1 ## Copyright (C) 1996-2021 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 time units
60 Time-related directives marked with either "time-units" or
61 "time-units-small" accept a time unit. The supported time units are:
63 nanosecond (time-units-small only)
64 microsecond (time-units-small only)
73 year - 31557790080 milliseconds (just over 365 days)
76 Values with spaces, quotes, and other special characters
78 Squid supports directive parameters with spaces, quotes, and other
79 special characters. Surround such parameters with "double quotes". Use
80 the configuration_includes_quoted_values directive to enable or
83 Squid supports reading configuration option parameters from external
84 files using the syntax:
85 parameters("/path/filename")
87 acl allowlist dstdomain parameters("/etc/squid/allowlist.txt")
89 Conditional configuration
91 If-statements can be used to make configuration directives
95 ... regular configuration directives ...
97 ... regular configuration directives ...]
100 The else part is optional. The keywords "if", "else", and "endif"
101 must be typed on their own lines, as if they were regular
102 configuration directives.
104 NOTE: An else-if condition is not supported.
106 These individual conditions types are supported:
109 Always evaluates to true.
111 Always evaluates to false.
112 <integer> = <integer>
113 Equality comparison of two integer numbers.
118 The following SMP-related preprocessor macros can be used.
120 ${process_name} expands to the current Squid process "name"
121 (e.g., squid1, squid2, or cache1).
123 ${process_number} expands to the current Squid process
124 identifier, which is an integer number (e.g., 1, 2, 3) unique
125 across all Squid processes of the current service instance.
127 ${service_name} expands into the current Squid service instance
128 name identifier which is provided by -n on the command line.
132 Logformat macros can be used in many places outside of the logformat
133 directive. In theory, all of the logformat codes can be used as %macros,
134 where they are supported. In practice, a %macro expands as a dash (-) when
135 the transaction does not yet have enough information and a value is needed.
137 There is no definitive list of what tokens are available at the various
138 stages of the transaction.
140 And some information may already be available to Squid but not yet
141 committed where the macro expansion code can access it (report
142 such instances!). The macro will be expanded into a single dash
143 ('-') in such cases. Not all macros have been tested.
147 # options still not yet ported from 2.7 to 3.x
148 NAME: broken_vary_encoding
151 This option is not yet supported by Squid-3.
157 This option is not yet supported by Squid-3.
163 This option is not yet supported by Squid-3.
166 NAME: external_refresh_check
169 This option is not yet supported by Squid-3.
172 NAME: location_rewrite_program location_rewrite_access location_rewrite_children location_rewrite_concurrency
175 This option is not yet supported by Squid-3.
178 NAME: refresh_stale_hit
181 This option is not yet supported by Squid-3.
184 # Options removed in 5.x
188 Remove this line. Squid no longer supports preferential treatment of DNS A records.
191 # Options removed in 4.x
192 NAME: cache_peer_domain cache_host_domain
195 Replace with dstdomain ACLs and cache_peer_access.
201 Remove this line. The behaviour enabled by this is no longer needed.
204 NAME: sslproxy_cafile
207 Remove this line. Use tls_outgoing_options cafile= instead.
210 NAME: sslproxy_capath
213 Remove this line. Use tls_outgoing_options capath= instead.
216 NAME: sslproxy_cipher
219 Remove this line. Use tls_outgoing_options cipher= instead.
222 NAME: sslproxy_client_certificate
225 Remove this line. Use tls_outgoing_options cert= instead.
228 NAME: sslproxy_client_key
231 Remove this line. Use tls_outgoing_options key= instead.
237 Remove this line. Use tls_outgoing_options flags= instead.
240 NAME: sslproxy_options
243 Remove this line. Use tls_outgoing_options options= instead.
246 NAME: sslproxy_version
249 Remove this line. Use tls_outgoing_options options= instead.
252 # Options removed in 3.5
253 NAME: hierarchy_stoplist
256 Remove this line. Use always_direct or cache_peer_access ACLs instead if you need to prevent cache_peer use.
259 # Options removed in 3.4
263 Remove this line. Use acls with access_log directives to control access logging
269 Remove this line. Use acls with icap_log directives to control icap logging
272 # Options Removed in 3.3
273 NAME: ignore_ims_on_miss
276 Remove this line. The HTTP/1.1 feature is now configured by 'cache_miss_revalidate'.
279 # Options Removed in 3.2
280 NAME: balance_on_multiple_ip
283 Remove this line. Squid performs a 'Happy Eyeballs' algorithm, this multiple-IP algorithm is not longer relevant.
286 NAME: chunked_request_body_max_size
289 Remove this line. Squid is now HTTP/1.1 compliant.
292 NAME: dns_v4_fallback
295 Remove this line. Squid performs a 'Happy Eyeballs' algorithm, the 'fallback' algorithm is no longer relevant.
298 NAME: emulate_httpd_log
301 Replace this with an access_log directive using the format 'common' or 'combined'.
307 Use a regular access.log with ACL limiting it to MISS events.
313 Remove this line. Configure FTP page display using the CSS controls in errorpages.css instead.
316 NAME: ignore_expect_100
319 Remove this line. The HTTP/1.1 feature is now fully supported by default.
325 Remove this option from your config. To log FQDN use %>A in the log format.
328 NAME: log_ip_on_direct
331 Remove this option from your config. To log server or peer names use %<A in the log format.
334 NAME: maximum_single_addr_tries
337 Replaced by connect_retries. The behaviour has changed, please read the documentation before altering.
340 NAME: referer_log referrer_log
343 Replace this with an access_log directive using the format 'referrer'.
349 Remove this line. The feature is supported by default in storage types where update is implemented.
352 NAME: url_rewrite_concurrency
355 Remove this line. Set the 'concurrency=' option of url_rewrite_children instead.
361 Replace this with an access_log directive using the format 'useragent'.
364 # Options Removed in 3.1
368 Remove this line. DNS is no longer tested on startup.
371 NAME: extension_methods
374 Remove this line. All valid methods for HTTP are accepted by default.
377 # 2.7 Options Removed/Replaced in 3.2
382 # 2.7 Options Removed/Replaced in 3.1
390 Remove this line. HTTP/1.1 is supported by default.
393 NAME: upgrade_http0.9
396 Remove this line. ICY/1.0 streaming protocol is supported by default.
399 NAME: zph_local zph_mode zph_option zph_parent zph_sibling
402 Alter these entries. Use the qos_flows directive instead.
405 # Options Removed in 3.0
409 Since squid-3.0 replace with request_header_access or reply_header_access
410 depending on whether you wish to match client requests or server replies.
413 NAME: httpd_accel_no_pmtu_disc
416 Since squid-3.0 use the 'disable-pmtu-discovery' flag on http_port instead.
419 NAME: wais_relay_host
422 Replace this line with 'cache_peer' configuration.
425 NAME: wais_relay_port
428 Replace this line with 'cache_peer' configuration.
433 -----------------------------------------------------------------------------
440 DEFAULT_DOC: SMP support disabled.
442 Number of main Squid processes or "workers" to fork and maintain.
443 0: "no daemon" mode, like running "squid -N ..."
444 1: "no SMP" mode, start one main Squid process daemon (default)
445 N: start N main Squid process daemons (i.e., SMP mode)
447 In SMP mode, each worker does nearly all what a single Squid daemon
448 does (e.g., listen on http_port and forward HTTP requests).
451 NAME: cpu_affinity_map
453 LOC: Config.cpuAffinityMap
455 DEFAULT_DOC: Let operating system decide.
457 Usage: cpu_affinity_map process_numbers=P1,P2,... cores=C1,C2,...
459 Sets 1:1 mapping between Squid processes and CPU cores. For example,
461 cpu_affinity_map process_numbers=1,2,3,4 cores=1,3,5,7
463 affects processes 1 through 4 only and places them on the first
464 four even cores, starting with core #1.
466 CPU cores are numbered starting from 1. Requires support for
467 sched_getaffinity(2) and sched_setaffinity(2) system calls.
469 Multiple cpu_affinity_map options are merged.
474 NAME: shared_memory_locking
477 LOC: Config.shmLocking
480 Whether to ensure that all required shared memory is available by
481 "locking" that shared memory into RAM when Squid starts. The
482 alternative is faster startup time followed by slightly slower
483 performance and, if not enough RAM is actually available during
484 runtime, mysterious crashes.
486 SMP Squid uses many shared memory segments. These segments are
487 brought into Squid memory space using an mmap(2) system call. During
488 Squid startup, the mmap() call often succeeds regardless of whether
489 the system has enough RAM. In general, Squid cannot tell whether the
490 kernel applies this "optimistic" memory allocation policy (but
491 popular modern kernels usually use it).
493 Later, if Squid attempts to actually access the mapped memory
494 regions beyond what the kernel is willing to allocate, the
495 "optimistic" kernel simply kills Squid kid with a SIGBUS signal.
496 Some of the memory limits enforced by the kernel are currently
497 poorly understood: We do not know how to detect and check them. This
498 option ensures that the mapped memory will be available.
500 This option may have a positive performance side-effect: Locking
501 memory at start avoids runtime paging I/O. Paging slows Squid down.
503 Locking memory may require a large enough RLIMIT_MEMLOCK OS limit,
504 CAP_IPC_LOCK capability, or equivalent.
507 NAME: hopeless_kid_revival_delay
510 LOC: Config.hopelessKidRevivalDelay
513 Normally, when a kid process dies, Squid immediately restarts the
514 kid. A kid experiencing frequent deaths is marked as "hopeless" for
515 the duration specified by this directive. Hopeless kids are not
516 automatically restarted.
518 Currently, zero values are not supported because they result in
519 misconfigured SMP Squid instances running forever, endlessly
520 restarting each dying kid. To effectively disable hopeless kids
521 revival, set the delay to a huge value (e.g., 1 year).
523 Reconfiguration also clears all hopeless kids designations, allowing
524 for manual revival of hopeless kids.
528 OPTIONS FOR AUTHENTICATION
529 -----------------------------------------------------------------------------
535 LOC: Auth::TheConfig.schemes
538 This is used to define parameters for the various authentication
539 schemes supported by Squid.
541 format: auth_param scheme parameter [setting]
543 The order in which authentication schemes are presented to the client is
544 dependent on the order the scheme first appears in config file. IE
545 has a bug (it's not RFC 2617 compliant) in that it will use the basic
546 scheme if basic is the first entry presented, even if more secure
547 schemes are presented. For now use the order in the recommended
548 settings section below. If other browsers have difficulties (don't
549 recognize the schemes offered even if you are using basic) either
550 put basic first, or disable the other schemes (by commenting out their
553 Once an authentication scheme is fully configured, it can only be
554 shutdown by shutting squid down and restarting. Changes can be made on
555 the fly and activated with a reconfigure. I.E. You can change to a
556 different helper, but not unconfigure the helper completely.
558 Please note that while this directive defines how Squid processes
559 authentication it does not automatically activate authentication.
560 To use authentication you must in addition make use of ACLs based
561 on login name in http_access (proxy_auth, proxy_auth_regex or
562 external with %LOGIN used in the format tag). The browser will be
563 challenged for authentication on the first such acl encountered
564 in http_access processing and will also be re-challenged for new
565 login credentials if the request is being denied by a proxy_auth
568 WARNING: authentication can't be used in a transparently intercepting
569 proxy as the client then thinks it is talking to an origin server and
570 not the proxy. This is a limitation of bending the TCP/IP protocol to
571 transparently intercepting port 80, not a limitation in Squid.
572 Ports flagged 'transparent', 'intercept', or 'tproxy' have
573 authentication disabled.
575 === Parameters common to all schemes. ===
578 Specifies the command for the external authenticator.
580 By default, each authentication scheme is not used unless a
581 program is specified.
583 See http://wiki.squid-cache.org/Features/AddonHelpers for
584 more details on helper operations and creating your own.
587 Specifies a string to be append to request line format for
588 the authentication helper. "Quoted" format values may contain
589 spaces and logformat %macros. In theory, any logformat %macro
590 can be used. In practice, a %macro expands as a dash (-) if
591 the helper request is sent before the required macro
592 information is available to Squid.
594 By default, Squid uses request formats provided in
595 scheme-specific examples below (search for %credentials).
597 The expanded key_extras value is added to the Squid credentials
598 cache and, hence, will affect authentication. It can be used to
599 authenticate different users with identical user names (e.g.,
600 when user authentication depends on http_port).
602 Avoid adding frequently changing information to key_extras. For
603 example, if you add user source IP, and it changes frequently
604 in your environment, then max_user_ip ACL is going to treat
605 every user+IP combination as a unique "user", breaking the ACL
606 and wasting a lot of memory on those user records. It will also
607 force users to authenticate from scratch whenever their IP
611 Specifies the protection scope (aka realm name) which is to be
612 reported to the client for the authentication scheme. It is
613 commonly part of the text the user will see when prompted for
614 their username and password.
616 For Basic the default is "Squid proxy-caching web server".
617 For Digest there is no default, this parameter is mandatory.
618 For NTLM and Negotiate this parameter is ignored.
620 "children" numberofchildren [startup=N] [idle=N] [concurrency=N]
621 [queue-size=N] [on-persistent-overload=action]
622 [reservation-timeout=seconds]
624 The maximum number of authenticator processes to spawn. If
625 you start too few Squid will have to wait for them to process
626 a backlog of credential verifications, slowing it down. When
627 password verifications are done via a (slow) network you are
628 likely to need lots of authenticator processes.
630 The startup= and idle= options permit some skew in the exact
631 amount run. A minimum of startup=N will begin during startup
632 and reconfigure. Squid will start more in groups of up to
633 idle=N in an attempt to meet traffic needs and to keep idle=N
634 free above those traffic needs up to the maximum.
636 The concurrency= option sets the number of concurrent requests
637 the helper can process. The default of 0 is used for helpers
638 who only supports one request at a time. Setting this to a
639 number greater than 0 changes the protocol used to include a
640 channel ID field first on the request/response line, allowing
641 multiple requests to be sent to the same helper in parallel
642 without waiting for the response.
644 Concurrency must not be set unless it's known the helper
645 supports the input format with channel-ID fields.
647 The queue-size option sets the maximum number of queued
648 requests. A request is queued when no existing child can
649 accept it due to concurrency limit and no new child can be
650 started due to numberofchildren limit. The default maximum is
651 2*numberofchildren. Squid is allowed to temporarily exceed the
652 configured maximum, marking the affected helper as
653 "overloaded". If the helper overload lasts more than 3
654 minutes, the action prescribed by the on-persistent-overload
657 The on-persistent-overload=action option specifies Squid
658 reaction to a new helper request arriving when the helper
659 has been overloaded for more that 3 minutes already. The number
660 of queued requests determines whether the helper is overloaded
661 (see the queue-size option).
663 Two actions are supported:
665 die Squid worker quits. This is the default behavior.
667 ERR Squid treats the helper request as if it was
668 immediately submitted, and the helper immediately
669 replied with an ERR response. This action has no effect
670 on the already queued and in-progress helper requests.
672 NOTE: NTLM and Negotiate schemes do not support concurrency
673 in the Squid code module even though some helpers can.
675 The reservation-timeout=seconds option allows NTLM and Negotiate
676 helpers to forget about clients that abandon their in-progress
677 connection authentication without closing the connection. The
678 timeout is measured since the last helper response received by
679 Squid for the client. Fractional seconds are not supported.
681 After the timeout, the helper will be used for other clients if
682 there are no unreserved helpers available. In the latter case,
683 the old client attempt to resume authentication will not be
684 forwarded to the helper (and the client should open a new HTTP
685 connection and retry authentication from scratch).
687 By default, reservations do not expire and clients that keep
688 their connections open without completing authentication may
689 exhaust all NTLM and Negotiate helpers.
692 If you experience problems with PUT/POST requests when using
693 the NTLM or Negotiate schemes then you can try setting this
694 to off. This will cause Squid to forcibly close the connection
695 on the initial request where the browser asks which schemes
696 are supported by the proxy.
698 For Basic and Digest this parameter is ignored.
701 Useful for sending credentials to authentication backends that
702 expect UTF-8 encoding (e.g., LDAP).
704 When this option is enabled, Squid uses HTTP Accept-Language
705 request header to guess the received credentials encoding
706 (ISO-Latin-1, CP1251, or UTF-8) and then converts the first
707 two encodings into UTF-8.
709 When this option is disabled and by default, Squid sends
710 credentials in their original (i.e. received) encoding.
712 This parameter is only honored for Basic and Digest schemes.
713 For Basic, the entire username:password credentials are
714 checked and, if necessary, re-encoded. For Digest -- just the
715 username component. For NTLM and Negotiate schemes, this
716 parameter is ignored.
718 IF HAVE_AUTH_MODULE_BASIC
719 === Basic authentication parameters ===
721 "credentialsttl" timetolive
722 Specifies how long squid assumes an externally validated
723 username:password pair is valid for - in other words how
724 often the helper program is called for that user. Set this
725 low to force revalidation with short lived passwords.
727 NOTE: setting this high does not impact your susceptibility
728 to replay attacks unless you are using an one-time password
729 system (such as SecureID). If you are using such a system,
730 you will be vulnerable to replay attacks unless you also
731 use the max_user_ip ACL in an http_access rule.
733 "casesensitive" on|off
734 Specifies if usernames are case sensitive. Most user databases
735 are case insensitive allowing the same username to be spelled
736 using both lower and upper case letters, but some are case
737 sensitive. This makes a big difference for user_max_ip ACL
738 processing and similar.
741 IF HAVE_AUTH_MODULE_DIGEST
742 === Digest authentication parameters ===
744 "nonce_garbage_interval" timeinterval
745 Specifies the interval that nonces that have been issued
746 to client_agent's are checked for validity.
748 "nonce_max_duration" timeinterval
749 Specifies the maximum length of time a given nonce will be
752 "nonce_max_count" number
753 Specifies the maximum number of times a given nonce can be
756 "nonce_strictness" on|off
757 Determines if squid requires strict increment-by-1 behavior
758 for nonce counts, or just incrementing (off - for use when
759 user agents generate nonce counts that occasionally miss 1
760 (ie, 1,2,4,6)). Default off.
762 "check_nonce_count" on|off
763 This directive if set to off can disable the nonce count check
764 completely to work around buggy digest qop implementations in
765 certain mainstream browser versions. Default on to check the
766 nonce count to protect from authentication replay attacks.
768 "post_workaround" on|off
769 This is a workaround to certain buggy browsers who send an
770 incorrect request digest in POST requests when reusing the
771 same nonce as acquired earlier on a GET request.
