even if they fail to
9376 Don't verify the icap server certificate
9377 matches the server name
9379 tls-default-ca[=off]
9380 Whether to use the system Trusted CAs. Default is ON.
9382 tls-domain= The icap server name as advertised in it's certificate.
9383 Used for verifying the correctness of the received icap
9384 server certificate. If not specified the icap server
9385 hostname extracted from ICAP URI will be used.
9387 Older icap_service format without optional named parameters is
9388 deprecated but supported for backward compatibility.
9391 icap_service svcBlocker reqmod_precache icap://icap1.mydomain.net:1344/reqmod bypass=0
9392 icap_service svcLogger reqmod_precache icaps://icap2.mydomain.net:11344/reqmod routing=on
9396 TYPE: icap_class_type
9401 This deprecated option was documented to define an ICAP service
9402 chain, even though it actually defined a set of similar, redundant
9403 services, and the chains were not supported.
9405 To define a set of redundant services, please use the
9406 adaptation_service_set directive. For service chains, use
9407 adaptation_service_chain.
9411 TYPE: icap_access_type
9416 This option is deprecated. Please use adaptation_access, which
9417 has the same ICAP functionality, but comes with better
9418 documentation, and eCAP support.
9423 -----------------------------------------------------------------------------
9430 LOC: Adaptation::Ecap::TheConfig.onoff
9433 Controls whether eCAP support is enabled.
9437 TYPE: ecap_service_type
9439 LOC: Adaptation::Ecap::TheConfig
9442 Defines a single eCAP service
9444 ecap_service id vectoring_point uri [option ...]
9447 an opaque identifier or name which is used to direct traffic to
9448 this specific service. Must be unique among all adaptation
9449 services in squid.conf.
9451 vectoring_point: reqmod_precache|reqmod_postcache|respmod_precache|respmod_postcache
9452 This specifies at which point of transaction processing the
9453 eCAP service should be activated. *_postcache vectoring points
9454 are not yet supported.
9456 uri: ecap://vendor/service_name?custom&cgi=style¶meters=optional
9457 Squid uses the eCAP service URI to match this configuration
9458 line with one of the dynamically loaded services. Each loaded
9459 eCAP service must have a unique URI. Obtain the right URI from
9460 the service provider.
9462 To activate a service, use the adaptation_access directive. To group
9463 services, use adaptation_service_chain and adaptation_service_set.
9465 Service options are separated by white space. eCAP services support
9466 the following name=value options:
9469 If set to 'on' or '1', the eCAP service is treated as optional.
9470 If the service cannot be reached or malfunctions, Squid will try
9471 to ignore any errors and process the message as if the service
9472 was not enabled. No all eCAP errors can be bypassed.
9473 If set to 'off' or '0', the eCAP service is treated as essential
9474 and all eCAP errors will result in an error page returned to the
9477 Bypass is off by default: services are treated as essential.
9480 If set to 'on' or '1', the eCAP service is allowed to
9481 dynamically change the current message adaptation plan by
9482 returning a chain of services to be used next.
9484 Dynamic adaptation plan may cross or cover multiple supported
9485 vectoring points in their natural processing order.
9487 Routing is not allowed by default.
9489 connection-encryption=on|off
9490 Determines the eCAP service effect on the connections_encrypted
9493 Defaults to "on", which does not taint the master transaction
9496 Does not affect eCAP API calls.
9498 Older ecap_service format without optional named parameters is
9499 deprecated but supported for backward compatibility.
9503 ecap_service s1 reqmod_precache ecap://filters.R.us/leakDetector?on_error=block bypass=off
9504 ecap_service s2 respmod_precache ecap://filters.R.us/virusFilter config=/etc/vf.cfg bypass=on
9507 NAME: loadable_modules
9509 IFDEF: USE_LOADABLE_MODULES
9510 LOC: Config.loadable_module_names
9513 Instructs Squid to load the specified dynamic module(s) or activate
9514 preloaded module(s).
9516 loadable_modules @DEFAULT_PREFIX@/lib/MinimalAdapter.so
9520 MESSAGE ADAPTATION OPTIONS
9521 -----------------------------------------------------------------------------
9524 NAME: adaptation_service_set
9525 TYPE: adaptation_service_set_type
9526 IFDEF: USE_ADAPTATION
9531 Configures an ordered set of similar, redundant services. This is
9532 useful when hot standby or backup adaptation servers are available.
9534 adaptation_service_set set_name service_name1 service_name2 ...