775 === Example Configuration ===
777 This configuration displays the recommended authentication scheme
778 order from most to least secure with recommended minimum configuration
779 settings for each scheme:
781 #auth_param negotiate program <uncomment and complete this line to activate>
782 #auth_param negotiate children 20 startup=0 idle=1
784 #auth_param digest program <uncomment and complete this line to activate>
785 #auth_param digest children 20 startup=0 idle=1
786 #auth_param digest realm Squid proxy-caching web server
787 #auth_param digest nonce_garbage_interval 5 minutes
788 #auth_param digest nonce_max_duration 30 minutes
789 #auth_param digest nonce_max_count 50
791 #auth_param ntlm program <uncomment and complete this line to activate>
792 #auth_param ntlm children 20 startup=0 idle=1
794 #auth_param basic program <uncomment and complete this line>
795 #auth_param basic children 5 startup=5 idle=1
796 #auth_param basic credentialsttl 2 hours
799 NAME: authenticate_cache_garbage_interval
803 LOC: Auth::TheConfig.garbageCollectInterval
805 The time period between garbage collection across the username cache.
806 This is a trade-off between memory utilization (long intervals - say
807 2 days) and CPU (short intervals - say 1 minute). Only change if you
811 NAME: authenticate_ttl
815 LOC: Auth::TheConfig.credentialsTtl
817 The time a user & their credentials stay in the logged in
818 user cache since their last request. When the garbage
819 interval passes, all user credentials that have passed their
820 TTL are removed from memory.
823 NAME: authenticate_ip_ttl
826 LOC: Auth::TheConfig.ipTtl
829 If you use proxy authentication and the 'max_user_ip' ACL,
830 this directive controls how long Squid remembers the IP
831 addresses associated with each user. Use a small value
832 (e.g., 60 seconds) if your users might change addresses
833 quickly, as is the case with dialup. You might be safe
834 using a larger value (e.g., 2 hours) in a corporate LAN
835 environment with relatively static address assignments.
840 -----------------------------------------------------------------------------
843 NAME: external_acl_type
844 TYPE: externalAclHelper
845 LOC: Config.externalAclHelperList
848 This option defines external acl classes using a helper program
849 to look up the status
851 external_acl_type name [options] FORMAT /path/to/helper [helper arguments]
855 ttl=n TTL in seconds for cached results (defaults to 3600
859 TTL for cached negative lookups (default same
862 grace=n Percentage remaining of TTL where a refresh of a
863 cached entry should be initiated without needing to
864 wait for a new reply. (default is for no grace period)
866 cache=n The maximum number of entries in the result cache. The
867 default limit is 262144 entries. Each cache entry usually
868 consumes at least 256 bytes. Squid currently does not remove
869 expired cache entries until the limit is reached, so a proxy
870 will sooner or later reach the limit. The expanded FORMAT
871 value is used as the cache key, so if the details in FORMAT
872 are highly variable, a larger cache may be needed to produce
873 reduction in helper load.
876 Maximum number of acl helper processes spawned to service
877 external acl lookups of this type. (default 5)
880 Minimum number of acl helper processes to spawn during
881 startup and reconfigure to service external acl lookups
882 of this type. (default 0)
885 Number of acl helper processes to keep ahead of traffic
886 loads. Squid will spawn this many at once whenever load
887 rises above the capabilities of existing processes.
888 Up to the value of children-max. (default 1)
890 concurrency=n concurrency level per process. Only used with helpers
891 capable of processing more than one query at a time.
893 queue-size=N The queue-size option sets the maximum number of
894 queued requests. A request is queued when no existing
895 helper can accept it due to concurrency limit and no
896 new helper can be started due to children-max limit.
897 If the queued requests exceed queue size, the acl is
898 ignored. The default value is set to 2*children-max.
900 protocol=2.5 Compatibility mode for Squid-2.5 external acl helpers.
902 ipv4 / ipv6 IP protocol used to communicate with this helper.
903 The default is to auto-detect IPv6 and use it when available.
906 FORMAT is a series of %macro codes. See logformat directive for a full list
907 of the accepted codes. Although note that at the time of any external ACL
908 being tested data may not be available and thus some %macro expand to '-'.
910 In addition to the logformat codes; when processing external ACLs these
911 additional macros are made available:
913 %ACL The name of the ACL being tested.
915 %DATA The ACL arguments specified in the referencing config
916 'acl ... external' line, separated by spaces (an
917 "argument string"). see acl external.
919 If there are no ACL arguments %DATA expands to '-'.
921 If you do not specify a DATA macro inside FORMAT,
922 Squid automatically appends %DATA to your FORMAT.
923 Note that Squid-3.x may expand %DATA to whitespace
924 or nothing in this case.
926 By default, Squid applies URL-encoding to each ACL
927 argument inside the argument string. If an explicit
928 encoding modifier is used (e.g., %#DATA), then Squid
929 encodes the whole argument string as a single token
930 (e.g., with %#DATA, spaces between arguments become
933 If SSL is enabled, the following formatting codes become available:
935 %USER_CERT SSL User certificate in PEM format
936 %USER_CERTCHAIN SSL User certificate chain in PEM format
937 %USER_CERT_xx SSL User certificate subject attribute xx
938 %USER_CA_CERT_xx SSL User certificate issuer attribute xx
941 NOTE: all other format codes accepted by older Squid versions
945 General request syntax:
947 [channel-ID] FORMAT-values
950 FORMAT-values consists of transaction details expanded with
951 whitespace separation per the config file FORMAT specification
952 using the FORMAT macros listed above.
954 Request values sent to the helper are URL escaped to protect
955 each value in requests against whitespaces.
957 If using protocol=2.5 then the request sent to the helper is not
958 URL escaped to protect against whitespace.
960 NOTE: protocol=3.0 is deprecated as no longer necessary.
962 When using the concurrency= option the protocol is changed by
963 introducing a query channel tag in front of the request/response.
964 The query channel tag is a number between 0 and concurrency-1.
965 This value must be echoed back unchanged to Squid as the first part
966 of the response relating to its request.
969 The helper receives lines expanded per the above format specification
970 and for each input line returns 1 line starting with OK/ERR/BH result
971 code and optionally followed by additional keywords with more details.
974 General result syntax:
976 [channel-ID] result keyword=value ...
978 Result consists of one of the codes:
981 the ACL test produced a match.
984 the ACL test does not produce a match.
987 An internal error occurred in the helper, preventing
988 a result being identified.
990 The meaning of 'a match' is determined by your squid.conf
991 access control configuration. See the Squid wiki for details.
995 user= The users name (login)
997 password= The users password (for login= cache_peer option)
999 message= Message describing the reason for this response.
1000 Available as %o in error pages.
1001 Useful on (ERR and BH results).
1003 tag= Apply a tag to a request. Only sets a tag once,
1004 does not alter existing tags.
1006 log= String to be logged in access.log. Available as
1007 %ea in logformat specifications.
1009 clt_conn_tag= Associates a TAG with the client TCP connection.
1010 Please see url_rewrite_program related documentation
1013 Any keywords may be sent on any response whether OK, ERR or BH.
1015 All response keyword values need to be a single token with URL
1016 escaping, or enclosed in double quotes (") and escaped using \ on
1017 any double quotes or \ characters within the value. The wrapping
1018 double quotes are removed before the value is interpreted by Squid.
1019 \r and \n are also replace by CR and LF.
1021 Some example key values:
1025 user="J. \"Bob\" Smith"
1032 DEFAULT: ssl::certHasExpired ssl_error X509_V_ERR_CERT_HAS_EXPIRED
1033 DEFAULT: ssl::certNotYetValid ssl_error X509_V_ERR_CERT_NOT_YET_VALID
1034 DEFAULT: ssl::certDomainMismatch ssl_error SQUID_X509_V_ERR_DOMAIN_MISMATCH
1035 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
1036 DEFAULT: ssl::certSelfSigned ssl_error X509_V_ERR_DEPTH_ZERO_SELF_SIGNED_CERT
1038 DEFAULT: all src all
1039 DEFAULT: manager url_regex -i ^cache_object:// +i ^https?://[^/]+/squid-internal-mgr/
1040 DEFAULT: localhost src 127.0.0.1/32 ::1
1041 DEFAULT: to_localhost dst 127.0.0.0/8 0.0.0.0/32 ::1/128 ::/128
1042 DEFAULT: CONNECT method CONNECT
1043 DEFAULT_DOC: ACLs all, manager, localhost, to_localhost, and CONNECT are predefined.
1045 Defining an Access List
1047 Every access list definition must begin with an aclname and acltype,
1048 followed by either type-specific arguments or a quoted filename that
1051 acl aclname acltype argument ...
1052 acl aclname acltype "file" ...
1054 When using "file", the file should contain one item per line.
1059 Some acl types supports options which changes their default behaviour:
1061 -i,+i By default, regular expressions are CASE-SENSITIVE. To make them
1062 case-insensitive, use the -i option. To return case-sensitive
1063 use the +i option between patterns, or make a new ACL line
1066 -n Disable lookups and address type conversions. If lookup or
1067 conversion is required because the parameter type (IP or
1068 domain name) does not match the message address type (domain
1069 name or IP), then the ACL would immediately declare a mismatch
1070 without any warnings or lookups.
1073 Perform a list membership test, interpreting values as
1074 comma-separated token lists and matching against individual
1075 tokens instead of whole values.
1076 The optional "delimiters" parameter specifies one or more
1077 alternative non-alphanumeric delimiter characters.
1078 non-alphanumeric delimiter characters.
1080 -- Used to stop processing all options, in the case the first acl
1081 value has '-' character as first character (for example the '-'
1082 is a valid domain name)
1084 Some acl types require suspending the current request in order
1085 to access some external data source.
1086 Those which do are marked with the tag [slow], those which
1087 don't are marked as [fast].
1088 See http://wiki.squid-cache.org/SquidFaq/SquidAcl
1089 for further information
1091 ***** ACL TYPES AVAILABLE *****
1093 acl aclname src ip-address/mask ... # clients IP address [fast]
1094 acl aclname src addr1-addr2/mask ... # range of addresses [fast]
1095 acl aclname dst [-n] ip-address/mask ... # URL host's IP address [slow]
1096 acl aclname localip ip-address/mask ... # IP address the client connected to [fast]
1099 acl aclname arp mac-address ...
1100 acl aclname eui64 eui64-address ...
1102 # MAC (EUI-48) and EUI-64 addresses use xx:xx:xx:xx:xx:xx notation.
1104 # The 'arp' ACL code is not portable to all operating systems.
1105 # It works on Linux, Solaris, Windows, FreeBSD, and some other
1108 # The eui_lookup directive is required to be 'on' (the default)
1109 # and Squid built with --enable-eui for MAC/EUI addresses to be
1110 # available for this ACL.
1112 # Squid can only determine the MAC/EUI address for IPv4
1113 # clients that are on the same subnet. If the client is on a
1114 # different subnet, then Squid cannot find out its address.
1116 # IPv6 protocol does not contain ARP. MAC/EUI is either
1117 # encoded directly in the IPv6 address or not available.
1119 acl aclname clientside_mark mark[/mask] ...
1120 # matches CONNMARK of an accepted connection [fast]
1121 # DEPRECATED. Use the 'client_connection_mark' instead.
1123 acl aclname client_connection_mark mark[/mask] ...
1124 # matches CONNMARK of an accepted connection [fast]
1126 # mark and mask are unsigned integers (hex, octal, or decimal).
1127 # If multiple marks are given, then the ACL matches if at least
1130 # Uses netfilter-conntrack library.
1131 # Requires building Squid with --enable-linux-netfilter.
1133 # The client, various intermediaries, and Squid itself may set
1134 # CONNMARK at various times. The last CONNMARK set wins. This ACL
1135 # checks the mark present on an accepted connection or set by
1136 # Squid afterwards, depending on the ACL check timing. This ACL
1137 # effectively ignores any mark set by other agents after Squid has
1138 # accepted the connection.
1140 acl aclname srcdomain .foo.com ...
1141 # reverse lookup, from client IP [slow]
1142 acl aclname dstdomain [-n] .foo.com ...
1143 # Destination server from URL [fast]
1144 acl aclname srcdom_regex [-i] \.foo\.com ...
1145 # regex matching client name [slow]
1146 acl aclname dstdom_regex [-n] [-i] \.foo\.com ...
1147 # regex matching server [fast]
1149 # For dstdomain and dstdom_regex a reverse lookup is tried if a IP
1150 # based URL is used and no match is found. The name "none" is used
1151 # if the reverse lookup fails.
1153 acl aclname src_as number ...
1154 acl aclname dst_as number ...
1156 # Except for access control, AS numbers can be used for
1157 # routing of requests to specific caches. Here's an
1158 # example for routing all requests for AS#1241 and only
1159 # those to mycache.mydomain.net:
1160 # acl asexample dst_as 1241
1161 # cache_peer_access mycache.mydomain.net allow asexample
1162 # cache_peer_access mycache_mydomain.net deny all
1164 acl aclname peername myPeer ...
1165 acl aclname peername_regex [-i] regex-pattern ...
1167 # match against a named cache_peer entry
1168 # set unique name= on cache_peer lines for reliable use.
1170 acl aclname time [day-abbrevs] [h1:m1-h2:m2]
1180 # h1:m1 must be less than h2:m2
1182 acl aclname url_regex [-i] ^http:// ...
1183 # regex matching on whole URL [fast]
1184 acl aclname urllogin [-i] [^a-zA-Z0-9] ...
1185 # regex matching on URL login field
1186 acl aclname urlpath_regex [-i] \.gif$ ...
1187 # regex matching on URL path [fast]
1189 acl aclname port 80 70 21 0-1024... # destination TCP port [fast]
1190 # ranges are allowed
1191 acl aclname localport 3128 ... # TCP port the client connected to [fast]
1192 # NP: for interception mode this is usually '80'
1194 acl aclname myportname 3128 ... # *_port name [fast]
1196 acl aclname proto HTTP FTP ... # request protocol [fast]
1198 acl aclname method GET POST ... # HTTP request method [fast]
1200 acl aclname http_status 200 301 500- 400-403 ...
1201 # status code in reply [fast]
1203 acl aclname browser [-i] regexp ...
1204 # pattern match on User-Agent header (see also req_header below) [fast]
1206 acl aclname referer_regex [-i] regexp ...
1207 # pattern match on Referer header [fast]
1208 # Referer is highly unreliable, so use with care
1210 acl aclname ident [-i] username ...
1211 acl aclname ident_regex [-i] pattern ...
1212 # string match on ident output [slow]
1213 # use REQUIRED to accept any non-null ident.
1215 acl aclname proxy_auth [-i] username ...
1216 acl aclname proxy_auth_regex [-i] pattern ...
1217 # perform http authentication challenge to the client and match against
1218 # supplied credentials [slow]
1220 # takes a list of allowed usernames.
1221 # use REQUIRED to accept any valid username.
1223 # Will use proxy authentication in forward-proxy scenarios, and plain
1224 # http authentication in reverse-proxy scenarios
1226 # NOTE: when a Proxy-Authentication header is sent but it is not
1227 # needed during ACL checking the username is NOT logged
1230 # NOTE: proxy_auth requires a EXTERNAL authentication program
1231 # to check username/password combinations (see
1232 # auth_param directive).
1234 # NOTE: proxy_auth can't be used in a transparent/intercepting proxy
1235 # as the browser needs to be configured for using a proxy in order
1236 # to respond to proxy authentication.
1238 acl aclname snmp_community string ...
1239 # A community string to limit access to your SNMP Agent [fast]
1242 # acl snmppublic snmp_community public
1244 acl aclname maxconn number
1245 # This will be matched when the client's IP address has
1246 # more than <number> TCP connections established. [fast]
1247 # NOTE: This only measures direct TCP links so X-Forwarded-For
1248 # indirect clients are not counted.
1250 acl aclname max_user_ip [-s] number
1251 # This will be matched when the user attempts to log in from more
1252 # than <number> different ip addresses. The authenticate_ip_ttl
1253 # parameter controls the timeout on the ip entries. [fast]
1254 # If -s is specified the limit is strict, denying browsing
1255 # from any further IP addresses until the ttl has expired. Without
1256 # -s Squid will just annoy the user by "randomly" denying requests.
1257 # (the counter is reset each time the limit is reached and a
1258 # request is denied)
1259 # NOTE: in acceleration mode or where there is mesh of child proxies,
1260 # clients may appear to come from multiple addresses if they are
1261 # going through proxy farms, so a limit of 1 may cause user problems.
1263 acl aclname random probability
1264 # Pseudo-randomly match requests. Based on the probability given.
1265 # Probability may be written as a decimal (0.333), fraction (1/3)
1266 # or ratio of matches:non-matches (3:5).
1268 acl aclname req_mime_type [-i] mime-type ...
1269 # regex match against the mime type of the request generated
1270 # by the client. Can be used to detect file upload or some
1271 # types HTTP tunneling requests [fast]
1272 # NOTE: This does NOT match the reply. You cannot use this
1273 # to match the returned file type.
1275 acl aclname req_header header-name [-i] any\.regex\.here
1276 # regex match against any of the known request headers. May be
1277 # thought of as a superset of "browser", "referer" and "mime-type"
1280 acl aclname rep_mime_type [-i] mime-type ...
1281 # regex match against the mime type of the reply received by
1282 # squid. Can be used to detect file download or some
1283 # types HTTP tunneling requests. [fast]
1284 # NOTE: This has no effect in http_access rules. It only has
1285 # effect in rules that affect the reply data stream such as
1286 # http_reply_access.
1288 acl aclname rep_header header-name [-i] any\.regex\.here
1289 # regex match against any of the known reply headers. May be
1290 # thought of as a superset of "browser", "referer" and "mime-type"
1293 acl aclname external class_name [arguments...]
1294 # external ACL lookup via a helper class defined by the
1295 # external_acl_type directive [slow]
1297 acl aclname user_cert attribute values...
1298 # match against attributes in a user SSL certificate
1299 # attribute is one of DN/C/O/CN/L/ST or a numerical OID [fast]
1301 acl aclname ca_cert attribute values...
1302 # match against attributes a users issuing CA SSL certificate
1303 # attribute is one of DN/C/O/CN/L/ST or a numerical OID [fast]
1305 acl aclname ext_user [-i] username ...
1306 acl aclname ext_user_regex [-i] pattern ...
1307 # string match on username returned by external acl helper [slow]
1308 # use REQUIRED to accept any non-null user name.
1310 acl aclname tag tagvalue ...
1311 # string match on tag returned by external acl helper [fast]
1312 # DEPRECATED. Only the first tag will match with this ACL.
1313 # Use the 'note' ACL instead for handling multiple tag values.
1315 acl aclname hier_code codename ...
1316 # string match against squid hierarchy code(s); [fast]
1317 # e.g., DIRECT, PARENT_HIT, NONE, etc.
1319 # NOTE: This has no effect in http_access rules. It only has
1320 # effect in rules that affect the reply data stream such as
1321 # http_reply_access.
1323 acl aclname note [-m[=delimiters]] name [value ...]
1324 # match transaction annotation [fast]
1325 # Without values, matches any annotation with a given name.
1326 # With value(s), matches any annotation with a given name that
1327 # also has one of the given values.
1328 # If the -m flag is used, then the value of the named
1329 # annotation is interpreted as a list of tokens, and the ACL
1330 # matches individual name=token pairs rather than whole
1331 # name=value pairs. See "ACL Options" above for more info.
1332 # Annotation sources include note and adaptation_meta directives
1333 # as well as helper and eCAP responses.
1335 acl aclname annotate_transaction [-m[=delimiters]] key=value ...
1336 acl aclname annotate_transaction [-m[=delimiters]] key+=value ...
1337 # Always matches. [fast]
1338 # Used for its side effect: This ACL immediately adds a
1339 # key=value annotation to the current master transaction.
1340 # The added annotation can then be tested using note ACL and
1341 # logged (or sent to helpers) using %note format code.