9536 The named services are used in the set declaration order. The first
9537 applicable adaptation service from the set is used first. The next
9538 applicable service is tried if and only if the transaction with the
9539 previous service fails and the message waiting to be adapted is still
9542 When adaptation starts, broken services are ignored as if they were
9543 not a part of the set. A broken service is a down optional service.
9545 The services in a set must be attached to the same vectoring point
9546 (e.g., pre-cache) and use the same adaptation method (e.g., REQMOD).
9548 If all services in a set are optional then adaptation failures are
9549 bypassable. If all services in the set are essential, then a
9550 transaction failure with one service may still be retried using
9551 another service from the set, but when all services fail, the master
9552 transaction fails as well.
9554 A set may contain a mix of optional and essential services, but that
9555 is likely to lead to surprising results because broken services become
9556 ignored (see above), making previously bypassable failures fatal.
9557 Technically, it is the bypassability of the last failed service that
9560 See also: adaptation_access adaptation_service_chain
9563 adaptation_service_set svcBlocker urlFilterPrimary urlFilterBackup
9564 adaptation service_set svcLogger loggerLocal loggerRemote
9567 NAME: adaptation_service_chain
9568 TYPE: adaptation_service_chain_type
9569 IFDEF: USE_ADAPTATION
9574 Configures a list of complementary services that will be applied
9575 one-by-one, forming an adaptation chain or pipeline. This is useful
9576 when Squid must perform different adaptations on the same message.
9578 adaptation_service_chain chain_name service_name1 svc_name2 ...
9580 The named services are used in the chain declaration order. The first
9581 applicable adaptation service from the chain is used first. The next
9582 applicable service is applied to the successful adaptation results of
9583 the previous service in the chain.
9585 When adaptation starts, broken services are ignored as if they were
9586 not a part of the chain. A broken service is a down optional service.
9588 Request satisfaction terminates the adaptation chain because Squid
9589 does not currently allow declaration of RESPMOD services at the
9590 "reqmod_precache" vectoring point (see icap_service or ecap_service).
9592 The services in a chain must be attached to the same vectoring point
9593 (e.g., pre-cache) and use the same adaptation method (e.g., REQMOD).
9595 A chain may contain a mix of optional and essential services. If an
9596 essential adaptation fails (or the failure cannot be bypassed for
9597 other reasons), the master transaction fails. Otherwise, the failure
9598 is bypassed as if the failed adaptation service was not in the chain.
9600 See also: adaptation_access adaptation_service_set
9603 adaptation_service_chain svcRequest requestLogger urlFilter leakDetector
9606 NAME: adaptation_access
9607 TYPE: adaptation_access_type
9608 IFDEF: USE_ADAPTATION
9611 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
9613 Sends an HTTP transaction to an ICAP or eCAP adaptation service.
9615 adaptation_access service_name allow|deny [!]aclname...
9616 adaptation_access set_name allow|deny [!]aclname...
9618 At each supported vectoring point, the adaptation_access
9619 statements are processed in the order they appear in this
9620 configuration file. Statements pointing to the following services
9621 are ignored (i.e., skipped without checking their ACL):
9623 - services serving different vectoring points
9624 - "broken-but-bypassable" services
9625 - "up" services configured to ignore such transactions
9626 (e.g., based on the ICAP Transfer-Ignore header).
9628 When a set_name is used, all services in the set are checked
9629 using the same rules, to find the first applicable one. See
9630 adaptation_service_set for details.
9632 If an access list is checked and there is a match, the
9633 processing stops: For an "allow" rule, the corresponding
9634 adaptation service is used for the transaction. For a "deny"
9635 rule, no adaptation service is activated.
9637 It is currently not possible to apply more than one adaptation
9638 service at the same vectoring point to the same HTTP transaction.
9640 See also: icap_service and ecap_service
9643 adaptation_access service_1 allow all
9646 NAME: adaptation_service_iteration_limit
9648 IFDEF: USE_ADAPTATION
9649 LOC: Adaptation::Config::service_iteration_limit
9652 Limits the number of iterations allowed when applying adaptation
9653 services to a message. If your longest adaptation set or chain
9654 may have more than 16 services, increase the limit beyond its
9655 default value of 16. If detecting infinite iteration loops sooner
9656 is critical, make the iteration limit match the actual number
9657 of services in your longest adaptation set or chain.
9659 Infinite adaptation loops are most likely with routing services.
9661 See also: icap_service routing=1
9664 NAME: adaptation_masterx_shared_names
9666 IFDEF: USE_ADAPTATION
9667 LOC: Adaptation::Config::masterx_shared_name
9670 For each master transaction (i.e., the HTTP request and response
9671 sequence, including all related ICAP and eCAP exchanges), Squid
9672 maintains a table of metadata. The table entries are (name, value)
9673 pairs shared among eCAP and ICAP exchanges. The table is destroyed
9674 with the master transaction.