1343 # Annotations can be specified using replacement and addition
1344 # formats. The key=value form replaces old same-key annotation
1345 # value(s). The key+=value form appends a new value to the old
1346 # same-key annotation. Both forms create a new key=value
1347 # annotation if no same-key annotation exists already. If
1348 # -m flag is used, then the value is interpreted as a list
1349 # and the annotation will contain key=token pair(s) instead of the
1350 # whole key=value pair.
1352 # This ACL is especially useful for recording complex multi-step
1353 # ACL-driven decisions. For example, the following configuration
1354 # avoids logging transactions accepted after aclX matched:
1356 # # First, mark transactions accepted after aclX matched
1357 # acl markSpecial annotate_transaction special=true
1358 # http_access allow acl001
1360 # http_access deny acl100
1361 # http_access allow aclX markSpecial
1363 # # Second, do not log marked transactions:
1364 # acl markedSpecial note special true
1365 # access_log ... deny markedSpecial
1367 # # Note that the following would not have worked because aclX
1368 # # alone does not determine whether the transaction was allowed:
1369 # access_log ... deny aclX # Wrong!
1371 # Warning: This ACL annotates the transaction even when negated
1372 # and even if subsequent ACLs fail to match. For example, the
1373 # following three rules will have exactly the same effect as far
1374 # as annotations set by the "mark" ACL are concerned:
1376 # some_directive acl1 ... mark # rule matches if mark is reached
1377 # some_directive acl1 ... !mark # rule never matches
1378 # some_directive acl1 ... mark !all # rule never matches
1380 acl aclname annotate_client [-m[=delimiters]] key=value ...
1381 acl aclname annotate_client [-m[=delimiters]] key+=value ...
1383 # Always matches. [fast]
1384 # Used for its side effect: This ACL immediately adds a
1385 # key=value annotation to the current client-to-Squid
1386 # connection. Connection annotations are propagated to the current
1387 # and all future master transactions on the annotated connection.
1388 # See the annotate_transaction ACL for details.
1390 # For example, the following configuration avoids rewriting URLs
1391 # of transactions bumped by SslBump:
1393 # # First, mark bumped connections:
1394 # acl markBumped annotate_client bumped=true
1395 # ssl_bump peek acl1
1396 # ssl_bump stare acl2
1397 # ssl_bump bump acl3 markBumped
1398 # ssl_bump splice all
1400 # # Second, do not send marked transactions to the redirector:
1401 # acl markedBumped note bumped true
1402 # url_rewrite_access deny markedBumped
1404 # # Note that the following would not have worked because acl3 alone
1405 # # does not determine whether the connection is going to be bumped:
1406 # url_rewrite_access deny acl3 # Wrong!
1408 acl aclname adaptation_service service ...
1409 # Matches the name of any icap_service, ecap_service,
1410 # adaptation_service_set, or adaptation_service_chain that Squid
1411 # has used (or attempted to use) for the master transaction.
1412 # This ACL must be defined after the corresponding adaptation
1413 # service is named in squid.conf. This ACL is usable with
1414 # adaptation_meta because it starts matching immediately after
1415 # the service has been selected for adaptation.
1417 acl aclname transaction_initiator initiator ...
1418 # Matches transaction's initiator [fast]
1420 # Supported initiators are:
1421 # esi: matches transactions fetching ESI resources
1422 # certificate-fetching: matches transactions fetching
1423 # a missing intermediate TLS certificate
1424 # cache-digest: matches transactions fetching Cache Digests
1426 # htcp: matches HTCP requests from peers
1427 # icp: matches ICP requests to peers
1428 # icmp: matches ICMP RTT database (NetDB) requests to peers
1429 # asn: matches asns db requests
1430 # internal: matches any of the above
1431 # client: matches transactions containing an HTTP or FTP
1432 # client request received at a Squid *_port
1433 # all: matches any transaction, including internal transactions
1434 # without a configurable initiator and hopefully rare
1435 # transactions without a known-to-Squid initiator
1437 # Multiple initiators are ORed.
1439 acl aclname has component
1440 # matches a transaction "component" [fast]
1442 # Supported transaction components are:
1443 # request: transaction has a request header (at least)
1444 # response: transaction has a response header (at least)
1445 # ALE: transaction has an internally-generated Access Log Entry
1446 # structure; bugs notwithstanding, all transaction have it
1448 # For example, the following configuration helps when dealing with HTTP
1449 # clients that close connections without sending a request header:
1451 # acl hasRequest has request
1452 # acl logMe note important_transaction
1453 # # avoid "logMe ACL is used in context without an HTTP request" warnings
1454 # access_log ... logformat=detailed hasRequest logMe
1455 # # log request-less transactions, instead of ignoring them
1456 # access_log ... logformat=brief !hasRequest
1458 # Multiple components are not supported for one "acl" rule, but
1459 # can be specified (and are ORed) using multiple same-name rules:
1461 # # OK, this strange logging daemon needs request or response,
1462 # # but can work without either a request or a response:
1463 # acl hasWhatMyLoggingDaemonNeeds has request
1464 # acl hasWhatMyLoggingDaemonNeeds has response
1466 acl aclname at_step step
1467 # match against the current request processing step [fast]
1469 # GeneratingCONNECT: Generating HTTP CONNECT request headers
1471 # The following ssl_bump processing steps are recognized:
1472 # SslBump1: After getting TCP-level and HTTP CONNECT info.
1473 # SslBump2: After getting SSL Client Hello info.
1474 # SslBump3: After getting SSL Server Hello info.
1478 acl aclname ssl_error errorname
1479 # match against SSL certificate validation error [fast]
1481 # For valid error names see in @DEFAULT_ERROR_DIR@/templates/error-details.txt
1484 # The following can be used as shortcuts for certificate properties:
1485 # [ssl::]certHasExpired: the "not after" field is in the past
1486 # [ssl::]certNotYetValid: the "not before" field is in the future
1487 # [ssl::]certUntrusted: The certificate issuer is not to be trusted.
1488 # [ssl::]certSelfSigned: The certificate is self signed.
1489 # [ssl::]certDomainMismatch: The certificate CN domain does not
1490 # match the name the name of the host we are connecting to.
1492 # The ssl::certHasExpired, ssl::certNotYetValid, ssl::certDomainMismatch,
1493 # ssl::certUntrusted, and ssl::certSelfSigned can also be used as
1494 # predefined ACLs, just like the 'all' ACL.
1496 # NOTE: The ssl_error ACL is only supported with sslproxy_cert_error,
1497 # sslproxy_cert_sign, and sslproxy_cert_adapt options.
1499 acl aclname server_cert_fingerprint [-sha1] fingerprint
1500 # match against server SSL certificate fingerprint [fast]
1502 # The fingerprint is the digest of the DER encoded version
1503 # of the whole certificate. The user should use the form: XX:XX:...
1504 # Optional argument specifies the digest algorithm to use.
1505 # The SHA1 digest algorithm is the default and is currently
1506 # the only algorithm supported (-sha1).
1508 acl aclname ssl::server_name [option] .foo.com ...
1509 # matches server name obtained from various sources [fast]
1511 # The ACL computes server name(s) using such information sources as
1512 # CONNECT request URI, TLS client SNI, and TLS server certificate
1513 # subject (CN and SubjectAltName). The computed server name(s) usually
1514 # change with each SslBump step, as more info becomes available:
1515 # * SNI is used as the server name instead of the request URI,
1516 # * subject name(s) from the server certificate (CN and
1517 # SubjectAltName) are used as the server names instead of SNI.
1519 # When the ACL computes multiple server names, matching any single
1520 # computed name is sufficient for the ACL to match.
1522 # The "none" name can be used to match transactions where the ACL
1523 # could not compute the server name using any information source
1524 # that was both available and allowed to be used by the ACL options at
1525 # the ACL evaluation time.
1527 # Unlike dstdomain, this ACL does not perform DNS lookups.
1529 # An ACL option below may be used to restrict what information
1530 # sources are used to extract the server names from:
1532 # --client-requested
1533 # The server name is SNI regardless of what the server says.
1535 # The server name(s) are the certificate subject name(s), regardless
1536 # of what the client has requested. If the server certificate is
1537 # unavailable, then the name is "none".
1539 # The server name is either SNI (if SNI matches at least one of the
1540 # certificate subject names) or "none" (otherwise). When the server
1541 # certificate is unavailable, the consensus server name is SNI.
1543 # Combining multiple options in one ACL is a fatal configuration
1546 # For all options: If no SNI is available, then the CONNECT request
1547 # target (a.k.a. URI) is used instead of SNI (for an intercepted
1548 # connection, this target is the destination IP address).
1550 acl aclname ssl::server_name_regex [-i] \.foo\.com ...
1551 # regex matches server name obtained from various sources [fast]
1553 acl aclname connections_encrypted
1554 # matches transactions with all HTTP messages received over TLS
1555 # transport connections. [fast]
1557 # The master transaction deals with HTTP messages received from
1558 # various sources. All sources used by the master transaction in the
1559 # past are considered by the ACL. The following rules define whether
1560 # a given message source taints the entire master transaction,
1561 # resulting in ACL mismatches:
1563 # * The HTTP client transport connection is not TLS.
1564 # * An adaptation service connection-encryption flag is off.
1565 # * The peer or origin server transport connection is not TLS.
1567 # Caching currently does not affect these rules. This cache ignorance
1568 # implies that only the current HTTP client transport and REQMOD
1569 # services status determine whether this ACL matches a from-cache
1570 # transaction. The source of the cached response does not have any
1571 # effect on future transaction that use the cached response without
1572 # revalidation. This may change.
1574 # DNS, ICP, and HTCP exchanges during the master transaction do not
1575 # affect these rules.
1577 acl aclname any-of acl1 acl2 ...
1578 # match any one of the acls [fast or slow]
1579 # The first matching ACL stops further ACL evaluation.
1581 # ACLs from multiple any-of lines with the same name are ORed.
1582 # For example, A = (a1 or a2) or (a3 or a4) can be written as
1583 # acl A any-of a1 a2
1584 # acl A any-of a3 a4
1586 # This group ACL is fast if all evaluated ACLs in the group are fast
1587 # and slow otherwise.
1589 acl aclname all-of acl1 acl2 ...
1590 # match all of the acls [fast or slow]
1591 # The first mismatching ACL stops further ACL evaluation.
1593 # ACLs from multiple all-of lines with the same name are ORed.
1594 # For example, B = (b1 and b2) or (b3 and b4) can be written as
1595 # acl B all-of b1 b2
1596 # acl B all-of b3 b4
1598 # This group ACL is fast if all evaluated ACLs in the group are fast
1599 # and slow otherwise.
1602 acl macaddress arp 09:00:2b:23:45:67
1603 acl myexample dst_as 1241
1604 acl password proxy_auth REQUIRED
1605 acl fileupload req_mime_type -i ^multipart/form-data$
1606 acl javascript rep_mime_type -i ^application/x-javascript$
1610 # Recommended minimum configuration:
1613 # Example rule allowing access from your local networks.
1614 # Adapt to list your (internal) IP networks from where browsing
1616 acl localnet src 0.0.0.1-0.255.255.255 # RFC 1122 "this" network (LAN)
1617 acl localnet src 10.0.0.0/8 # RFC 1918 local private network (LAN)
1618 acl localnet src 100.64.0.0/10 # RFC 6598 shared address space (CGN)
1619 acl localnet src 169.254.0.0/16 # RFC 3927 link-local (directly plugged) machines
1620 acl localnet src 172.16.0.0/12 # RFC 1918 local private network (LAN)
1621 acl localnet src 192.168.0.0/16 # RFC 1918 local private network (LAN)
1622 acl localnet src fc00::/7 # RFC 4193 local private network range
1623 acl localnet src fe80::/10 # RFC 4291 link-local (directly plugged) machines
1625 acl SSL_ports port 443
1626 acl Safe_ports port 80 # http
1627 acl Safe_ports port 21 # ftp
1628 acl Safe_ports port 443 # https
1629 acl Safe_ports port 70 # gopher
1630 acl Safe_ports port 210 # wais
1631 acl Safe_ports port 1025-65535 # unregistered ports
1632 acl Safe_ports port 280 # http-mgmt
1633 acl Safe_ports port 488 # gss-http
1634 acl Safe_ports port 591 # filemaker
1635 acl Safe_ports port 777 # multiling http
1639 NAME: proxy_protocol_access
1641 LOC: Config.accessList.proxyProtocol
1643 DEFAULT_DOC: all TCP connections to ports with require-proxy-header will be denied
1645 Determine which client proxies can be trusted to provide correct
1646 information regarding real client IP address using PROXY protocol.
1648 Requests may pass through a chain of several other proxies
1649 before reaching us. The original source details may by sent in:
1650 * HTTP message Forwarded header, or
1651 * HTTP message X-Forwarded-For header, or
1652 * PROXY protocol connection header.
1654 This directive is solely for validating new PROXY protocol
1655 connections received from a port flagged with require-proxy-header.
1656 It is checked only once after TCP connection setup.
1658 A deny match results in TCP connection closure.
1660 An allow match is required for Squid to permit the corresponding
1661 TCP connection, before Squid even looks for HTTP request headers.
1662 If there is an allow match, Squid starts using PROXY header information
1663 to determine the source address of the connection for all future ACL
1664 checks, logging, etc.
1666 SECURITY CONSIDERATIONS:
1668 Any host from which we accept client IP details can place
1669 incorrect information in the relevant header, and Squid
1670 will use the incorrect information as if it were the
1671 source address of the request. This may enable remote
1672 hosts to bypass any access control restrictions that are
1673 based on the client's source addresses.
1675 This clause only supports fast acl types.
1676 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1679 NAME: follow_x_forwarded_for
1681 IFDEF: FOLLOW_X_FORWARDED_FOR
1682 LOC: Config.accessList.followXFF
1683 DEFAULT_IF_NONE: deny all
1684 DEFAULT_DOC: X-Forwarded-For header will be ignored.
1686 Determine which client proxies can be trusted to provide correct
1687 information regarding real client IP address.
1689 Requests may pass through a chain of several other proxies
1690 before reaching us. The original source details may by sent in:
1691 * HTTP message Forwarded header, or
1692 * HTTP message X-Forwarded-For header, or
1693 * PROXY protocol connection header.
1695 PROXY protocol connections are controlled by the proxy_protocol_access
1696 directive which is checked before this.
1698 If a request reaches us from a source that is allowed by this
1699 directive, then we trust the information it provides regarding
1700 the IP of the client it received from (if any).
1702 For the purpose of ACLs used in this directive the src ACL type always
1703 matches the address we are testing and srcdomain matches its rDNS.
1705 On each HTTP request Squid checks for X-Forwarded-For header fields.
1706 If found the header values are iterated in reverse order and an allow
1707 match is required for Squid to continue on to the next value.
1708 The verification ends when a value receives a deny match, cannot be
1709 tested, or there are no more values to test.
1710 NOTE: Squid does not yet follow the Forwarded HTTP header.
1712 The end result of this process is an IP address that we will
1713 refer to as the indirect client address. This address may
1714 be treated as the client address for access control, ICAP, delay
1715 pools and logging, depending on the acl_uses_indirect_client,
1716 icap_uses_indirect_client, delay_pool_uses_indirect_client,
1717 log_uses_indirect_client and tproxy_uses_indirect_client options.
1719 This clause only supports fast acl types.
1720 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1722 SECURITY CONSIDERATIONS:
1724 Any host from which we accept client IP details can place
1725 incorrect information in the relevant header, and Squid
1726 will use the incorrect information as if it were the
1727 source address of the request. This may enable remote
1728 hosts to bypass any access control restrictions that are
1729 based on the client's source addresses.
1733 acl localhost src 127.0.0.1
1734 acl my_other_proxy srcdomain .proxy.example.com
1735 follow_x_forwarded_for allow localhost
1736 follow_x_forwarded_for allow my_other_proxy
1739 NAME: acl_uses_indirect_client
1742 IFDEF: FOLLOW_X_FORWARDED_FOR
1744 LOC: Config.onoff.acl_uses_indirect_client
1746 Controls whether the indirect client address
1747 (see follow_x_forwarded_for) is used instead of the
1748 direct client address in acl matching.
1750 NOTE: maxconn ACL considers direct TCP links and indirect
1751 clients will always have zero. So no match.
1754 NAME: delay_pool_uses_indirect_client
1757 IFDEF: FOLLOW_X_FORWARDED_FOR&&USE_DELAY_POOLS
1759 LOC: Config.onoff.delay_pool_uses_indirect_client
1761 Controls whether the indirect client address
1762 (see follow_x_forwarded_for) is used instead of the
1763 direct client address in delay pools.
1766 NAME: log_uses_indirect_client
1769 IFDEF: FOLLOW_X_FORWARDED_FOR
1771 LOC: Config.onoff.log_uses_indirect_client
1773 Controls whether the indirect client address
1774 (see follow_x_forwarded_for) is used instead of the
1775 direct client address in the access log.
1778 NAME: tproxy_uses_indirect_client
1781 IFDEF: FOLLOW_X_FORWARDED_FOR&&LINUX_NETFILTER
1783 LOC: Config.onoff.tproxy_uses_indirect_client
1785 Controls whether the indirect client address
1786 (see follow_x_forwarded_for) is used instead of the
1787 direct client address when spoofing the outgoing client.
1789 This has no effect on requests arriving in non-tproxy
1792 SECURITY WARNING: Usage of this option is dangerous
1793 and should not be used trivially. Correct configuration
1794 of follow_x_forwarded_for with a limited set of trusted
1795 sources is required to prevent abuse of your proxy.
1798 NAME: spoof_client_ip
1800 LOC: Config.accessList.spoof_client_ip
1802 DEFAULT_DOC: Allow spoofing on all TPROXY traffic.
1804 Control client IP address spoofing of TPROXY traffic based on
1805 defined access lists.
1807 spoof_client_ip allow|deny [!]aclname ...
1809 If there are no "spoof_client_ip" lines present, the default
1810 is to "allow" spoofing of any suitable request.
1812 Note that the cache_peer "no-tproxy" option overrides this ACL.
1814 This clause supports fast acl types.
1815 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1820 LOC: Config.accessList.http
1821 DEFAULT_IF_NONE: deny all
1822 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
1824 Allowing or Denying access based on defined access lists
1826 To allow or deny a message received on an HTTP, HTTPS, or FTP port:
1827 http_access allow|deny [!]aclname ...
1829 NOTE on default values:
1831 If there are no "access" lines present, the default is to deny
1834 If none of the "access" lines cause a match, the default is the
1835 opposite of the last line in the list. If the last line was
1836 deny, the default is allow. Conversely, if the last line
1837 is allow, the default will be deny. For these reasons, it is a
1838 good idea to have an "deny all" entry at the end of your access
1839 lists to avoid potential confusion.
1841 This clause supports both fast and slow acl types.
1842 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1847 # Recommended minimum Access Permission configuration:
1849 # Deny requests to certain unsafe ports
1850 http_access deny !Safe_ports
1852 # Deny CONNECT to other than secure SSL ports
1853 http_access deny CONNECT !SSL_ports
1855 # Only allow cachemgr access from localhost
1856 http_access allow localhost manager
1857 http_access deny manager
1859 # We strongly recommend the following be uncommented to protect innocent
1860 # web applications running on the proxy server who think the only
1861 # one who can access services on "localhost" is a local user
1862 #http_access deny to_localhost
1865 # INSERT YOUR OWN RULE(S) HERE TO ALLOW ACCESS FROM YOUR CLIENTS
1868 # Example rule allowing access from your local networks.