9676 This option specifies the table entry names that Squid must accept
9677 from and forward to the adaptation transactions.
9679 An ICAP REQMOD or RESPMOD transaction may set an entry in the
9680 shared table by returning an ICAP header field with a name
9681 specified in adaptation_masterx_shared_names.
9683 An eCAP REQMOD or RESPMOD transaction may set an entry in the
9684 shared table by implementing the libecap::visitEachOption() API
9685 to provide an option with a name specified in
9686 adaptation_masterx_shared_names.
9688 Squid will store and forward the set entry to subsequent adaptation
9689 transactions within the same master transaction scope.
9691 Only one shared entry name is supported at this time.
9694 # share authentication information among ICAP services
9695 adaptation_masterx_shared_names X-Subscriber-ID
9698 NAME: adaptation_meta
9700 IFDEF: USE_ADAPTATION
9701 LOC: Adaptation::Config::metaHeaders
9704 This option allows Squid administrator to add custom ICAP request
9705 headers or eCAP options to Squid ICAP requests or eCAP transactions.
9706 Use it to pass custom authentication tokens and other
9707 transaction-state related meta information to an ICAP/eCAP service.
9709 The addition of a meta header is ACL-driven:
9710 adaptation_meta name value [!]aclname ...
9712 Processing for a given header name stops after the first ACL list match.
9713 Thus, it is impossible to add two headers with the same name. If no ACL
9714 lists match for a given header name, no such header is added. For
9717 # do not debug transactions except for those that need debugging
9718 adaptation_meta X-Debug 1 needs_debugging
9720 # log all transactions except for those that must remain secret
9721 adaptation_meta X-Log 1 !keep_secret
9723 # mark transactions from users in the "G 1" group
9724 adaptation_meta X-Authenticated-Groups "G 1" authed_as_G1
9726 The "value" parameter may be a regular squid.conf token or a "double
9727 quoted string". Within the quoted string, use backslash (\) to escape
9728 any character, which is currently only useful for escaping backslashes
9729 and double quotes. For example,
9730 "this string has one backslash (\\) and two \"quotes\""
9732 Used adaptation_meta header values may be logged via %note
9733 logformat code. If multiple adaptation_meta headers with the same name
9734 are used during master transaction lifetime, the header values are
9735 logged in the order they were used and duplicate values are ignored
9736 (only the first repeated value will be logged).
9742 LOC: Adaptation::Icap::TheConfig.repeat
9743 DEFAULT_IF_NONE: deny all
9745 This ACL determines which retriable ICAP transactions are
9746 retried. Transactions that received a complete ICAP response
9747 and did not have to consume or produce HTTP bodies to receive
9748 that response are usually retriable.
9750 icap_retry allow|deny [!]aclname ...
9752 Squid automatically retries some ICAP I/O timeouts and errors
9753 due to persistent connection race conditions.
9755 See also: icap_retry_limit
9758 NAME: icap_retry_limit
9761 LOC: Adaptation::Icap::TheConfig.repeat_limit
9763 DEFAULT_DOC: No retries are allowed.
9765 Limits the number of retries allowed.
9767 Communication errors due to persistent connection race
9768 conditions are unavoidable, automatically retried, and do not
9769 count against this limit.
9771 See also: icap_retry
9777 -----------------------------------------------------------------------------
9780 NAME: check_hostnames
9783 LOC: Config.onoff.check_hostnames
9785 For security and stability reasons Squid can check
9786 hostnames for Internet standard RFC compliance. If you want
9787 Squid to perform these checks turn this directive on.
9790 NAME: allow_underscore
9793 LOC: Config.onoff.allow_underscore
9795 Underscore characters is not strictly allowed in Internet hostnames
9796 but nevertheless used by many sites. Set this to off if you want
9797 Squid to be strict about the standard.
9798 This check is performed only when check_hostnames is set to on.
9801 NAME: dns_retransmit_interval
9804 LOC: Config.Timeout.idns_retransmit
9806 Initial retransmit interval for DNS queries. The interval is
9807 doubled each time all configured DNS servers have been tried.
9813 LOC: Config.Timeout.idns_query
9815 DNS Query timeout. If no response is received to a DNS query
9816 within this time all DNS servers for the queried domain
9817 are assumed to be unavailable.
9820 NAME: dns_packet_max
9822 DEFAULT_DOC: EDNS disabled
9824 LOC: Config.dns.packet_max
9826 Maximum number of bytes packet size to advertise via EDNS.
9827 Set to "none" to disable EDNS large packet support.
9829 For legacy reasons DNS UDP replies will default to 512 bytes which
9830 is too small for many responses. EDNS provides a means for Squid to
9831 negotiate receiving larger responses back immediately without having
9832 to failover with repeat requests. Responses larger than this limit
9833 will retain the old behaviour of failover to TCP DNS.
9835 Squid has no real fixed limit internally, but allowing packet sizes
9836 over 1500 bytes requires network jumbogram support and is usually not
9839 WARNING: The RFC also indicates that some older resolvers will reply
9840 with failure of the whole request if the extension is added. Some
9841 resolvers have already been identified which will reply with mangled
9842 EDNS response on occasion. Usually in response to many-KB jumbogram
9843 sizes being advertised by Squid.
9844 Squid will currently treat these both as an unable-to-resolve domain
9845 even if it would be resolvable without EDNS.
9852 DEFAULT_DOC: Search for single-label domain names is disabled.
9853 LOC: Config.onoff.res_defnames
9855 Normally the RES_DEFNAMES resolver option is disabled
9856 (see res_init(3)). This prevents caches in a hierarchy
9857 from interpreting single-component hostnames locally. To allow
9858 Squid to handle single-component names, enable this option.
9861 NAME: dns_multicast_local
9865 DEFAULT_DOC: Search for .local and .arpa names is disabled.
9866 LOC: Config.onoff.dns_mdns
9868 When set to on, Squid sends multicast DNS lookups on the local
9869 network for domains ending in .local and .arpa.
9870 This enables local servers and devices to be contacted in an
9871 ad-hoc or zero-configuration network environment.
9874 NAME: dns_nameservers
9877 DEFAULT_DOC: Use operating system definitions
9878 LOC: Config.dns.nameservers
9880 Use this if you want to specify a list of DNS name servers
9881 (IP addresses) to use instead of those given in your
9882 /etc/resolv.conf file.
9884 On Windows platforms, if no value is specified here or in
9885 the /etc/resolv.conf file, the list of DNS name servers are
9886 taken from the Windows registry, both static and dynamic DHCP
9887 configurations are supported.
9889 Example: dns_nameservers 10.0.0.1 192.172.0.4
9894 DEFAULT: @DEFAULT_HOSTS@
9895 LOC: Config.etcHostsPath
9897 Location of the host-local IP name-address associations
9898 database. Most Operating Systems have such a file on different
9900 - Un*X & Linux: /etc/hosts
9901 - Windows NT/2000: %SystemRoot%\system32\drivers\etc\hosts
9902 (%SystemRoot% value install default is c:\winnt)
9903 - Windows XP/2003: %SystemRoot%\system32\drivers\etc\hosts
9904 (%SystemRoot% value install default is c:\windows)
9905 - Windows 9x/Me: %windir%\hosts
9906 (%windir% value is usually c:\windows)
9907 - Cygwin: /etc/hosts
9909 The file contains newline-separated definitions, in the
9910 form ip_address_in_dotted_form name [name ...] names are
9911 whitespace-separated. Lines beginning with an hash (#)
9912 character are comments.
9914 The file is checked at startup and upon configuration.
9915 If set to 'none', it won't be checked.
9916 If append_domain is used, that domain will be added to
9917 domain-local (i.e. not containing any dot character) host
9923 LOC: Config.appendDomain
9925 DEFAULT_DOC: Use operating system definitions
9927 Appends local domain name to hostnames without any dots in
9928 them. append_domain must begin with a period.
9930 Be warned there are now Internet names with no dots in
9931 them using only top-domain names, so setting this may
9932 cause some Internet sites to become unavailable.
9935 append_domain .yourdomain.com
9938 NAME: ignore_unknown_nameservers
9940 LOC: Config.onoff.ignore_unknown_nameservers
9943 By default Squid checks that DNS responses are received
9944 from the same IP addresses they are sent to. If they
9945 don't match, Squid ignores the response and writes a warning
9946 message to cache.log. You can allow responses from unknown
9947 nameservers by setting this option to 'off'.
9953 LOC: Config.dns.v4_first
9955 With the IPv6 Internet being as fast or faster than IPv4 Internet
9956 for most networks Squid prefers to contact websites over IPv6.
9958 This option reverses the order of preference to make Squid contact
9959 dual-stack websites over IPv4 first. Squid will still perform both
9960 IPv6 and IPv4 DNS lookups before connecting.
9963 This option will restrict the situations under which IPv6
9964 connectivity is used (and tested). Hiding network problems
9965 which would otherwise be detected and warned about.