1869 # Adapt localnet in the ACL section to list your (internal) IP networks
1870 # from where browsing should be allowed
1871 http_access allow localnet
1872 http_access allow localhost
1874 # And finally deny all other access to this proxy
1875 http_access deny all
1879 NAME: adapted_http_access http_access2
1881 LOC: Config.accessList.adapted_http
1883 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
1885 Allowing or Denying access based on defined access lists
1887 Essentially identical to http_access, but runs after redirectors
1888 and ICAP/eCAP adaptation. Allowing access control based on their
1891 If not set then only http_access is used.
1894 NAME: http_reply_access
1896 LOC: Config.accessList.reply
1898 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
1900 Allow replies to client requests. This is complementary to http_access.
1902 http_reply_access allow|deny [!] aclname ...
1904 NOTE: if there are no access lines present, the default is to allow
1907 If none of the access lines cause a match the opposite of the
1908 last line will apply. Thus it is good practice to end the rules
1909 with an "allow all" or "deny all" entry.
1911 This clause supports both fast and slow acl types.
1912 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1917 LOC: Config.accessList.icp
1919 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
1921 Allowing or Denying access to the ICP port based on defined
1924 icp_access allow|deny [!]aclname ...
1926 NOTE: The default if no icp_access lines are present is to
1927 deny all traffic. This default may cause problems with peers
1930 This clause only supports fast acl types.
1931 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1933 # Allow ICP queries from local networks only
1934 #icp_access allow localnet
1935 #icp_access deny all
1941 LOC: Config.accessList.htcp
1943 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
1945 Allowing or Denying access to the HTCP port based on defined
1948 htcp_access allow|deny [!]aclname ...
1950 See also htcp_clr_access for details on access control for
1951 cache purge (CLR) HTCP messages.
1953 NOTE: The default if no htcp_access lines are present is to
1954 deny all traffic. This default may cause problems with peers
1955 using the htcp option.
1957 This clause only supports fast acl types.
1958 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1960 # Allow HTCP queries from local networks only
1961 #htcp_access allow localnet
1962 #htcp_access deny all
1965 NAME: htcp_clr_access
1968 LOC: Config.accessList.htcp_clr
1970 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
1972 Allowing or Denying access to purge content using HTCP based
1973 on defined access lists.
1974 See htcp_access for details on general HTCP access control.
1976 htcp_clr_access allow|deny [!]aclname ...
1978 This clause only supports fast acl types.
1979 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1981 # Allow HTCP CLR requests from trusted peers
1982 acl htcp_clr_peer src 192.0.2.2 2001:DB8::2
1983 htcp_clr_access allow htcp_clr_peer
1984 htcp_clr_access deny all
1989 LOC: Config.accessList.miss
1991 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
1993 Determines whether network access is permitted when satisfying a request.
1996 to force your neighbors to use you as a sibling instead of
1999 acl localclients src 192.0.2.0/24 2001:DB8::a:0/64
2000 miss_access deny !localclients
2001 miss_access allow all
2003 This means only your local clients are allowed to fetch relayed/MISS
2004 replies from the network and all other clients can only fetch cached
2007 The default for this setting allows all clients who passed the
2008 http_access rules to relay via this proxy.
2010 This clause only supports fast acl types.
2011 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2014 NAME: ident_lookup_access
2018 DEFAULT_DOC: Unless rules exist in squid.conf, IDENT is not fetched.
2019 LOC: Ident::TheConfig.identLookup
2021 A list of ACL elements which, if matched, cause an ident
2022 (RFC 931) lookup to be performed for this request. For
2023 example, you might choose to always perform ident lookups
2024 for your main multi-user Unix boxes, but not for your Macs
2025 and PCs. By default, ident lookups are not performed for
2028 To enable ident lookups for specific client addresses, you
2029 can follow this example:
2031 acl ident_aware_hosts src 198.168.1.0/24
2032 ident_lookup_access allow ident_aware_hosts
2033 ident_lookup_access deny all
2035 Only src type ACL checks are fully supported. A srcdomain
2036 ACL might work at times, but it will not always provide
2039 This clause only supports fast acl types.
2040 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2043 NAME: reply_body_max_size
2044 COMMENT: size [acl acl...]
2047 DEFAULT_DOC: No limit is applied.
2048 LOC: Config.ReplyBodySize
2050 This option specifies the maximum size of a reply body. It can be
2051 used to prevent users from downloading very large files, such as
2052 MP3's and movies. When the reply headers are received, the
2053 reply_body_max_size lines are processed, and the first line where
2054 all (if any) listed ACLs are true is used as the maximum body size
2057 This size is checked twice. First when we get the reply headers,
2058 we check the content-length value. If the content length value exists
2059 and is larger than the allowed size, the request is denied and the
2060 user receives an error message that says "the request or reply
2061 is too large." If there is no content-length, and the reply
2062 size exceeds this limit, the client's connection is just closed
2063 and they will receive a partial reply.
2065 WARNING: downstream caches probably can not detect a partial reply
2066 if there is no content-length header, so they will cache
2067 partial responses and give them out as hits. You should NOT
2068 use this option if you have downstream caches.
2070 WARNING: A maximum size smaller than the size of squid's error messages
2071 will cause an infinite loop and crash squid. Ensure that the smallest
2072 non-zero value you use is greater that the maximum header size plus
2073 the size of your largest error page.
2075 If you set this parameter none (the default), there will be
2078 Configuration Format is:
2079 reply_body_max_size SIZE UNITS [acl ...]
2081 reply_body_max_size 10 MB
2085 NAME: on_unsupported_protocol
2086 TYPE: on_unsupported_protocol
2087 LOC: Config.accessList.on_unsupported_protocol
2089 DEFAULT_DOC: Respond with an error message to unidentifiable traffic
2091 Determines Squid behavior when encountering strange requests at the
2092 beginning of an accepted TCP connection or the beginning of a bumped
2093 CONNECT tunnel. Controlling Squid reaction to unexpected traffic is
2094 especially useful in interception environments where Squid is likely
2095 to see connections for unsupported protocols that Squid should either
2096 terminate or tunnel at TCP level.
2098 on_unsupported_protocol <action> [!]acl ...
2100 The first matching action wins. Only fast ACLs are supported.
2102 Supported actions are:
2104 tunnel: Establish a TCP connection with the intended server and
2105 blindly shovel TCP packets between the client and server.
2107 respond: Respond with an error message, using the transfer protocol
2108 for the Squid port that received the request (e.g., HTTP
2109 for connections intercepted at the http_port). This is the
2112 Squid expects the following traffic patterns:
2114 http_port: a plain HTTP request
2115 https_port: SSL/TLS handshake followed by an [encrypted] HTTP request
2116 ftp_port: a plain FTP command (no on_unsupported_protocol support yet!)
2117 CONNECT tunnel on http_port: same as https_port
2118 CONNECT tunnel on https_port: same as https_port
2120 Currently, this directive has effect on intercepted connections and
2121 bumped tunnels only. Other cases are not supported because Squid
2122 cannot know the intended destination of other traffic.
2125 # define what Squid errors indicate receiving non-HTTP traffic:
2126 acl foreignProtocol squid_error ERR_PROTOCOL_UNKNOWN ERR_TOO_BIG
2127 # define what Squid errors indicate receiving nothing:
2128 acl serverTalksFirstProtocol squid_error ERR_REQUEST_START_TIMEOUT
2129 # tunnel everything that does not look like HTTP:
2130 on_unsupported_protocol tunnel foreignProtocol
2131 # tunnel if we think the client waits for the server to talk first:
2132 on_unsupported_protocol tunnel serverTalksFirstProtocol
2133 # in all other error cases, just send an HTTP "error page" response:
2134 on_unsupported_protocol respond all
2136 See also: squid_error ACL
2142 LOC: Auth::TheConfig.schemeAccess
2144 DEFAULT_DOC: use all auth_param schemes in their configuration order
2146 Use this directive to customize authentication schemes presence and
2147 order in Squid's Unauthorized and Authentication Required responses.
2149 auth_schemes scheme1,scheme2,... [!]aclname ...
2151 where schemeN is the name of one of the authentication schemes
2152 configured using auth_param directives. At least one scheme name is
2153 required. Multiple scheme names are separated by commas. Either
2154 avoid whitespace or quote the entire schemes list.
2156 A special "ALL" scheme name expands to all auth_param-configured
2157 schemes in their configuration order. This directive cannot be used
2158 to configure Squid to offer no authentication schemes at all.
2160 The first matching auth_schemes rule determines the schemes order
2161 for the current Authentication Required transaction. Note that the
2162 future response is not yet available during auth_schemes evaluation.
2164 If this directive is not used or none of its rules match, then Squid
2165 responds with all configured authentication schemes in the order of
2166 auth_param directives in the configuration file.
2168 This directive does not determine when authentication is used or
2169 how each authentication scheme authenticates clients.
2171 The following example sends basic and negotiate authentication
2172 schemes, in that order, when requesting authentication of HTTP
2173 requests matching the isIE ACL (not shown) while sending all
2174 auth_param schemes in their configuration order to other clients:
2176 auth_schemes basic,negotiate isIE
2177 auth_schemes ALL all # explicit default
2179 This directive supports fast ACLs only.
2181 See also: auth_param.
2186 -----------------------------------------------------------------------------
2189 NAME: http_port ascii_port
2194 Usage: port [mode] [options]
2195 hostname:port [mode] [options]
2196 1.2.3.4:port [mode] [options]
2198 The socket addresses where Squid will listen for HTTP client
2199 requests. You may specify multiple socket addresses.
2200 There are three forms: port alone, hostname with port, and
2201 IP address with port. If you specify a hostname or IP
2202 address, Squid binds the socket to that specific
2203 address. Most likely, you do not need to bind to a specific
2204 address, so you can use the port number alone.
2206 If you are running Squid in accelerator mode, you
2207 probably want to listen on port 80 also, or instead.
2209 The -a command line option may be used to specify additional
2210 port(s) where Squid listens for proxy request. Such ports will
2211 be plain proxy ports with no options.
2213 You may specify multiple socket addresses on multiple lines.
2217 intercept Support for IP-Layer NAT interception delivering
2218 traffic to this Squid port.
2219 NP: disables authentication on the port.
2221 tproxy Support Linux TPROXY (or BSD divert-to) with spoofing
2222 of outgoing connections using the client IP address.
2223 NP: disables authentication on the port.
2225 accel Accelerator / reverse proxy mode
2227 ssl-bump For each CONNECT request allowed by ssl_bump ACLs,
2228 establish secure connection with the client and with
2229 the server, decrypt HTTPS messages as they pass through
2230 Squid, and treat them as unencrypted HTTP messages,
2231 becoming the man-in-the-middle.
2233 The ssl_bump option is required to fully enable
2234 bumping of CONNECT requests.
2236 Omitting the mode flag causes default forward proxy mode to be used.
2239 Accelerator Mode Options:
2241 defaultsite=domainname
2242 What to use for the Host: header if it is not present
2243 in a request. Determines what site (not origin server)
2244 accelerators should consider the default.
2246 no-vhost Disable using HTTP/1.1 Host header for virtual domain support.
2248 protocol= Protocol to reconstruct accelerated and intercepted
2249 requests with. Defaults to HTTP/1.1 for http_port and
2250 HTTPS/1.1 for https_port.
2251 When an unsupported value is configured Squid will
2252 produce a FATAL error.
2253 Values: HTTP or HTTP/1.1, HTTPS or HTTPS/1.1
2255 vport Virtual host port support. Using the http_port number
2256 instead of the port passed on Host: headers.
2258 vport=NN Virtual host port support. Using the specified port
2259 number instead of the port passed on Host: headers.
2262 Act as if this Squid is the origin server.
2263 This currently means generate new Date: and Expires:
2264 headers on HIT instead of adding Age:.
2266 ignore-cc Ignore request Cache-Control headers.
2268 WARNING: This option violates HTTP specifications if
2269 used in non-accelerator setups.
2271 allow-direct Allow direct forwarding in accelerator mode. Normally
2272 accelerated requests are denied direct forwarding as if
2273 never_direct was used.
2275 WARNING: this option opens accelerator mode to security
2276 vulnerabilities usually only affecting in interception
2277 mode. Make sure to protect forwarding with suitable
2278 http_access rules when using this.
2281 SSL Bump Mode Options:
2282 In addition to these options ssl-bump requires TLS/SSL options.
2284 generate-host-certificates[=<on|off>]
2285 Dynamically create SSL server certificates for the
2286 destination hosts of bumped CONNECT requests.When
2287 enabled, the cert and key options are used to sign
2288 generated certificates. Otherwise generated
2289 certificate will be selfsigned.
2290 If there is a CA certificate lifetime of the generated
2291 certificate equals lifetime of the CA certificate. If
2292 generated certificate is selfsigned lifetime is three
2294 This option is enabled by default when ssl-bump is used.
2295 See the ssl-bump option above for more information.
2297 dynamic_cert_mem_cache_size=SIZE
2298 Approximate total RAM size spent on cached generated
2299 certificates. If set to zero, caching is disabled. The
2300 default value is 4MB.
2304 tls-cert= Path to file containing an X.509 certificate (PEM format)
2305 to be used in the TLS handshake ServerHello.
2307 If this certificate is constrained by KeyUsage TLS
2308 feature it must allow HTTP server usage, along with
2309 any additional restrictions imposed by your choice
2310 of options= settings.
2312 When OpenSSL is used this file may also contain a
2313 chain of intermediate CA certificates to send in the
2316 When GnuTLS is used this option (and any paired
2317 tls-key= option) may be repeated to load multiple
2318 certificates for different domains.
2320 Also, when generate-host-certificates=on is configured
2321 the first tls-cert= option must be a CA certificate
2322 capable of signing the automatically generated
2325 tls-key= Path to a file containing private key file (PEM format)
2326 for the previous tls-cert= option.
2328 If tls-key= is not specified tls-cert= is assumed to
2329 reference a PEM file containing both the certificate
2332 cipher= Colon separated list of supported ciphers.
2333 NOTE: some ciphers such as EDH ciphers depend on
2334 additional settings. If those settings are
2335 omitted the ciphers may be silently ignored
2336 by the OpenSSL library.
2338 options= Various SSL implementation options. The most important
2341 NO_SSLv3 Disallow the use of SSLv3
2343 NO_TLSv1 Disallow the use of TLSv1.0
2345 NO_TLSv1_1 Disallow the use of TLSv1.1
2347 NO_TLSv1_2 Disallow the use of TLSv1.2
2350 Always create a new key when using
2351 temporary/ephemeral DH key exchanges
2354 Enable ephemeral ECDH key exchange.
2355 The adopted curve should be specified
2356 using the tls-dh option.
2359 Disable use of RFC5077 session tickets.
2360 Some servers may have problems
2361 understanding the TLS extension due
2362 to ambiguous specification in RFC4507.
2364 ALL Enable various bug workarounds
2365 suggested as "harmless" by OpenSSL
2366 Be warned that this reduces SSL/TLS
2367 strength to some attacks.
2369 See the OpenSSL SSL_CTX_set_options documentation for a
2372 clientca= File containing the list of CAs to use when
2373 requesting a client certificate.
2375 tls-cafile= PEM file containing CA certificates to use when verifying
2376 client certificates. If not configured clientca will be
2377 used. May be repeated to load multiple files.
2379 capath= Directory containing additional CA certificates
2380 and CRL lists to use when verifying client certificates.
2381 Requires OpenSSL or LibreSSL.
2383 crlfile= File of additional CRL lists to use when verifying
2384 the client certificate, in addition to CRLs stored in
2385 the capath. Implies VERIFY_CRL flag below.
2388 File containing DH parameters for temporary/ephemeral DH key
2389 exchanges, optionally prefixed by a curve for ephemeral ECDH
2391 See OpenSSL documentation for details on how to create the
2392 DH parameter file. Supported curves for ECDH can be listed
2393 using the "openssl ecparam -list_curves" command.
2394 WARNING: EDH and EECDH ciphers will be silently disabled if
2395 this option is not set.
2397 sslflags= Various flags modifying the use of SSL:
2399 Don't request client certificates
2400 immediately, but wait until acl processing
2401 requires a certificate (not yet implemented).
2403 Request a client certificate during the TLS
2404 handshake, but ignore certificate absence in
2405 the TLS client Hello. If the client does
2406 supply a certificate, it is validated.
2408 Don't allow for session reuse. Each connection
2409 will result in a new SSL session.
2411 Verify CRL lists when accepting client
2414 Verify CRL lists for all certificates in the
2415 client certificate chain.
2417 tls-default-ca[=off]
2418 Whether to use the system Trusted CAs. Default is OFF.
2420 tls-no-npn Do not use the TLS NPN extension to advertise HTTP/1.1.
2422 sslcontext= SSL session ID context identifier.
2426 connection-auth[=on|off]
2427 use connection-auth=off to tell Squid to prevent
2428 forwarding Microsoft connection oriented authentication
2429 (NTLM, Negotiate and Kerberos)
2431 disable-pmtu-discovery=
2432 Control Path-MTU discovery usage:
2433 off lets OS decide on what to do (default).
2434 transparent disable PMTU discovery when transparent
2436 always disable always PMTU discovery.
2438 In many setups of transparently intercepting proxies
2439 Path-MTU discovery can not work on traffic towards the
2440 clients. This is the case when the intercepting device
2441 does not fully track connections and fails to forward
2442 ICMP must fragment messages to the cache server. If you
2443 have such setup and experience that certain clients
2444 sporadically hang or never complete requests set
2445 disable-pmtu-discovery option to 'transparent'.
2447 name= Specifies a internal name for the port. Defaults to
2448 the port specification (port or addr:port)
2450 tcpkeepalive[=idle,interval,timeout]
2451 Enable TCP keepalive probes of idle connections.
2452 In seconds; idle is the initial time before TCP starts
2453 probing the connection, interval how often to probe, and
2454 timeout the time before giving up.
2456 require-proxy-header
2457 Require PROXY protocol version 1 or 2 connections.
2458 The proxy_protocol_access is required to permit
2459 downstream proxies which can be trusted.
2462 Ask TCP stack to maintain a dedicated listening queue
2463 for each worker accepting requests at this port.
2464 Requires TCP stack that supports the SO_REUSEPORT socket
2467 SECURITY WARNING: Enabling worker-specific queues
2468 allows any process running as Squid's effective user to
2469 easily accept requests destined to this port.
2471 If you run Squid on a dual-homed machine with an internal
2472 and an external interface we recommend you to specify the
2473 internal address:port in http_port. This way Squid will only be
2474 visible on the internal address.
2478 # Squid normally listens to port 3128
2479 http_port @DEFAULT_HTTP_PORT@
2484 IFDEF: USE_GNUTLS||USE_OPENSSL
2489 Usage: [ip:]port [mode] tls-cert=certificate.pem [options]
2491 The socket address where Squid will listen for client requests made
2492 over TLS or SSL connections. Commonly referred to as HTTPS.