9969 COMMENT: (number of entries)
9972 LOC: Config.ipcache.size
9974 Maximum number of DNS IP cache entries.
9981 LOC: Config.ipcache.low
9988 LOC: Config.ipcache.high
9990 The size, low-, and high-water marks for the IP cache.
9993 NAME: fqdncache_size
9994 COMMENT: (number of entries)
9997 LOC: Config.fqdncache.size
9999 Maximum number of FQDN cache entries.
10004 -----------------------------------------------------------------------------
10007 NAME: configuration_includes_quoted_values
10009 TYPE: configuration_includes_quoted_values
10011 LOC: ConfigParser::RecognizeQuotedValues
10013 If set, Squid will recognize each "quoted string" after a configuration
10014 directive as a single parameter. The quotes are stripped before the
10015 parameter value is interpreted or used.
10016 See "Values with spaces, quotes, and other special characters"
10017 section for more details.
10024 LOC: Config.onoff.mem_pools
10026 If set, Squid will keep pools of allocated (but unused) memory
10027 available for future use. If memory is a premium on your
10028 system and you believe your malloc library outperforms Squid
10029 routines, disable this.
10032 NAME: memory_pools_limit
10036 LOC: Config.MemPools.limit
10038 Used only with memory_pools on:
10039 memory_pools_limit 50 MB
10041 If set to a non-zero value, Squid will keep at most the specified
10042 limit of allocated (but unused) memory in memory pools. All free()
10043 requests that exceed this limit will be handled by your malloc
10044 library. Squid does not pre-allocate any memory, just safe-keeps
10045 objects that otherwise would be free()d. Thus, it is safe to set
10046 memory_pools_limit to a reasonably high value even if your
10047 configuration will use less memory.
10049 If set to none, Squid will keep all memory it can. That is, there
10050 will be no limit on the total amount of memory used for safe-keeping.
10052 To disable memory allocation optimization, do not set
10053 memory_pools_limit to 0 or none. Set memory_pools to "off" instead.
10055 An overhead for maintaining memory pools is not taken into account
10056 when the limit is checked. This overhead is close to four bytes per
10057 object kept. However, pools may actually _save_ memory because of
10058 reduced memory thrashing in your malloc library.
10061 NAME: forwarded_for
10062 COMMENT: on|off|transparent|truncate|delete
10065 LOC: opt_forwarded_for
10067 If set to "on", Squid will append your client's IP address
10068 in the HTTP requests it forwards. By default it looks like:
10070 X-Forwarded-For: 192.1.2.3
10072 If set to "off", it will appear as
10074 X-Forwarded-For: unknown
10076 If set to "transparent", Squid will not alter the
10077 X-Forwarded-For header in any way.
10079 If set to "delete", Squid will delete the entire
10080 X-Forwarded-For header.
10082 If set to "truncate", Squid will remove all existing
10083 X-Forwarded-For entries, and place the client IP as the sole entry.
10086 NAME: cachemgr_passwd
10087 TYPE: cachemgrpasswd
10089 DEFAULT_DOC: No password. Actions which require password are denied.
10090 LOC: Config.passwd_list
10092 Specify passwords for cachemgr operations.
10094 Usage: cachemgr_passwd password action action ...
10096 Some valid actions are (see cache manager menu for a full list):
10136 * Indicates actions which will not be performed without a
10137 valid password, others can be performed if not listed here.
10139 To disable an action, set the password to "disable".
10140 To allow performing an action without a password, set the
10141 password to "none".
10143 Use the keyword "all" to set the same password for all actions.
10146 cachemgr_passwd secret shutdown
10147 cachemgr_passwd lesssssssecret info stats/objects
10148 cachemgr_passwd disable all
10155 LOC: Config.onoff.client_db
10157 If you want to disable collecting per-client statistics,
10158 turn off client_db here.
10161 NAME: refresh_all_ims
10165 LOC: Config.onoff.refresh_all_ims
10167 When you enable this option, squid will always check
10168 the origin server for an update when a client sends an
10169 If-Modified-Since request. Many browsers use IMS
10170 requests when the user requests a reload, and this
10171 ensures those clients receive the latest version.
10173 By default (off), squid may return a Not Modified response
10174 based on the age of the cached version.
10177 NAME: reload_into_ims
10178 IFDEF: USE_HTTP_VIOLATIONS
10182 LOC: Config.onoff.reload_into_ims
10184 When you enable this option, client no-cache or ``reload''
10185 requests will be changed to If-Modified-Since requests.
10186 Doing this VIOLATES the HTTP standard. Enabling this
10187 feature could make you liable for problems which it
10190 see also refresh_pattern for a more selective approach.