2494 This is most useful for situations where you are running squid in
2495 accelerator mode and you want to do the TLS work at the accelerator
2498 You may specify multiple socket addresses on multiple lines,
2499 each with their own certificate and/or options.
2501 The tls-cert= option is mandatory on HTTPS ports.
2503 See http_port for a list of modes and options.
2511 Enables Native FTP proxy by specifying the socket address where Squid
2512 listens for FTP client requests. See http_port directive for various
2513 ways to specify the listening address and mode.
2515 Usage: ftp_port address [mode] [options]
2517 WARNING: This is a new, experimental, complex feature that has seen
2518 limited production exposure. Some Squid modules (e.g., caching) do not
2519 currently work with native FTP proxying, and many features have not
2520 even been tested for compatibility. Test well before deploying!
2522 Native FTP proxying differs substantially from proxying HTTP requests
2523 with ftp:// URIs because Squid works as an FTP server and receives
2524 actual FTP commands (rather than HTTP requests with FTP URLs).
2526 Native FTP commands accepted at ftp_port are internally converted or
2527 wrapped into HTTP-like messages. The same happens to Native FTP
2528 responses received from FTP origin servers. Those HTTP-like messages
2529 are shoveled through regular access control and adaptation layers
2530 between the FTP client and the FTP origin server. This allows Squid to
2531 examine, adapt, block, and log FTP exchanges. Squid reuses most HTTP
2532 mechanisms when shoveling wrapped FTP messages. For example,
2533 http_access and adaptation_access directives are used.
2537 intercept Same as http_port intercept. The FTP origin address is
2538 determined based on the intended destination of the
2539 intercepted connection.
2541 tproxy Support Linux TPROXY for spoofing outgoing
2542 connections using the client IP address.
2543 NP: disables authentication and maybe IPv6 on the port.
2545 By default (i.e., without an explicit mode option), Squid extracts the
2546 FTP origin address from the login@origin parameter of the FTP USER
2547 command. Many popular FTP clients support such native FTP proxying.
2551 name=token Specifies an internal name for the port. Defaults to
2552 the port address. Usable with myportname ACL.
2555 Enables tracking of FTP directories by injecting extra
2556 PWD commands and adjusting Request-URI (in wrapping
2557 HTTP requests) to reflect the current FTP server
2558 directory. Tracking is disabled by default.
2560 protocol=FTP Protocol to reconstruct accelerated and intercepted
2561 requests with. Defaults to FTP. No other accepted
2562 values have been tested with. An unsupported value
2563 results in a FATAL error. Accepted values are FTP,
2564 HTTP (or HTTP/1.1), and HTTPS (or HTTPS/1.1).
2566 Other http_port modes and options that are not specific to HTTP and
2567 HTTPS may also work.
2570 NAME: tcp_outgoing_tos tcp_outgoing_ds tcp_outgoing_dscp
2573 LOC: Ip::Qos::TheConfig.tosToServer
2575 Allows you to select a TOS/Diffserv value for packets outgoing
2576 on the server side, based on an ACL.
2578 tcp_outgoing_tos ds-field [!]aclname ...
2580 Example where normal_service_net uses the TOS value 0x00
2581 and good_service_net uses 0x20
2583 acl normal_service_net src 10.0.0.0/24
2584 acl good_service_net src 10.0.1.0/24
2585 tcp_outgoing_tos 0x00 normal_service_net
2586 tcp_outgoing_tos 0x20 good_service_net
2588 TOS/DSCP values really only have local significance - so you should
2589 know what you're specifying. For more information, see RFC2474,
2590 RFC2475, and RFC3260.
2592 The TOS/DSCP byte must be exactly that - a octet value 0 - 255, or
2593 "default" to use whatever default your host has.
2594 Note that only multiples of 4 are usable as the two rightmost bits have
2595 been redefined for use by ECN (RFC 3168 section 23.1).
2596 The squid parser will enforce this by masking away the ECN bits.
2598 Processing proceeds in the order specified, and stops at first fully
2601 Only fast ACLs are supported.
2604 NAME: clientside_tos
2607 LOC: Ip::Qos::TheConfig.tosToClient
2609 Allows you to select a TOS/DSCP value for packets being transmitted
2610 on the client-side, based on an ACL.
2612 clientside_tos ds-field [!]aclname ...
2614 Example where normal_service_net uses the TOS value 0x00
2615 and good_service_net uses 0x20
2617 acl normal_service_net src 10.0.0.0/24
2618 acl good_service_net src 10.0.1.0/24
2619 clientside_tos 0x00 normal_service_net
2620 clientside_tos 0x20 good_service_net
2622 Note: This feature is incompatible with qos_flows. Any TOS values set here
2623 will be overwritten by TOS values in qos_flows.
2625 The TOS/DSCP byte must be exactly that - a octet value 0 - 255, or
2626 "default" to use whatever default your host has.
2627 Note that only multiples of 4 are usable as the two rightmost bits have
2628 been redefined for use by ECN (RFC 3168 section 23.1).
2629 The squid parser will enforce this by masking away the ECN bits.
2631 This clause only supports fast acl types.
2632 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2635 NAME: tcp_outgoing_mark
2637 IFDEF: SO_MARK&&USE_LIBCAP
2639 LOC: Ip::Qos::TheConfig.nfmarkToServer
2641 Allows you to apply a Netfilter mark value to outgoing packets
2642 on the server side, based on an ACL.
2644 tcp_outgoing_mark mark-value [!]aclname ...
2646 Example where normal_service_net uses the mark value 0x00
2647 and good_service_net uses 0x20
2649 acl normal_service_net src 10.0.0.0/24
2650 acl good_service_net src 10.0.1.0/24
2651 tcp_outgoing_mark 0x00 normal_service_net
2652 tcp_outgoing_mark 0x20 good_service_net
2654 Only fast ACLs are supported.
2657 NAME: mark_client_packet clientside_mark
2659 IFDEF: SO_MARK&&USE_LIBCAP
2661 LOC: Ip::Qos::TheConfig.nfmarkToClient
2663 Allows you to apply a Netfilter MARK value to packets being transmitted
2664 on the client-side, based on an ACL.
2666 mark_client_packet mark-value [!]aclname ...
2668 Example where normal_service_net uses the MARK value 0x00
2669 and good_service_net uses 0x20
2671 acl normal_service_net src 10.0.0.0/24
2672 acl good_service_net src 10.0.1.0/24
2673 mark_client_packet 0x00 normal_service_net
2674 mark_client_packet 0x20 good_service_net
2676 Note: This feature is incompatible with qos_flows. Any mark values set here
2677 will be overwritten by mark values in qos_flows.
2679 This clause only supports fast acl types.
2680 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2683 NAME: mark_client_connection
2685 IFDEF: SO_MARK&&USE_LIBCAP
2687 LOC: Ip::Qos::TheConfig.nfConnmarkToClient
2689 Allows you to apply a Netfilter CONNMARK value to a connection
2690 on the client-side, based on an ACL.
2692 mark_client_connection mark-value[/mask] [!]aclname ...
2694 The mark-value and mask are unsigned integers (hex, octal, or decimal).
2695 The mask may be used to preserve marking previously set by other agents
2698 A matching rule replaces the CONNMARK value. If a mask is also
2699 specified, then the masked bits of the original value are zeroed, and
2700 the configured mark-value is ORed with that adjusted value.
2701 For example, applying a mark-value 0xAB/0xF to 0x5F CONNMARK, results
2702 in a 0xFB marking (rather than a 0xAB or 0x5B).
2704 This directive semantics is similar to iptables --set-mark rather than
2705 --set-xmark functionality.
2707 The directive does not interfere with qos_flows (which uses packet MARKs,
2710 Example where squid marks intercepted FTP connections:
2712 acl proto_ftp proto FTP
2713 mark_client_connection 0x200/0xff00 proto_ftp
2715 This clause only supports fast acl types.
2716 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2723 LOC: Ip::Qos::TheConfig
2725 Allows you to select a TOS/DSCP value to mark outgoing
2726 connections to the client, based on where the reply was sourced.
2727 For platforms using netfilter, allows you to set a netfilter mark
2728 value instead of, or in addition to, a TOS value.
2730 By default this functionality is disabled. To enable it with the default
2731 settings simply use "qos_flows mark" or "qos_flows tos". Default
2732 settings will result in the netfilter mark or TOS value being copied
2733 from the upstream connection to the client. Note that it is the connection
2734 CONNMARK value not the packet MARK value that is copied.
2736 It is not currently possible to copy the mark or TOS value from the
2737 client to the upstream connection request.
2739 TOS values really only have local significance - so you should
2740 know what you're specifying. For more information, see RFC2474,
2741 RFC2475, and RFC3260.
2743 The TOS/DSCP byte must be exactly that - a octet value 0 - 255.
2744 Note that only multiples of 4 are usable as the two rightmost bits have
2745 been redefined for use by ECN (RFC 3168 section 23.1).
2746 The squid parser will enforce this by masking away the ECN bits.
2748 Mark values can be any unsigned 32-bit integer value.
2750 This setting is configured by setting the following values:
2752 tos|mark Whether to set TOS or netfilter mark values
2754 local-hit=0xFF Value to mark local cache hits.
2756 sibling-hit=0xFF Value to mark hits from sibling peers.
2758 parent-hit=0xFF Value to mark hits from parent peers.
2760 miss=0xFF[/mask] Value to mark cache misses. Takes precedence
2761 over the preserve-miss feature (see below), unless
2762 mask is specified, in which case only the bits
2763 specified in the mask are written.
2765 The TOS variant of the following features are only possible on Linux
2766 and require your kernel to be patched with the TOS preserving ZPH
2767 patch, available from http://zph.bratcheda.org
2768 No patch is needed to preserve the netfilter mark, which will work
2769 with all variants of netfilter.
2771 disable-preserve-miss
2772 This option disables the preservation of the TOS or netfilter
2773 mark. By default, the existing TOS or netfilter mark value of
2774 the response coming from the remote server will be retained
2775 and masked with miss-mark.
2776 NOTE: in the case of a netfilter mark, the mark must be set on
2777 the connection (using the CONNMARK target) not on the packet
2781 Allows you to mask certain bits in the TOS or mark value
2782 received from the remote server, before copying the value to
2783 the TOS sent towards clients.
2784 Default for tos: 0xFF (TOS from server is not changed).
2785 Default for mark: 0xFFFFFFFF (mark from server is not changed).
2787 All of these features require the --enable-zph-qos compilation flag
2788 (enabled by default). Netfilter marking also requires the
2789 libnetfilter_conntrack libraries (--with-netfilter-conntrack) and
2790 libcap 2.09+ (--with-libcap).
2794 NAME: tcp_outgoing_address
2797 DEFAULT_DOC: Address selection is performed by the operating system.
2798 LOC: Config.accessList.outgoing_address
2800 Allows you to map requests to different outgoing IP addresses
2801 based on the username or source address of the user making
2804 tcp_outgoing_address ipaddr [[!]aclname] ...
2807 Forwarding clients with dedicated IPs for certain subnets.
2809 acl normal_service_net src 10.0.0.0/24
2810 acl good_service_net src 10.0.2.0/24
2812 tcp_outgoing_address 2001:db8::c001 good_service_net
2813 tcp_outgoing_address 10.1.0.2 good_service_net
2815 tcp_outgoing_address 2001:db8::beef normal_service_net
2816 tcp_outgoing_address 10.1.0.1 normal_service_net
2818 tcp_outgoing_address 2001:db8::1
2819 tcp_outgoing_address 10.1.0.3
2821 Processing proceeds in the order specified, and stops at first fully
2824 Squid will add an implicit IP version test to each line.
2825 Requests going to IPv4 websites will use the outgoing 10.1.0.* addresses.
2826 Requests going to IPv6 websites will use the outgoing 2001:db8:* addresses.
2829 NOTE: The use of this directive using client dependent ACLs is
2830 incompatible with the use of server side persistent connections. To
2831 ensure correct results it is best to set server_persistent_connections
2832 to off when using this directive in such configurations.
2834 NOTE: The use of this directive to set a local IP on outgoing TCP links
2835 is incompatible with using TPROXY to set client IP out outbound TCP links.
2836 When needing to contact peers use the no-tproxy cache_peer option and the
2837 client_dst_passthru directive re-enable normal forwarding such as this.
2839 This clause only supports fast acl types.
2840 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2843 NAME: host_verify_strict
2846 LOC: Config.onoff.hostStrictVerify
2848 Regardless of this option setting, when dealing with intercepted
2849 traffic, Squid always verifies that the destination IP address matches
2850 the Host header domain or IP (called 'authority form URL').
2852 This enforcement is performed to satisfy a MUST-level requirement in
2853 RFC 2616 section 14.23: "The Host field value MUST represent the naming
2854 authority of the origin server or gateway given by the original URL".
2857 Squid always responds with an HTTP 409 (Conflict) error
2858 page and logs a security warning if there is no match.
2860 Squid verifies that the destination IP address matches
2861 the Host header for forward-proxy and reverse-proxy traffic
2862 as well. For those traffic types, Squid also enables the
2863 following checks, comparing the corresponding Host header
2864 and Request-URI components:
2866 * The host names (domain or IP) must be identical,
2867 but valueless or missing Host header disables all checks.
2868 For the two host names to match, both must be either IP
2871 * Port numbers must be identical, but if a port is missing
2872 the scheme-default port is assumed.
2875 When set to OFF (the default):
2876 Squid allows suspicious requests to continue but logs a
2877 security warning and blocks caching of the response.
2879 * Forward-proxy traffic is not checked at all.
2881 * Reverse-proxy traffic is not checked at all.
2883 * Intercepted traffic which passes verification is handled
2884 according to client_dst_passthru.
2886 * Intercepted requests which fail verification are sent
2887 to the client original destination instead of DIRECT.
2888 This overrides 'client_dst_passthru off'.
2890 For now suspicious intercepted CONNECT requests are always
2891 responded to with an HTTP 409 (Conflict) error page.
2896 As described in CVE-2009-0801 when the Host: header alone is used
2897 to determine the destination of a request it becomes trivial for
2898 malicious scripts on remote websites to bypass browser same-origin
2899 security policy and sandboxing protections.
2901 The cause of this is that such applets are allowed to perform their
2902 own HTTP stack, in which case the same-origin policy of the browser
2903 sandbox only verifies that the applet tries to contact the same IP
2904 as from where it was loaded at the IP level. The Host: header may
2905 be different from the connected IP and approved origin.
2909 NAME: client_dst_passthru
2912 LOC: Config.onoff.client_dst_passthru
2914 With NAT or TPROXY intercepted traffic Squid may pass the request
2915 directly to the original client destination IP or seek a faster
2916 source using the HTTP Host header.
2918 Using Host to locate alternative servers can provide faster
2919 connectivity with a range of failure recovery options.
2920 But can also lead to connectivity trouble when the client and
2921 server are attempting stateful interactions unaware of the proxy.
2923 This option (on by default) prevents alternative DNS entries being
2924 located to send intercepted traffic DIRECT to an origin server.
2925 The clients original destination IP and port will be used instead.
2927 Regardless of this option setting, when dealing with intercepted
2928 traffic Squid will verify the Host: header and any traffic which
2929 fails Host verification will be treated as if this option were ON.
2931 see host_verify_strict for details on the verification process.
2936 -----------------------------------------------------------------------------
2939 NAME: tls_outgoing_options
2940 IFDEF: USE_GNUTLS||USE_OPENSSL
2941 TYPE: securePeerOptions
2942 DEFAULT: min-version=1.0
2943 LOC: Security::ProxyOutgoingConfig
2945 disable Do not support https:// URLs.
2947 cert=/path/to/client/certificate
2948 A client X.509 certificate to use when connecting.
2950 key=/path/to/client/private_key
2951 The private key corresponding to the cert= above.
2953 If key= is not specified cert= is assumed to
2954 reference a PEM file containing both the certificate
2957 cipher=... The list of valid TLS ciphers to use.
2960 The minimum TLS protocol version to permit.
2961 To control SSLv3 use the options= parameter.
2962 Supported Values: 1.0 (default), 1.1, 1.2, 1.3
2964 options=... Specify various TLS/SSL implementation options.
2966 OpenSSL options most important are:
2968 NO_SSLv3 Disallow the use of SSLv3
2971 Always create a new key when using
2972 temporary/ephemeral DH key exchanges
2975 Disable use of RFC5077 session tickets.
2976 Some servers may have problems
2977 understanding the TLS extension due
2978 to ambiguous specification in RFC4507.
2980 ALL Enable various bug workarounds
2981 suggested as "harmless" by OpenSSL
2982 Be warned that this reduces SSL/TLS
2983 strength to some attacks.
2985 See the OpenSSL SSL_CTX_set_options documentation
2986 for a more complete list.
2988 GnuTLS options most important are:
2991 Disable use of RFC5077 session tickets.
2992 Some servers may have problems
2993 understanding the TLS extension due
2994 to ambiguous specification in RFC4507.
2996 See the GnuTLS Priority Strings documentation
2997 for a more complete list.
2998 http://www.gnutls.org/manual/gnutls.html#Priority-Strings
3001 cafile= PEM file containing CA certificates to use when verifying
3002 the peer certificate. May be repeated to load multiple files.
3004 capath= A directory containing additional CA certificates to
3005 use when verifying the peer certificate.
3006 Requires OpenSSL or LibreSSL.
3008 crlfile=... A certificate revocation list file to use when
3009 verifying the peer certificate.
3011 flags=... Specify various flags modifying the TLS implementation:
3014 Accept certificates even if they fail to
3017 Don't verify the peer certificate
3018 matches the server name
3021 Whether to use the system Trusted CAs. Default is ON.
3023 domain= The peer name as advertised in its certificate.
3024 Used for verifying the correctness of the received peer
3025 certificate. If not specified the peer hostname will be
3031 -----------------------------------------------------------------------------
3034 NAME: ssl_unclean_shutdown
3038 LOC: Config.SSL.unclean_shutdown
3040 Some browsers (especially MSIE) bugs out on SSL shutdown
3047 LOC: Config.SSL.ssl_engine
3050 The OpenSSL engine to use. You will need to set this if you
3051 would like to use hardware SSL acceleration for example.
3054 NAME: sslproxy_session_ttl
3057 LOC: Config.SSL.session_ttl
3060 Sets the timeout value for SSL sessions
3063 NAME: sslproxy_session_cache_size
3066 LOC: Config.SSL.sessionCacheSize
3069 Sets the cache size to use for ssl session
3072 NAME: sslproxy_foreign_intermediate_certs
3075 LOC: Config.ssl_client.foreignIntermediateCertsPath
3078 Many origin servers fail to send their full server certificate
3079 chain for verification, assuming the client already has or can
3080 easily locate any missing intermediate certificates.
3082 Squid uses the certificates from the specified file to fill in
3083 these missing chains when trying to validate origin server
3086 The file is expected to contain zero or more PEM-encoded
3087 intermediate certificates. These certificates are not treated
3088 as trusted root certificates, and any self-signed certificate in
3089 this file will be ignored.