10193 NAME: connect_retries
10195 LOC: Config.connect_retries
10197 DEFAULT_DOC: Do not retry failed connections.
10199 Limits the number of reopening attempts when establishing a single
10200 TCP connection. All these attempts must still complete before the
10201 applicable connection opening timeout expires.
10203 By default and when connect_retries is set to zero, Squid does not
10204 retry failed connection opening attempts.
10206 The (not recommended) maximum is 10 tries. An attempt to configure a
10207 higher value results in the value of 10 being used (with a warning).
10209 Squid may open connections to retry various high-level forwarding
10210 failures. For an outside observer, that activity may look like a
10211 low-level connection reopening attempt, but those high-level retries
10212 are governed by forward_max_tries instead.
10214 See also: connect_timeout, forward_timeout, icap_connect_timeout,
10215 ident_timeout, and forward_max_tries.
10218 NAME: retry_on_error
10220 LOC: Config.retry.onerror
10223 If set to ON Squid will automatically retry requests when
10224 receiving an error response with status 403 (Forbidden),
10225 500 (Internal Error), 501 or 503 (Service not available).
10226 Status 502 and 504 (Gateway errors) are always retried.
10228 This is mainly useful if you are in a complex cache hierarchy to
10229 work around access control errors.
10231 NOTE: This retry will attempt to find another working destination.
10232 Which is different from the server which just failed.
10235 NAME: as_whois_server
10237 LOC: Config.as_whois_server
10238 DEFAULT: whois.ra.net
10240 WHOIS server to query for AS numbers. NOTE: AS numbers are
10241 queried only when Squid starts up, not for every request.
10246 LOC: Config.onoff.offline
10249 Enable this option and Squid will never try to validate cached
10253 NAME: uri_whitespace
10254 TYPE: uri_whitespace
10255 LOC: Config.uri_whitespace
10258 What to do with requests that have whitespace characters in the
10261 strip: The whitespace characters are stripped out of the URL.
10262 This is the behavior recommended by RFC2396 and RFC3986
10263 for tolerant handling of generic URI.
10264 NOTE: This is one difference between generic URI and HTTP URLs.
10266 deny: The request is denied. The user receives an "Invalid
10268 This is the behaviour recommended by RFC2616 for safe
10269 handling of HTTP request URL.
10271 allow: The request is allowed and the URI is not changed. The
10272 whitespace characters remain in the URI. Note the
10273 whitespace is passed to redirector processes if they
10275 Note this may be considered a violation of RFC2616
10276 request parsing where whitespace is prohibited in the
10279 encode: The request is allowed and the whitespace characters are
10280 encoded according to RFC1738.
10282 chop: The request is allowed and the URI is chopped at the
10286 NOTE the current Squid implementation of encode and chop violates
10287 RFC2616 by not using a 301 redirect after altering the URL.
10292 LOC: Config.chroot_dir
10295 Specifies a directory where Squid should do a chroot() while
10296 initializing. This also causes Squid to fully drop root
10297 privileges after initializing. This means, for example, if you
10298 use a HTTP port less than 1024 and try to reconfigure, you may
10299 get an error saying that Squid can not open the port.
10302 NAME: pipeline_prefetch
10303 TYPE: pipelinePrefetch
10304 LOC: Config.pipeline_max_prefetch
10306 DEFAULT_DOC: Do not pre-parse pipelined requests.
10308 HTTP clients may send a pipeline of 1+N requests to Squid using a
10309 single connection, without waiting for Squid to respond to the first
10310 of those requests. This option limits the number of concurrent
10311 requests Squid will try to handle in parallel. If set to N, Squid
10312 will try to receive and process up to 1+N requests on the same
10313 connection concurrently.
10315 Defaults to 0 (off) for bandwidth management and access logging
10318 NOTE: pipelining requires persistent connections to clients.
10320 WARNING: pipelining breaks NTLM and Negotiate/Kerberos authentication.
10323 NAME: high_response_time_warning
10326 LOC: Config.warnings.high_rptm
10328 DEFAULT_DOC: disabled.
10330 If the one-minute median response time exceeds this value,
10331 Squid prints a WARNING with debug level 0 to get the
10332 administrators attention. The value is in milliseconds.
10335 NAME: high_page_fault_warning
10337 LOC: Config.warnings.high_pf
10339 DEFAULT_DOC: disabled.
10341 If the one-minute average page fault rate exceeds this
10342 value, Squid prints a WARNING with debug level 0 to get
10343 the administrators attention. The value is in page faults
10347 NAME: high_memory_warning
10349 LOC: Config.warnings.high_memory
10350 IFDEF: HAVE_MSTATS&&HAVE_GNUMALLOC_H
10352 DEFAULT_DOC: disabled.
10354 If the memory usage (as determined by gnumalloc, if available and used)
10355 exceeds this amount, Squid prints a WARNING with debug level 0 to get
10356 the administrators attention.