3092 NAME: sslproxy_cert_sign_hash
3095 LOC: Config.SSL.certSignHash
3098 Sets the hashing algorithm to use when signing generated certificates.
3099 Valid algorithm names depend on the OpenSSL library used. The following
3100 names are usually available: sha1, sha256, sha512, and md5. Please see
3101 your OpenSSL library manual for the available hashes. By default, Squids
3102 that support this option use sha256 hashes.
3104 Squid does not forcefully purge cached certificates that were generated
3105 with an algorithm other than the currently configured one. They remain
3106 in the cache, subject to the regular cache eviction policy, and become
3107 useful if the algorithm changes again.
3112 TYPE: sslproxy_ssl_bump
3113 LOC: Config.accessList.ssl_bump
3114 DEFAULT_DOC: Become a TCP tunnel without decrypting proxied traffic.
3117 This option is consulted when a CONNECT request is received on
3118 an http_port (or a new connection is intercepted at an
3119 https_port), provided that port was configured with an ssl-bump
3120 flag. The subsequent data on the connection is either treated as
3121 HTTPS and decrypted OR tunneled at TCP level without decryption,
3122 depending on the first matching bumping "action".
3124 ssl_bump <action> [!]acl ...
3126 The following bumping actions are currently supported:
3129 Become a TCP tunnel without decrypting proxied traffic.
3130 This is the default action.
3133 When used on step SslBump1, establishes a secure connection
3134 with the client first, then connect to the server.
3135 When used on step SslBump2 or SslBump3, establishes a secure
3136 connection with the server and, using a mimicked server
3137 certificate, with the client.
3140 Receive client (step SslBump1) or server (step SslBump2)
3141 certificate while preserving the possibility of splicing the
3142 connection. Peeking at the server certificate (during step 2)
3143 usually precludes bumping of the connection at step 3.
3146 Receive client (step SslBump1) or server (step SslBump2)
3147 certificate while preserving the possibility of bumping the
3148 connection. Staring at the server certificate (during step 2)
3149 usually precludes splicing of the connection at step 3.
3152 Close client and server connections.
3154 Backward compatibility actions available at step SslBump1:
3157 Bump the connection. Establish a secure connection with the
3158 client first, then connect to the server. This old mode does
3159 not allow Squid to mimic server SSL certificate and does not
3160 work with intercepted SSL connections.
3163 Bump the connection. Establish a secure connection with the
3164 server first, then establish a secure connection with the
3165 client, using a mimicked server certificate. Works with both
3166 CONNECT requests and intercepted SSL connections, but does
3167 not allow to make decisions based on SSL handshake info.
3170 Decide whether to bump or splice the connection based on
3171 client-to-squid and server-to-squid SSL hello messages.
3175 Same as the "splice" action.
3177 All ssl_bump rules are evaluated at each of the supported bumping
3178 steps. Rules with actions that are impossible at the current step are
3179 ignored. The first matching ssl_bump action wins and is applied at the
3180 end of the current step. If no rules match, the splice action is used.
3181 See the at_step ACL for a list of the supported SslBump steps.
3183 This clause supports both fast and slow acl types.
3184 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
3186 See also: http_port ssl-bump, https_port ssl-bump, and acl at_step.
3189 # Example: Bump all TLS connections except those originating from
3190 # localhost or those going to example.com.
3192 acl broken_sites ssl::server_name .example.com
3193 ssl_bump splice localhost
3194 ssl_bump splice broken_sites
3198 NAME: sslproxy_cert_error
3201 DEFAULT_DOC: Server certificate errors terminate the transaction.
3202 LOC: Config.ssl_client.cert_error
3205 Use this ACL to bypass server certificate validation errors.
3207 For example, the following lines will bypass all validation errors
3208 when talking to servers for example.com. All other
3209 validation errors will result in ERR_SECURE_CONNECT_FAIL error.
3211 acl BrokenButTrustedServers dstdomain example.com
3212 sslproxy_cert_error allow BrokenButTrustedServers
3213 sslproxy_cert_error deny all
3215 This clause only supports fast acl types.
3216 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
3217 Using slow acl types may result in server crashes
3219 Without this option, all server certificate validation errors
3220 terminate the transaction to protect Squid and the client.
3222 SQUID_X509_V_ERR_INFINITE_VALIDATION error cannot be bypassed
3223 but should not happen unless your OpenSSL library is buggy.
3226 Bypassing validation errors is dangerous because an
3227 error usually implies that the server cannot be trusted
3228 and the connection may be insecure.
3230 See also: sslproxy_flags and DONT_VERIFY_PEER.
3233 NAME: sslproxy_cert_sign
3236 POSTSCRIPTUM: signUntrusted ssl::certUntrusted
3237 POSTSCRIPTUM: signSelf ssl::certSelfSigned
3238 POSTSCRIPTUM: signTrusted all
3239 TYPE: sslproxy_cert_sign
3240 LOC: Config.ssl_client.cert_sign
3243 sslproxy_cert_sign <signing algorithm> acl ...
3245 The following certificate signing algorithms are supported:
3248 Sign using the configured CA certificate which is usually
3249 placed in and trusted by end-user browsers. This is the
3250 default for trusted origin server certificates.
3253 Sign to guarantee an X509_V_ERR_CERT_UNTRUSTED browser error.
3254 This is the default for untrusted origin server certificates
3255 that are not self-signed (see ssl::certUntrusted).
3258 Sign using a self-signed certificate with the right CN to
3259 generate a X509_V_ERR_DEPTH_ZERO_SELF_SIGNED_CERT error in the
3260 browser. This is the default for self-signed origin server
3261 certificates (see ssl::certSelfSigned).
3263 This clause only supports fast acl types.
3265 When sslproxy_cert_sign acl(s) match, Squid uses the corresponding
3266 signing algorithm to generate the certificate and ignores all
3267 subsequent sslproxy_cert_sign options (the first match wins). If no
3268 acl(s) match, the default signing algorithm is determined by errors
3269 detected when obtaining and validating the origin server certificate.
3271 WARNING: SQUID_X509_V_ERR_DOMAIN_MISMATCH and ssl:certDomainMismatch can
3272 be used with sslproxy_cert_adapt, but if and only if Squid is bumping a
3273 CONNECT request that carries a domain name. In all other cases (CONNECT
3274 to an IP address or an intercepted SSL connection), Squid cannot detect
3275 the domain mismatch at certificate generation time when
3276 bump-server-first is used.
3279 NAME: sslproxy_cert_adapt
3282 TYPE: sslproxy_cert_adapt
3283 LOC: Config.ssl_client.cert_adapt
3286 sslproxy_cert_adapt <adaptation algorithm> acl ...
3288 The following certificate adaptation algorithms are supported:
3291 Sets the "Not After" property to the "Not After" property of
3292 the CA certificate used to sign generated certificates.
3295 Sets the "Not Before" property to the "Not Before" property of
3296 the CA certificate used to sign generated certificates.
3298 setCommonName or setCommonName{CN}
3299 Sets Subject.CN property to the host name specified as a
3300 CN parameter or, if no explicit CN parameter was specified,
3301 extracted from the CONNECT request. It is a misconfiguration
3302 to use setCommonName without an explicit parameter for
3303 intercepted or tproxied SSL connections.
3305 This clause only supports fast acl types.
3307 Squid first groups sslproxy_cert_adapt options by adaptation algorithm.
3308 Within a group, when sslproxy_cert_adapt acl(s) match, Squid uses the
3309 corresponding adaptation algorithm to generate the certificate and
3310 ignores all subsequent sslproxy_cert_adapt options in that algorithm's
3311 group (i.e., the first match wins within each algorithm group). If no
3312 acl(s) match, the default mimicking action takes place.
3314 WARNING: SQUID_X509_V_ERR_DOMAIN_MISMATCH and ssl:certDomainMismatch can
3315 be used with sslproxy_cert_adapt, but if and only if Squid is bumping a
3316 CONNECT request that carries a domain name. In all other cases (CONNECT
3317 to an IP address or an intercepted SSL connection), Squid cannot detect
3318 the domain mismatch at certificate generation time when
3319 bump-server-first is used.
3322 NAME: sslpassword_program
3325 LOC: Config.Program.ssl_password
3328 Specify a program used for entering SSL key passphrases
3329 when using encrypted SSL certificate keys. If not specified
3330 keys must either be unencrypted, or Squid started with the -N
3331 option to allow it to query interactively for the passphrase.
3333 The key file name is given as argument to the program allowing
3334 selection of the right password if you have multiple encrypted
3339 OPTIONS RELATING TO EXTERNAL SSL_CRTD
3340 -----------------------------------------------------------------------------
3343 NAME: sslcrtd_program
3346 DEFAULT: @DEFAULT_SSL_CRTD@ -s @DEFAULT_SSL_DB_DIR@ -M 4MB
3347 LOC: Ssl::TheConfig.ssl_crtd
3349 Specify the location and options of the executable for certificate
3352 @DEFAULT_SSL_CRTD@ program can use a disk cache to improve response
3353 times on repeated requests. To enable caching, specify -s and -M
3354 parameters. If those parameters are not given, the program generates
3355 a new certificate on every request.
3357 For more information use:
3358 @DEFAULT_SSL_CRTD@ -h
3361 NAME: sslcrtd_children
3362 TYPE: HelperChildConfig
3364 DEFAULT: 32 startup=5 idle=1
3365 LOC: Ssl::TheConfig.ssl_crtdChildren
3367 Specifies the maximum number of certificate generation processes that
3368 Squid may spawn (numberofchildren) and several related options. Using
3369 too few of these helper processes (a.k.a. "helpers") creates request
3370 queues. Using too many helpers wastes your system resources. Squid
3371 does not support spawning more than 32 helpers.
3373 Usage: numberofchildren [option]...
3375 The startup= and idle= options allow some measure of skew in your
3380 Sets the minimum number of processes to spawn when Squid
3381 starts or reconfigures. When set to zero the first request will
3382 cause spawning of the first child process to handle it.
3384 Starting too few children temporary slows Squid under load while it
3385 tries to spawn enough additional processes to cope with traffic.
3389 Sets a minimum of how many processes Squid is to try and keep available
3390 at all times. When traffic begins to rise above what the existing
3391 processes can handle this many more will be spawned up to the maximum
3392 configured. A minimum setting of 1 is required.
3396 Sets the maximum number of queued requests. A request is queued when
3397 no existing child is idle and no new child can be started due to
3398 numberofchildren limit. If the queued requests exceed queue size for
3399 more than 3 minutes squid aborts its operation. The default value is
3400 set to 2*numberofchildren.
3402 You must have at least one ssl_crtd process.
3405 NAME: sslcrtvalidator_program
3409 LOC: Ssl::TheConfig.ssl_crt_validator
3411 Specify the location and options of the executable for ssl_crt_validator
3414 Usage: sslcrtvalidator_program [ttl=...] [cache=n] path ...
3419 Limits how much memory Squid can use for caching validator
3420 responses. The default is 67108864 (i.e. 64 MB).
3421 Reconfiguration purges any excess entries. To disable caching,
3422 use cache=0. Currently, cache entry sizes are seriously
3423 underestimated. Even with that bug, a typical estimate for a
3424 single cache entry size would be at least a few kilobytes (the
3425 size of the PEM certificates sent to the validator).
3427 ttl=<seconds|"infinity">
3428 Approximately how long Squid may reuse the validator results
3429 for. The default is 3600 (i.e. 1 hour). Using ttl=infinity
3430 disables TTL checks. Reconfiguration does not affect TTLs of
3431 the already cached entries. To disable caching, use zero cache
3432 size, not zero TTL -- zero TTL allows reuse for the remainder
3433 of the second when the result was cached.
3436 NAME: sslcrtvalidator_children
3437 TYPE: HelperChildConfig
3439 DEFAULT: 32 startup=5 idle=1 concurrency=1
3440 LOC: Ssl::TheConfig.ssl_crt_validator_Children
3442 Specifies the maximum number of certificate validation processes that
3443 Squid may spawn (numberofchildren) and several related options. Using
3444 too few of these helper processes (a.k.a. "helpers") creates request
3445 queues. Using too many helpers wastes your system resources. Squid
3446 does not support spawning more than 32 helpers.
3448 Usage: numberofchildren [option]...
3450 The startup= and idle= options allow some measure of skew in your
3455 Sets the minimum number of processes to spawn when Squid
3456 starts or reconfigures. When set to zero the first request will
3457 cause spawning of the first child process to handle it.
3459 Starting too few children temporary slows Squid under load while it
3460 tries to spawn enough additional processes to cope with traffic.
3464 Sets a minimum of how many processes Squid is to try and keep available
3465 at all times. When traffic begins to rise above what the existing
3466 processes can handle this many more will be spawned up to the maximum
3467 configured. A minimum setting of 1 is required.
3471 The number of requests each certificate validator helper can handle in
3472 parallel. A value of 0 indicates the certificate validator does not
3473 support concurrency. Defaults to 1.
3475 When this directive is set to a value >= 1 then the protocol
3476 used to communicate with the helper is modified to include
3477 a request ID in front of the request/response. The request
3478 ID from the request must be echoed back with the response
3483 Sets the maximum number of queued requests. A request is queued when
3484 no existing child can accept it due to concurrency limit and no new
3485 child can be started due to numberofchildren limit. If the queued
3486 requests exceed queue size for more than 3 minutes squid aborts its
3487 operation. The default value is set to 2*numberofchildren.
3489 You must have at least one ssl_crt_validator process.
3493 OPTIONS WHICH AFFECT THE NEIGHBOR SELECTION ALGORITHM
3494 -----------------------------------------------------------------------------
3502 To specify other caches in a hierarchy, use the format:
3504 cache_peer hostname type http-port icp-port [options]
3509 # hostname type port port options
3510 # -------------------- -------- ----- ----- -----------
3511 cache_peer parent.foo.net parent 3128 3130 default
3512 cache_peer sib1.foo.net sibling 3128 3130 proxy-only
3513 cache_peer sib2.foo.net sibling 3128 3130 proxy-only
3514 cache_peer example.com parent 80 0 default
3515 cache_peer cdn.example.com sibling 3128 0
3517 type: either 'parent', 'sibling', or 'multicast'.
3519 proxy-port: The port number where the peer accept HTTP requests.
3520 For other Squid proxies this is usually 3128
3521 For web servers this is usually 80
3523 icp-port: Used for querying neighbor caches about objects.
3524 Set to 0 if the peer does not support ICP or HTCP.
3525 See ICP and HTCP options below for additional details.
3528 ==== ICP OPTIONS ====
3530 You MUST also set icp_port and icp_access explicitly when using these options.
3531 The defaults will prevent peer traffic using ICP.
3534 no-query Disable ICP queries to this neighbor.
3537 Indicates the named peer is a member of a multicast group.
3538 ICP queries will not be sent directly to the peer, but ICP
3539 replies will be accepted from it.
3541 closest-only Indicates that, for ICP_OP_MISS replies, we'll only forward
3542 CLOSEST_PARENT_MISSes and never FIRST_PARENT_MISSes.
3545 To only send ICP queries to this neighbor infrequently.
3546 This is used to keep the neighbor round trip time updated
3547 and is usually used in conjunction with weighted-round-robin.
3550 ==== HTCP OPTIONS ====
3552 You MUST also set htcp_port and htcp_access explicitly when using these options.
3553 The defaults will prevent peer traffic using HTCP.
3556 htcp Send HTCP, instead of ICP, queries to the neighbor.
3557 You probably also want to set the "icp-port" to 4827
3558 instead of 3130. This directive accepts a comma separated
3559 list of options described below.
3561 htcp=oldsquid Send HTCP to old Squid versions (2.5 or earlier).
3563 htcp=no-clr Send HTCP to the neighbor but without
3564 sending any CLR requests. This cannot be used with
3567 htcp=only-clr Send HTCP to the neighbor but ONLY CLR requests.
3568 This cannot be used with no-clr.
3571 Send HTCP to the neighbor including CLRs but only when
3572 they do not result from PURGE requests.
3575 Forward any HTCP CLR requests this proxy receives to the peer.
3578 ==== PEER SELECTION METHODS ====
3580 The default peer selection method is ICP, with the first responding peer
3581 being used as source. These options can be used for better load balancing.
3584 default This is a parent cache which can be used as a "last-resort"
3585 if a peer cannot be located by any of the peer-selection methods.
3586 If specified more than once, only the first is used.
3588 round-robin Load-Balance parents which should be used in a round-robin
3589 fashion in the absence of any ICP queries.
3590 weight=N can be used to add bias.
3592 weighted-round-robin
3593 Load-Balance parents which should be used in a round-robin
3594 fashion with the frequency of each parent being based on the
3595 round trip time. Closer parents are used more often.
3596 Usually used for background-ping parents.
3597 weight=N can be used to add bias.
3599 carp Load-Balance parents which should be used as a CARP array.
3600 The requests will be distributed among the parents based on the
3601 CARP load balancing hash function based on their weight.
3603 userhash Load-balance parents based on the client proxy_auth or ident username.
3605 sourcehash Load-balance parents based on the client source IP.
3608 To be used only for cache peers of type "multicast".
3609 ALL members of this multicast group have "sibling"
3610 relationship with it, not "parent". This is to a multicast
3611 group when the requested object would be fetched only from
3612 a "parent" cache, anyway. It's useful, e.g., when
3613 configuring a pool of redundant Squid proxies, being
3614 members of the same multicast group.
3617 ==== PEER SELECTION OPTIONS ====
3619 weight=N use to affect the selection of a peer during any weighted
3620 peer-selection mechanisms.
3621 The weight must be an integer; default is 1,
3622 larger weights are favored more.
3623 This option does not affect parent selection if a peering
3624 protocol is not in use.
3626 basetime=N Specify a base amount to be subtracted from round trip
3628 It is subtracted before division by weight in calculating
3629 which parent to fectch from. If the rtt is less than the
3630 base time the rtt is set to a minimal value.
3632 ttl=N Specify a TTL to use when sending multicast ICP queries
3634 Only useful when sending to a multicast group.
3635 Because we don't accept ICP replies from random
3636 hosts, you must configure other group members as
3637 peers with the 'multicast-responder' option.
3639 no-delay To prevent access to this neighbor from influencing the
3642 digest-url=URL Tell Squid to fetch the cache digest (if digests are
3643 enabled) for this host from the specified URL rather
3644 than the Squid default location.
3647 ==== CARP OPTIONS ====
3649 carp-key=key-specification
3650 use a different key than the full URL to hash against the peer.
3651 the key-specification is a comma-separated list of the keywords
3652 scheme, host, port, path, params
3653 Order is not important.
3655 ==== ACCELERATOR / REVERSE-PROXY OPTIONS ====
3657 originserver Causes this parent to be contacted as an origin server.
3658 Meant to be used in accelerator setups when the peer
3662 Set the Host header of requests forwarded to this peer.
3663 Useful in accelerator setups where the server (peer)
3664 expects a certain domain name but clients may request
3665 others. ie example.com or www.example.com
3667 no-digest Disable request of cache digests.
3670 Disables requesting ICMP RTT database (NetDB).
3673 ==== AUTHENTICATION OPTIONS ====
3676 If this is a personal/workgroup proxy and your parent
3677 requires proxy authentication.
3679 Note: The string can include URL escapes (i.e. %20 for
3680 spaces). This also means % must be written as %%.