10358 # TODO: link high_memory_warning to mempools?
10360 NAME: sleep_after_fork
10361 COMMENT: (microseconds)
10363 LOC: Config.sleep_after_fork
10366 When this is set to a non-zero value, the main Squid process
10367 sleeps the specified number of microseconds after a fork()
10368 system call. This sleep may help the situation where your
10369 system reports fork() failures due to lack of (virtual)
10370 memory. Note, however, if you have a lot of child
10371 processes, these sleep delays will add up and your
10372 Squid will not service requests for some amount of time
10373 until all the child processes have been started.
10374 On Windows value less then 1000 (1 milliseconds) are
10378 NAME: windows_ipaddrchangemonitor
10379 IFDEF: _SQUID_WINDOWS_
10383 LOC: Config.onoff.WIN32_IpAddrChangeMonitor
10385 On Windows Squid by default will monitor IP address changes and will
10386 reconfigure itself after any detected event. This is very useful for
10387 proxies connected to internet with dial-up interfaces.
10388 In some cases (a Proxy server acting as VPN gateway is one) it could be
10389 desiderable to disable this behaviour setting this to 'off'.
10390 Note: after changing this, Squid service must be restarted.
10395 IFDEF: USE_SQUID_EUI
10397 LOC: Eui::TheConfig.euiLookup
10399 Whether to lookup the EUI or MAC address of a connected client.
10402 NAME: max_filedescriptors max_filedesc
10405 DEFAULT_DOC: Use operating system soft limit set by ulimit.
10406 LOC: Config.max_filedescriptors
10408 Set the maximum number of filedescriptors, either below the
10409 operating system default or up to the hard limit.
10411 Remove from squid.conf to inherit the current ulimit soft
10414 Note: Changing this requires a restart of Squid. Also
10415 not all I/O types supports large values (eg on Windows).
10418 NAME: force_request_body_continuation
10420 LOC: Config.accessList.forceRequestBodyContinuation
10422 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
10424 This option controls how Squid handles data upload requests from HTTP
10425 and FTP agents that require a "Please Continue" control message response
10426 to actually send the request body to Squid. It is mostly useful in
10427 adaptation environments.
10429 When Squid receives an HTTP request with an "Expect: 100-continue"
10430 header or an FTP upload command (e.g., STOR), Squid normally sends the
10431 request headers or FTP command information to an adaptation service (or
10432 peer) and waits for a response. Most adaptation services (and some
10433 broken peers) may not respond to Squid at that stage because they may
10434 decide to wait for the HTTP request body or FTP data transfer. However,
10435 that request body or data transfer may never come because Squid has not
10436 responded with the HTTP 100 or FTP 150 (Please Continue) control message
10437 to the request sender yet!
10439 An allow match tells Squid to respond with the HTTP 100 or FTP 150
10440 (Please Continue) control message on its own, before forwarding the
10441 request to an adaptation service or peer. Such a response usually forces
10442 the request sender to proceed with sending the body. A deny match tells
10443 Squid to delay that control response until the origin server confirms
10444 that the request body is needed. Delaying is the default behavior.
10447 NAME: server_pconn_for_nonretriable
10450 DEFAULT_DOC: Open new connections for forwarding requests Squid cannot retry safely.
10451 LOC: Config.accessList.serverPconnForNonretriable
10453 This option provides fine-grained control over persistent connection
10454 reuse when forwarding HTTP requests that Squid cannot retry. It is useful
10455 in environments where opening new connections is very expensive
10456 (e.g., all connections are secured with TLS with complex client and server
10457 certificate validation) and race conditions associated with persistent
10458 connections are very rare and/or only cause minor problems.
10460 HTTP prohibits retrying unsafe and non-idempotent requests (e.g., POST).
10461 Squid limitations also prohibit retrying all requests with bodies (e.g., PUT).
10462 By default, when forwarding such "risky" requests, Squid opens a new
10463 connection to the server or cache_peer, even if there is an idle persistent
10464 connection available. When Squid is configured to risk sending a non-retriable
10465 request on a previously used persistent connection, and the server closes
10466 the connection before seeing that risky request, the user gets an error response
10467 from Squid. In most cases, that error response will be HTTP 502 (Bad Gateway)
10468 with ERR_ZERO_SIZE_OBJECT or ERR_WRITE_ERROR (peer connection reset) error detail.
10470 If an allow rule matches, Squid reuses an available idle persistent connection
10471 (if any) for the request that Squid cannot retry. If a deny rule matches, then
10472 Squid opens a new connection for the request that Squid cannot retry.
10474 This option does not affect requests that Squid can retry. They will reuse idle
10475 persistent connections (if any).
10477 This clause only supports fast acl types.
10478 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
10481 acl SpeedIsWorthTheRisk method POST
10482 server_pconn_for_nonretriable allow SpeedIsWorthTheRisk
10485 NAME: happy_eyeballs_connect_timeout
10489 LOC: Config.happyEyeballs.connect_timeout
10491 This Happy Eyeballs (RFC 8305) tuning directive specifies the minimum
10492 delay between opening a primary to-server connection and opening a
10493 spare to-server connection for the same master transaction. This delay
10494 is similar to the Connection Attempt Delay in RFC 8305, but it is only
10495 applied to the first spare connection attempt. Subsequent spare
10496 connection attempts use happy_eyeballs_connect_gap, and primary
10497 connection attempts are not artificially delayed at all.
10499 Terminology: The "primary" and "spare" designations are determined by
10500 the order of DNS answers received by Squid: If Squid DNS AAAA query
10501 was answered first, then primary connections are connections to IPv6
10502 peer addresses (while spare connections use IPv4 addresses).
10503 Similarly, if Squid DNS A query was answered first, then primary
10504 connections are connections to IPv4 peer addresses (while spare
10505 connections use IPv6 addresses).
10507 Shorter happy_eyeballs_connect_timeout values reduce master
10508 transaction response time, potentially improving user-perceived
10509 response times (i.e., making user eyeballs happier). Longer delays
10510 reduce both concurrent connection level and server bombardment with
10511 connection requests, potentially improving overall Squid performance
10512 and reducing the chance of being blocked by servers for opening too
10513 many unused connections.
10515 RFC 8305 prohibits happy_eyeballs_connect_timeout values smaller than
10516 10 (milliseconds) to "avoid congestion collapse in the presence of
10517 high packet-loss rates".
10519 The following Happy Eyeballs directives place additional connection
10520 opening restrictions: happy_eyeballs_connect_gap and
10521 happy_eyeballs_connect_limit.
10524 NAME: happy_eyeballs_connect_gap
10528 DEFAULT_DOC: no artificial delays between spare attempts
10529 LOC: Config.happyEyeballs.connect_gap
10531 This Happy Eyeballs (RFC 8305) tuning directive specifies the
10532 minimum delay between opening spare to-server connections (to any
10533 server; i.e. across all concurrent master transactions in a Squid
10534 instance). Each SMP worker currently multiplies the configured gap
10535 by the total number of workers so that the combined spare connection
10536 opening rate of a Squid instance obeys the configured limit. The
10537 workers do not coordinate connection openings yet; a micro burst
10538 of spare connection openings may violate the configured gap.
10540 This directive has similar trade-offs as
10541 happy_eyeballs_connect_timeout, but its focus is on limiting traffic
10542 amplification effects for Squid as a whole, while
10543 happy_eyeballs_connect_timeout works on an individual master
10546 The following Happy Eyeballs directives place additional connection
10547 opening restrictions: happy_eyeballs_connect_timeout and
10548 happy_eyeballs_connect_limit. See the former for related terminology.
10551 NAME: happy_eyeballs_connect_limit
10554 DEFAULT_DOC: no artificial limit on the number of concurrent spare attempts
10555 LOC: Config.happyEyeballs.connect_limit
10557 This Happy Eyeballs (RFC 8305) tuning directive specifies the
10558 maximum number of spare to-server connections (to any server; i.e.
10559 across all concurrent master transactions in a Squid instance).
10560 Each SMP worker gets an equal share of the total limit. However,
10561 the workers do not share the actual connection counts yet, so one
10562 (busier) worker cannot "borrow" spare connection slots from another
10563 (less loaded) worker.
10565 Setting this limit to zero disables concurrent use of primary and
10566 spare TCP connections: Spare connection attempts are made only after
10567 all primary attempts fail. However, Squid would still use the
10568 DNS-related optimizations of the Happy Eyeballs approach.
10570 This directive has similar trade-offs as happy_eyeballs_connect_gap,
10571 but its focus is on limiting Squid overheads, while
10572 happy_eyeballs_connect_gap focuses on the origin server and peer
10575 The following Happy Eyeballs directives place additional connection
10576 opening restrictions: happy_eyeballs_connect_timeout and
10577 happy_eyeballs_connect_gap. See the former for related terminology.