3683 Send login details received from client to this peer.
3684 Both Proxy- and WWW-Authorization headers are passed
3685 without alteration to the peer.
3686 Authentication is not required by Squid for this to work.
3688 Note: This will pass any form of authentication but
3689 only Basic auth will work through a proxy unless the
3690 connection-auth options are also used.
3692 login=PASS Send login details received from client to this peer.
3693 Authentication is not required by this option.
3695 If there are no client-provided authentication headers
3696 to pass on, but username and password are available
3697 from an external ACL user= and password= result tags
3698 they may be sent instead.
3700 Note: To combine this with proxy_auth both proxies must
3701 share the same user database as HTTP only allows for
3702 a single login (one for proxy, one for origin server).
3703 Also be warned this will expose your users proxy
3704 password to the peer. USE WITH CAUTION
3707 Send the username to the upstream cache, but with a
3708 fixed password. This is meant to be used when the peer
3709 is in another administrative domain, but it is still
3710 needed to identify each user.
3711 The star can optionally be followed by some extra
3712 information which is added to the username. This can
3713 be used to identify this proxy to the peer, similar to
3714 the login=username:password option above.
3717 If this is a personal/workgroup proxy and your parent
3718 requires a secure proxy authentication.
3719 The first principal from the default keytab or defined by
3720 the environment variable KRB5_KTNAME will be used.
3722 WARNING: The connection may transmit requests from multiple
3723 clients. Negotiate often assumes end-to-end authentication
3724 and a single-client. Which is not strictly true here.
3726 login=NEGOTIATE:principal_name
3727 If this is a personal/workgroup proxy and your parent
3728 requires a secure proxy authentication.
3729 The principal principal_name from the default keytab or
3730 defined by the environment variable KRB5_KTNAME will be
3733 WARNING: The connection may transmit requests from multiple
3734 clients. Negotiate often assumes end-to-end authentication
3735 and a single-client. Which is not strictly true here.
3737 connection-auth=on|off
3738 Tell Squid that this peer does or not support Microsoft
3739 connection oriented authentication, and any such
3740 challenges received from there should be ignored.
3741 Default is auto to automatically determine the status
3745 Do not use a keytab to authenticate to a peer when
3746 login=NEGOTIATE is specified. Let the GSSAPI
3747 implementation determine which already existing
3748 credentials cache to use instead.
3751 ==== SSL / HTTPS / TLS OPTIONS ====
3753 tls Encrypt connections to this peer with TLS.
3755 sslcert=/path/to/ssl/certificate
3756 A client X.509 certificate to use when connecting to
3759 sslkey=/path/to/ssl/key
3760 The private key corresponding to sslcert above.
3762 If sslkey= is not specified sslcert= is assumed to
3763 reference a PEM file containing both the certificate
3766 sslcipher=... The list of valid SSL ciphers to use when connecting
3770 The minimum TLS protocol version to permit. To control
3771 SSLv3 use the tls-options= parameter.
3772 Supported Values: 1.0 (default), 1.1, 1.2
3774 tls-options=... Specify various TLS implementation options.
3776 OpenSSL options most important are:
3778 NO_SSLv3 Disallow the use of SSLv3
3781 Always create a new key when using
3782 temporary/ephemeral DH key exchanges
3785 Disable use of RFC5077 session tickets.
3786 Some servers may have problems
3787 understanding the TLS extension due
3788 to ambiguous specification in RFC4507.
3790 ALL Enable various bug workarounds
3791 suggested as "harmless" by OpenSSL
3792 Be warned that this reduces SSL/TLS
3793 strength to some attacks.
3795 See the OpenSSL SSL_CTX_set_options documentation for a
3798 GnuTLS options most important are:
3801 Disable use of RFC5077 session tickets.
3802 Some servers may have problems
3803 understanding the TLS extension due
3804 to ambiguous specification in RFC4507.
3806 See the GnuTLS Priority Strings documentation
3807 for a more complete list.
3808 http://www.gnutls.org/manual/gnutls.html#Priority-Strings
3810 tls-cafile= PEM file containing CA certificates to use when verifying
3811 the peer certificate. May be repeated to load multiple files.
3813 sslcapath=... A directory containing additional CA certificates to
3814 use when verifying the peer certificate.
3815 Requires OpenSSL or LibreSSL.
3817 sslcrlfile=... A certificate revocation list file to use when
3818 verifying the peer certificate.
3820 sslflags=... Specify various flags modifying the SSL implementation:
3823 Accept certificates even if they fail to
3827 Don't verify the peer certificate
3828 matches the server name
3830 ssldomain= The peer name as advertised in it's certificate.
3831 Used for verifying the correctness of the received peer
3832 certificate. If not specified the peer hostname will be
3835 front-end-https[=off|on|auto]
3836 Enable the "Front-End-Https: On" header needed when
3837 using Squid as a SSL frontend in front of Microsoft OWA.
3838 See MS KB document Q307347 for details on this header.
3839 If set to auto the header will only be added if the
3840 request is forwarded as a https:// URL.
3842 tls-default-ca[=off]
3843 Whether to use the system Trusted CAs. Default is ON.
3845 tls-no-npn Do not use the TLS NPN extension to advertise HTTP/1.1.
3847 ==== GENERAL OPTIONS ====
3850 A peer-specific connect timeout.
3851 Also see the peer_connect_timeout directive.
3853 connect-fail-limit=N
3854 How many times connecting to a peer must fail before
3855 it is marked as down. Standby connection failures
3856 count towards this limit. Default is 10.
3858 allow-miss Disable Squid's use of only-if-cached when forwarding
3859 requests to siblings. This is primarily useful when
3860 icp_hit_stale is used by the sibling. Excessive use
3861 of this option may result in forwarding loops. One way
3862 to prevent peering loops when using this option, is to
3863 deny cache peer usage on requests from a peer:
3865 cache_peer_access peerName deny fromPeer
3867 max-conn=N Limit the number of concurrent connections the Squid
3868 may open to this peer, including already opened idle
3869 and standby connections. There is no peer-specific
3870 connection limit by default.
3872 A peer exceeding the limit is not used for new
3873 requests unless a standby connection is available.
3875 max-conn currently works poorly with idle persistent
3876 connections: When a peer reaches its max-conn limit,
3877 and there are idle persistent connections to the peer,
3878 the peer may not be selected because the limiting code
3879 does not know whether Squid can reuse those idle
3882 standby=N Maintain a pool of N "hot standby" connections to an
3883 UP peer, available for requests when no idle
3884 persistent connection is available (or safe) to use.
3885 By default and with zero N, no such pool is maintained.
3886 N must not exceed the max-conn limit (if any).
3888 At start or after reconfiguration, Squid opens new TCP
3889 standby connections until there are N connections
3890 available and then replenishes the standby pool as
3891 opened connections are used up for requests. A used
3892 connection never goes back to the standby pool, but
3893 may go to the regular idle persistent connection pool
3894 shared by all peers and origin servers.
3896 Squid never opens multiple new standby connections
3897 concurrently. This one-at-a-time approach minimizes
3898 flooding-like effect on peers. Furthermore, just a few
3899 standby connections should be sufficient in most cases
3900 to supply most new requests with a ready-to-use
3903 Standby connections obey server_idle_pconn_timeout.
3904 For the feature to work as intended, the peer must be
3905 configured to accept and keep them open longer than
3906 the idle timeout at the connecting Squid, to minimize
3907 race conditions typical to idle used persistent
3908 connections. Default request_timeout and
3909 server_idle_pconn_timeout values ensure such a
3912 name=xxx Unique name for the peer.
3913 Required if you have multiple peers on the same host
3914 but different ports.
3915 This name can be used in cache_peer_access and similar
3916 directives to identify the peer.
3917 Can be used by outgoing access controls through the
3920 no-tproxy Do not use the client-spoof TPROXY support when forwarding
3921 requests to this peer. Use normal address selection instead.
3922 This overrides the spoof_client_ip ACL.
3924 proxy-only objects fetched from the peer will not be stored locally.
3928 NAME: cache_peer_access
3931 DEFAULT_DOC: No peer usage restrictions.
3934 Restricts usage of cache_peer proxies.
3937 cache_peer_access peer-name allow|deny [!]aclname ...
3939 For the required peer-name parameter, use either the value of the
3940 cache_peer name=value parameter or, if name=value is missing, the
3941 cache_peer hostname parameter.
3943 This directive narrows down the selection of peering candidates, but
3944 does not determine the order in which the selected candidates are
3945 contacted. That order is determined by the peer selection algorithms
3946 (see PEER SELECTION sections in the cache_peer documentation).
3948 If a deny rule matches, the corresponding peer will not be contacted
3949 for the current transaction -- Squid will not send ICP queries and
3950 will not forward HTTP requests to that peer. An allow match leaves
3951 the corresponding peer in the selection. The first match for a given
3952 peer wins for that peer.
3954 The relative order of cache_peer_access directives for the same peer
3955 matters. The relative order of any two cache_peer_access directives
3956 for different peers does not matter. To ease interpretation, it is a
3957 good idea to group cache_peer_access directives for the same peer
3960 A single cache_peer_access directive may be evaluated multiple times
3961 for a given transaction because individual peer selection algorithms
3962 may check it independently from each other. These redundant checks
3963 may be optimized away in future Squid versions.
3965 This clause only supports fast acl types.
3966 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
3970 NAME: neighbor_type_domain
3971 TYPE: hostdomaintype
3973 DEFAULT_DOC: The peer type from cache_peer directive is used for all requests to that peer.
3976 Modify the cache_peer neighbor type when passing requests
3977 about specific domains to the peer.
3980 neighbor_type_domain neighbor parent|sibling domain domain ...
3983 cache_peer foo.example.com parent 3128 3130
3984 neighbor_type_domain foo.example.com sibling .au .de
3986 The above configuration treats all requests to foo.example.com as a
3987 parent proxy unless the request is for a .au or .de ccTLD domain name.
3990 NAME: dead_peer_timeout
3994 LOC: Config.Timeout.deadPeer
3996 This controls how long Squid waits to declare a peer cache
3997 as "dead." If there are no ICP replies received in this
3998 amount of time, Squid will declare the peer dead and not
3999 expect to receive any further ICP replies. However, it
4000 continues to send ICP queries, and will mark the peer as
4001 alive upon receipt of the first subsequent ICP reply.
4003 This timeout also affects when Squid expects to receive ICP
4004 replies from peers. If more than 'dead_peer' seconds have
4005 passed since the last ICP reply was received, Squid will not
4006 expect to receive an ICP reply on the next query. Thus, if
4007 your time between requests is greater than this timeout, you
4008 will see a lot of requests sent DIRECT to origin servers
4009 instead of to your parents.
4012 NAME: forward_max_tries
4015 LOC: Config.forward_max_tries
4017 Limits the number of attempts to forward the request.
4019 For the purpose of this limit, Squid counts all high-level request
4020 forwarding attempts, including any same-destination retries after
4021 certain persistent connection failures and any attempts to use a
4022 different peer. However, low-level connection reopening attempts
4023 (enabled using connect_retries) are not counted.
4025 See also: forward_timeout and connect_retries.
4029 MEMORY CACHE OPTIONS
4030 -----------------------------------------------------------------------------
4037 LOC: Config.memMaxSize
4039 NOTE: THIS PARAMETER DOES NOT SPECIFY THE MAXIMUM PROCESS SIZE.
4040 IT ONLY PLACES A LIMIT ON HOW MUCH ADDITIONAL MEMORY SQUID WILL
4041 USE AS A MEMORY CACHE OF OBJECTS. SQUID USES MEMORY FOR OTHER
4042 THINGS AS WELL. SEE THE SQUID FAQ SECTION 8 FOR DETAILS.
4044 'cache_mem' specifies the ideal amount of memory to be used
4046 * In-Transit objects
4048 * Negative-Cached objects
4050 Data for these objects are stored in 4 KB blocks. This
4051 parameter specifies the ideal upper limit on the total size of
4052 4 KB blocks allocated. In-Transit objects take the highest
4055 In-transit objects have priority over the others. When
4056 additional space is needed for incoming data, negative-cached
4057 and hot objects will be released. In other words, the
4058 negative-cached and hot objects will fill up any unused space
4059 not needed for in-transit objects.
4061 If circumstances require, this limit will be exceeded.
4062 Specifically, if your incoming request rate requires more than
4063 'cache_mem' of memory to hold in-transit objects, Squid will
4064 exceed this limit to satisfy the new requests. When the load
4065 decreases, blocks will be freed until the high-water mark is
4066 reached. Thereafter, blocks will be used to store hot
4069 If shared memory caching is enabled, Squid does not use the shared
4070 cache space for in-transit objects, but they still consume as much
4071 local memory as they need. For more details about the shared memory
4072 cache, see memory_cache_shared.
4075 NAME: maximum_object_size_in_memory
4079 LOC: Config.Store.maxInMemObjSize
4081 Objects greater than this size will not be attempted to kept in
4082 the memory cache. This should be set high enough to keep objects
4083 accessed frequently in memory to improve performance whilst low
4084 enough to keep larger objects from hoarding cache_mem.
4087 NAME: memory_cache_shared
4090 LOC: Config.memShared
4092 DEFAULT_DOC: "on" where supported if doing memory caching with multiple SMP workers.
4094 Controls whether the memory cache is shared among SMP workers.
4096 The shared memory cache is meant to occupy cache_mem bytes and replace
4097 the non-shared memory cache, although some entities may still be
4098 cached locally by workers for now (e.g., internal and in-transit
4099 objects may be served from a local memory cache even if shared memory
4100 caching is enabled).
4102 By default, the memory cache is shared if and only if all of the
4103 following conditions are satisfied: Squid runs in SMP mode with
4104 multiple workers, cache_mem is positive, and Squid environment
4105 supports required IPC primitives (e.g., POSIX shared memory segments
4106 and GCC-style atomic operations).
4108 To avoid blocking locks, shared memory uses opportunistic algorithms
4109 that do not guarantee that every cachable entity that could have been
4110 shared among SMP workers will actually be shared.
4113 NAME: memory_cache_mode
4117 DEFAULT_DOC: Keep the most recently fetched objects in memory
4119 Controls which objects to keep in the memory cache (cache_mem)
4121 always Keep most recently fetched objects in memory (default)
4123 disk Only disk cache hits are kept in memory, which means
4124 an object must first be cached on disk and then hit
4125 a second time before cached in memory.
4127 network Only objects fetched from network is kept in memory
4130 NAME: memory_replacement_policy
4132 LOC: Config.memPolicy
4135 The memory replacement policy parameter determines which
4136 objects are purged from memory when memory space is needed.
4138 See cache_replacement_policy for details on algorithms.
4143 -----------------------------------------------------------------------------
4146 NAME: cache_replacement_policy
4148 LOC: Config.replPolicy
4151 The cache replacement policy parameter determines which
4152 objects are evicted (replaced) when disk space is needed.
4154 lru : Squid's original list based LRU policy
4155 heap GDSF : Greedy-Dual Size Frequency
4156 heap LFUDA: Least Frequently Used with Dynamic Aging
4157 heap LRU : LRU policy implemented using a heap
4159 Applies to any cache_dir lines listed below this directive.
4161 The LRU policies keeps recently referenced objects.
4163 The heap GDSF policy optimizes object hit rate by keeping smaller
4164 popular objects in cache so it has a better chance of getting a
4165 hit. It achieves a lower byte hit rate than LFUDA though since
4166 it evicts larger (possibly popular) objects.
4168 The heap LFUDA policy keeps popular objects in cache regardless of
4169 their size and thus optimizes byte hit rate at the expense of
4170 hit rate since one large, popular object will prevent many
4171 smaller, slightly less popular objects from being cached.
4173 Both policies utilize a dynamic aging mechanism that prevents
4174 cache pollution that can otherwise occur with frequency-based
4175 replacement policies.
4177 NOTE: if using the LFUDA replacement policy you should increase
4178 the value of maximum_object_size above its default of 4 MB to
4179 to maximize the potential byte hit rate improvement of LFUDA.
4181 For more information about the GDSF and LFUDA cache replacement
4182 policies see http://www.hpl.hp.com/techreports/1999/HPL-1999-69.html
4183 and http://fog.hpl.external.hp.com/techreports/98/HPL-98-173.html.
4186 NAME: minimum_object_size
4190 DEFAULT_DOC: no limit
4191 LOC: Config.Store.minObjectSize
4193 Objects smaller than this size will NOT be saved on disk. The
4194 value is specified in bytes, and the default is 0 KB, which
4195 means all responses can be stored.
4198 NAME: maximum_object_size
4202 LOC: Config.Store.maxObjectSize
4204 Set the default value for max-size parameter on any cache_dir.
4205 The value is specified in bytes, and the default is 4 MB.
4207 If you wish to get a high BYTES hit ratio, you should probably
4208 increase this (one 32 MB object hit counts for 3200 10KB
4211 If you wish to increase hit ratio more than you want to
4212 save bandwidth you should leave this low.
4214 NOTE: if using the LFUDA replacement policy you should increase
4215 this value to maximize the byte hit rate improvement of LFUDA!
4216 See cache_replacement_policy for a discussion of this policy.
4222 DEFAULT_DOC: No disk cache. Store cache objects only in memory.
4223 LOC: Config.cacheSwap
4226 cache_dir Type Directory-Name Fs-specific-data [options]
4228 You can specify multiple cache_dir lines to spread the
4229 cache among different disk partitions.
4231 Type specifies the kind of storage system to use. Only "ufs"
4232 is built by default. To enable any of the other storage systems
4233 see the --enable-storeio configure option.
4235 'Directory' is a top-level directory where cache swap
4236 files will be stored. If you want to use an entire disk
4237 for caching, this can be the mount-point directory.
4238 The directory must exist and be writable by the Squid
4239 process. Squid will NOT create this directory for you.
4241 In SMP configurations, cache_dir must not precede the workers option
4242 and should use configuration macros or conditionals to give each
4243 worker interested in disk caching a dedicated cache directory.
4246 ==== The ufs store type ====
4248 "ufs" is the old well-known Squid storage format that has always
4252 cache_dir ufs Directory-Name Mbytes L1 L2 [options]
4254 'Mbytes' is the amount of disk space (MB) to use under this
4255 directory. The default is 100 MB. Change this to suit your
4256 configuration. Do NOT put the size of your disk drive here.
4257 Instead, if you want Squid to use the entire disk drive,
4258 subtract 20% and use that value.
4260 'L1' is the number of first-level subdirectories which
4261 will be created under the 'Directory'. The default is 16.
4263 'L2' is the number of second-level subdirectories which
4264 will be created under each first-level directory. The default
4268 ==== The aufs store type ====
4270 "aufs" uses the same storage format as "ufs", utilizing
4271 POSIX-threads to avoid blocking the main Squid process on
4272 disk-I/O. This was formerly known in Squid as async-io.
4275 cache_dir aufs Directory-Name Mbytes L1 L2 [options]
4277 see argument descriptions under ufs above
4280 ==== The diskd store type ====
4282 "diskd" uses the same storage format as "ufs", utilizing a
4283 separate process to avoid blocking the main Squid process on
4287 cache_dir diskd Directory-Name Mbytes L1 L2 [options] [Q1=n] [Q2=n]
4289 see argument descriptions under ufs above
4291 Q1 specifies the number of unacknowledged I/O requests when Squid
4292 stops opening new files. If this many messages are in the queues,
4293 Squid won't open new files. Default is 64
4295 Q2 specifies the number of unacknowledged messages when Squid
4296 starts blocking. If this many messages are in the queues,
4297 Squid blocks until it receives some replies. Default is 72
4299 When Q1 < Q2 (the default), the cache directory is optimized
4300 for lower response time at the expense of a decrease in hit
4301 ratio. If Q1 > Q2, the cache directory is optimized for
4302 higher hit ratio at the expense of an increase in response
4306 ==== The rock store type ====
4309 cache_dir rock Directory-Name Mbytes [options]
4311 The Rock Store type is a database-style storage. All cached
4312 entries are stored in a "database" file, using fixed-size slots.
4313 A single entry occupies one or more slots.
4315 If possible, Squid using Rock Store creates a dedicated kid
4316 process called "disker" to avoid blocking Squid worker(s) on disk
4317 I/O. One disker kid is created for each rock cache_dir. Diskers
4318 are created only when Squid, running in daemon mode, has support
4319 for the IpcIo disk I/O module.
4321 swap-timeout=msec: Squid will not start writing a miss to or
4322 reading a hit from disk if it estimates that the swap operation
4323 will take more than the specified number of milliseconds. By
4324 default and when set to zero, disables the disk I/O time limit
4325 enforcement. Ignored when using blocking I/O module because
4326 blocking synchronous I/O does not allow Squid to estimate the
4327 expected swap wait time.
4329 max-swap-rate=swaps/sec: Artificially limits disk access using
4330 the specified I/O rate limit. Swap out requests that
4331 would cause the average I/O rate to exceed the limit are
4332 delayed. Individual swap in requests (i.e., hits or reads) are
4333 not delayed, but they do contribute to measured swap rate and
4334 since they are placed in the same FIFO queue as swap out
4335 requests, they may wait longer if max-swap-rate is smaller.
4336 This is necessary on file systems that buffer "too
4337 many" writes and then start blocking Squid and other processes
4338 while committing those writes to disk. Usually used together
4339 with swap-timeout to avoid excessive delays and queue overflows
4340 when disk demand exceeds available disk "bandwidth". By default
4341 and when set to zero, disables the disk I/O rate limit
4342 enforcement. Currently supported by IpcIo module only.
4344 slot-size=bytes: The size of a database "record" used for
4345 storing cached responses. A cached response occupies at least
4346 one slot and all database I/O is done using individual slots so
4347 increasing this parameter leads to more disk space waste while
4348 decreasing it leads to more disk I/O overheads. Should be a
4349 multiple of your operating system I/O page size. Defaults to
4350 16KBytes. A housekeeping header is stored with each slot and
4351 smaller slot-sizes will be rejected. The header is smaller than
4355 ==== COMMON OPTIONS ====
4357 no-store no new objects should be stored to this cache_dir.
4359 min-size=n the minimum object size in bytes this cache_dir
4360 will accept. It's used to restrict a cache_dir
4361 to only store large objects (e.g. AUFS) while
4362 other stores are optimized for smaller objects
4366 max-size=n the maximum object size in bytes this cache_dir
4368 The value in maximum_object_size directive sets
4369 the default unless more specific details are
4370 available (ie a small store capacity).
4372 Note: To make optimal use of the max-size limits you should order
4373 the cache_dir lines with the smallest max-size value first.
4377 # Uncomment and adjust the following to add a disk cache directory.
4378 #cache_dir ufs @DEFAULT_SWAP_DIR@ 100 16 256
4382 NAME: store_dir_select_algorithm
4384 LOC: Config.store_dir_select_algorithm
4387 How Squid selects which cache_dir to use when the response
4388 object will fit into more than one.
4390 Regardless of which algorithm is used the cache_dir min-size
4391 and max-size parameters are obeyed. As such they can affect
4392 the selection algorithm by limiting the set of considered
4399 This algorithm is suited to caches with similar cache_dir
4400 sizes and disk speeds.
4402 The disk with the least I/O pending is selected.
4403 When there are multiple disks with the same I/O load ranking
4404 the cache_dir with most available capacity is selected.
4406 When a mix of cache_dir sizes are configured the faster disks
4407 have a naturally lower I/O loading and larger disks have more
4408 capacity. So space used to store objects and data throughput
4409 may be very unbalanced towards larger disks.
4414 This algorithm is suited to caches with unequal cache_dir
4417 Each cache_dir is selected in a rotation. The next suitable
4420 Available cache_dir capacity is only considered in relation
4421 to whether the object will fit and meets the min-size and
4422 max-size parameters.
4424 Disk I/O loading is only considered to prevent overload on slow
4425 disks. This algorithm does not spread objects by size, so any
4426 I/O loading per-disk may appear very unbalanced and volatile.
4428 If several cache_dirs use similar min-size, max-size, or other
4429 limits to to reject certain responses, then do not group such
4430 cache_dir lines together, to avoid round-robin selection bias
4431 towards the first cache_dir after the group. Instead, interleave
4432 cache_dir lines from different groups. For example:
4434 store_dir_select_algorithm round-robin
4435 cache_dir rock /hdd1 ... min-size=100000
4436 cache_dir rock /ssd1 ... max-size=99999
4437 cache_dir rock /hdd2 ... min-size=100000
4438 cache_dir rock /ssd2 ... max-size=99999
4439 cache_dir rock /hdd3 ... min-size=100000
4440 cache_dir rock /ssd3 ... max-size=99999
4443 NAME: paranoid_hit_validation
4444 COMMENT: time-units-small
4445 TYPE: time_nanoseconds
4447 DEFAULT_DOC: validation disabled
4448 LOC: Config.paranoid_hit_validation
4450 Controls whether Squid should perform paranoid validation of cache entry
4451 metadata integrity every time a cache entry is hit. This low-level
4452 validation should always succeed. Each failed validation results in a
4453 cache miss, a BUG line reported to cache.log, and the invalid entry
4454 marked as unusable (and eventually purged from the cache).
4456 Squid can only validate shared cache memory and rock cache_dir entries.
4458 * Zero (default) value means that the validation is disabled.
4460 * Positive values enable validation:
4461 - values less than 1 day approximate the maximum time that Squid is allowed
4462 to spend validating a single cache hit.
4463 - values greater or equal to 1 day are considered as no limitation:
4464 in this case all checks will be performed, regardless of how much time
4467 Hits are usually stored using 16KB slots (for rock, the size is
4468 configurable via cache_dir slot-size). Larger hits require scanning more
4469 slots and, hence, take more time. When validation is enabled, at least one
4470 slot is always validated, regardless of the configured time limit.
4472 A worker process validating an entry cannot do anything else (i.e. the
4473 validation is blocking). The validation overhead is environment dependent,
4474 but developers have observed Squid spending 3-10 microseconds to check each
4475 slot of a Rock or shared memory hit entry. If Squid cuts validation short
4476 because it runs out of configured time, it treats the entry as valid.
4478 When hit validation is enabled, its statistics is included in Cache
4479 Manager mgr:counters, mgr:5min, and mgr:60min reports.
4482 NAME: max_open_disk_fds
4484 LOC: Config.max_open_disk_fds
4486 DEFAULT_DOC: no limit
4488 To avoid having disk as the I/O bottleneck Squid can optionally
4489 bypass the on-disk cache if more than this amount of disk file
4490 descriptors are open.
4492 A value of 0 indicates no limit.
4495 NAME: cache_swap_low
4496 COMMENT: (percent, 0-100)
4499 LOC: Config.Swap.lowWaterMark
4501 The low-water mark for AUFS/UFS/diskd cache object eviction by
4502 the cache_replacement_policy algorithm.
4504 Removal begins when the swap (disk) usage of a cache_dir is
4505 above this low-water mark and attempts to maintain utilization
4506 near the low-water mark.
4508 As swap utilization increases towards the high-water mark set
4509 by cache_swap_high object eviction becomes more aggressive.
4511 The value difference in percentages between low- and high-water
4512 marks represent an eviction rate of 300 objects per second and
4513 the rate continues to scale in aggressiveness by multiples of
4514 this above the high-water mark.
4516 Defaults are 90% and 95%. If you have a large cache, 5% could be
4517 hundreds of MB. If this is the case you may wish to set these
4518 numbers closer together.
4520 See also cache_swap_high and cache_replacement_policy
4523 NAME: cache_swap_high
4524 COMMENT: (percent, 0-100)
4527 LOC: Config.Swap.highWaterMark
4529 The high-water mark for AUFS/UFS/diskd cache object eviction by
4530 the cache_replacement_policy algorithm.
4532 Removal begins when the swap (disk) usage of a cache_dir is
4533 above the low-water mark set by cache_swap_low and attempts to
4534 maintain utilization near the low-water mark.
4536 As swap utilization increases towards this high-water mark object
4537 eviction becomes more aggressive.
4539 The value difference in percentages between low- and high-water
4540 marks represent an eviction rate of 300 objects per second and
4541 the rate continues to scale in aggressiveness by multiples of
4542 this above the high-water mark.
4544 Defaults are 90% and 95%. If you have a large cache, 5% could be
4545 hundreds of MB. If this is the case you may wish to set these
4546 numbers closer together.
4548 See also cache_swap_low and cache_replacement_policy
4553 -----------------------------------------------------------------------------
4560 DEFAULT_DOC: The format definitions squid, common, combined, referrer, useragent are built in.
4564 logformat <name> <format specification>
4566 Defines an access log format.
4568 The <format specification> is a string with embedded % format codes
4570 % format codes all follow the same basic structure where all
4571 components but the formatcode are optional and usually unnecessary,
4572 especially when dealing with common codes.
4574 % [encoding] [-] [[0]width] [{arg}] formatcode [{arg}]
4576 encoding escapes or otherwise protects "special" characters:
4578 " Quoted string encoding where quote(") and
4579 backslash(\) characters are \-escaped while
4580 CR, LF, and TAB characters are encoded as \r,
4581 \n, and \t two-character sequences.
4583 [ Custom Squid encoding where percent(%), square
4584 brackets([]), backslash(\) and characters with
4585 codes outside of [32,126] range are %-encoded.
4586 SP is not encoded. Used by log_mime_hdrs.
4588 # URL encoding (a.k.a. percent-encoding) where
4589 all URL unsafe and control characters (per RFC
4590 1738) are %-encoded.
4592 / Shell-like encoding where quote(") and
4593 backslash(\) characters are \-escaped while CR
4594 and LF characters are encoded as \r and \n
4595 two-character sequences. Values containing SP
4596 character(s) are surrounded by quotes(").
4598 ' Raw/as-is encoding with no escaping/quoting.
4600 Default encoding: When no explicit encoding is
4601 specified, each %code determines its own encoding.
4602 Most %codes use raw/as-is encoding, but some codes use
4603 a so called "pass-through URL encoding" where all URL
4604 unsafe and control characters (per RFC 1738) are
4605 %-encoded, but the percent character(%) is left as is.
4609 width minimum and/or maximum field width:
4610 [width_min][.width_max]
4611 When minimum starts with 0, the field is zero-padded.
4612 String values exceeding maximum width are truncated.
4614 {arg} argument such as header name etc. This field may be
4615 placed before or after the token, but not both at once.
4619 % a literal % character
4620 sn Unique sequence number per log line entry
4621 err_code The ID of an error response served by Squid or
4622 a similar internal error identifier.
4623 err_detail Additional err_code-dependent error information.
4624 note The annotation specified by the argument. Also
4625 logs the adaptation meta headers set by the
4626 adaptation_meta configuration parameter.
4627 If no argument given all annotations logged.
4628 The argument may include a separator to use with
4631 By default, multiple note values are separated with ","
4632 and multiple notes are separated with "\r\n".
4633 When logging named notes with %{name}note, the
4634 explicitly configured separator is used between note
4635 values. When logging all notes with %note, the
4636 explicitly configured separator is used between
4637 individual notes. There is currently no way to
4638 specify both value and notes separators when logging
4639 all notes with %note.
4640 master_xaction The master transaction identifier is an unsigned
4641 integer. These IDs are guaranteed to monotonically
4642 increase within a single worker process lifetime, with
4643 higher values corresponding to transactions that were
4644 accepted or initiated later. Due to current implementation
4645 deficiencies, some IDs are skipped (i.e. never logged).
4646 Concurrent workers and restarted workers use similar,
4647 overlapping sequences of master transaction IDs.
4649 Connection related format codes:
4651 >a Client source IP address
4653 >p Client source port
4654 >eui Client source EUI (MAC address, EUI-48 or EUI-64 identifier)
4655 >la Local IP address the client connected to
4656 >lp Local port number the client connected to
4657 >qos Client connection TOS/DSCP value set by Squid
4658 >nfmark Client connection netfilter packet MARK set by Squid
4660 transport::>connection_id Identifies a transport connection
4661 accepted by Squid (e.g., a connection carrying the
4662 logged HTTP request). Currently, Squid only supports
4663 TCP transport connections.
4665 The logged identifier is an unsigned integer. These
4666 IDs are guaranteed to monotonically increase within a
4667 single worker process lifetime, with higher values
4668 corresponding to connections that were accepted later.
4669 Many IDs are skipped (i.e. never logged). Concurrent
4670 workers and restarted workers use similar, partially
4671 overlapping sequences of IDs.
4673 la Local listening IP address the client connection was connected to.
4674 lp Local listening port number the client connection was connected to.
4676 <a Server IP address of the last server or peer connection
4677 <A Server FQDN or peer name
4678 <p Server port number of the last server or peer connection
4679 <la Local IP address of the last server or peer connection
4680 <lp Local port number of the last server or peer connection
4681 <qos Server connection TOS/DSCP value set by Squid
4682 <nfmark Server connection netfilter packet MARK set by Squid
4684 >handshake Raw client handshake
4685 Initial client bytes received by Squid on a newly
4686 accepted TCP connection or inside a just established
4687 CONNECT tunnel. Squid stops accumulating handshake
4688 bytes as soon as the handshake parser succeeds or
4689 fails (determining whether the client is using the
4692 For HTTP clients, the handshake is the request line.
4693 For TLS clients, the handshake consists of all TLS
4694 records up to and including the TLS record that
4695 contains the last byte of the first ClientHello
4696 message. For clients using an unsupported protocol,
4697 this field contains the bytes received by Squid at the
4698 time of the handshake parsing failure.
4700 See the on_unsupported_protocol directive for more
4701 information on Squid handshake traffic expectations.
4703 Current support is limited to these contexts:
4704 - http_port connections, but only when the
4705 on_unsupported_protocol directive is in use.
4706 - https_port connections (and CONNECT tunnels) that
4707 are subject to the ssl_bump peek or stare action.
4709 To protect binary handshake data, this field is always
4710 base64-encoded (RFC 4648 Section 4). If logformat
4711 field encoding is configured, that encoding is applied
4712 on top of base64. Otherwise, the computed base64 value
4715 Time related format codes:
4717 ts Seconds since epoch
4718 tu subsecond time (milliseconds)
4719 tl Local time. Optional strftime format argument
4720 default %d/%b/%Y:%H:%M:%S %z
4721 tg GMT time. Optional strftime format argument
4722 default %d/%b/%Y:%H:%M:%S %z
4723 tr Response time (milliseconds)
4724 dt Total time spent making DNS lookups (milliseconds)
4725 tS Approximate master transaction start time in
4726 <full seconds since epoch>.<fractional seconds> format.
4727 Currently, Squid considers the master transaction
4728 started when a complete HTTP request header initiating
4729 the transaction is received from the client. This is
4730 the same value that Squid uses to calculate transaction
4731 response time when logging %tr to access.log. Currently,
4732 Squid uses millisecond resolution for %tS values,
4733 similar to the default access.log "current time" field
4736 Access Control related format codes:
4738 et Tag returned by external acl
4739 ea Log string returned by external acl
4740 un User name (any available)
4741 ul User name from authentication
4742 ue User name from external acl helper
4743 ui User name from ident
4744 un A user name. Expands to the first available name
4745 from the following list of information sources:
4746 - authenticated user name, like %ul
4747 - user name supplied by an external ACL, like %ue
4748 - SSL client name, like %us
4749 - ident user name, like %ui
4750 credentials Client credentials. The exact meaning depends on
4751 the authentication scheme: For Basic authentication,
4752 it is the password; for Digest, the realm sent by the
4753 client; for NTLM and Negotiate, the client challenge
4754 or client credentials prefixed with "YR " or "KK ".
4756 HTTP related format codes:
4760 [http::]rm Request method (GET/POST etc)
4761 [http::]>rm Request method from client
4762 [http::]<rm Request method sent to server or peer
4764 [http::]ru Request URL received (or computed) and sanitized
4766 Logs request URI received from the client, a
4767 request adaptation service, or a request
4768 redirector (whichever was applied last).
4770 Computed URLs are URIs of internally generated
4771 requests and various "error:..." URIs.
4773 Honors strip_query_terms and uri_whitespace.
4775 This field is not encoded by default. Encoding
4776 this field using variants of %-encoding will
4777 clash with uri_whitespace modifications that
4778 also use %-encoding.
4780 [http::]>ru Request URL received from the client (or computed)
4782 Computed URLs are URIs of internally generated
4783 requests and various "error:..." URIs.
4785 Unlike %ru, this request URI is not affected
4786 by request adaptation, URL rewriting services,
4787 and strip_query_terms.
4789 Honors uri_whitespace.
4791 This field is using pass-through URL encoding
4792 by default. Encoding this field using other
4793 variants of %-encoding will clash with
4794 uri_whitespace modifications that also use
4797 [http::]<ru Request URL sent to server or peer
4798 [http::]>rs Request URL scheme from client
4799 [http::]<rs Request URL scheme sent to server or peer
4800 [http::]>rd Request URL domain from client
4801 [http::]<rd Request URL domain sent to server or peer
4802 [http::]>rP Request URL port from client
4803 [http::]<rP Request URL port sent to server or peer
4804 [http::]rp Request URL path excluding hostname
4805 [http::]>rp Request URL path excluding hostname from client
4806 [http::]<rp Request URL path excluding hostname sent to server or peer
4807 [http::]rv Request protocol version
4808 [http::]>rv Request protocol version from client
4809 [http::]<rv Request protocol version sent to server or peer
4811 [http::]>h Original received request header.
4812 Usually differs from the request header sent by
4813 Squid, although most fields are often preserved.
4814 Accepts optional header field name/value filter
4815 argument using name[:[separator]element] format.
4816 [http::]>ha Received request header after adaptation and
4817 redirection (pre-cache REQMOD vectoring point).
4818 Usually differs from the request header sent by
4819 Squid, although most fields are often preserved.
4820 Optional header name argument as for >h
4824 [http::]<Hs HTTP status code received from the next hop
4825 [http::]>Hs HTTP status code sent to the client
4827 [http::]<h Reply header. Optional header name argument
4830 [http::]mt MIME content type
4835 [http::]st Total size of request + reply traffic with client
4836 [http::]>st Total size of request received from client.
4837 Excluding chunked encoding bytes.
4838 [http::]<st Total size of reply sent to client (after adaptation)
4840 [http::]>sh Size of request headers received from client
4841 [http::]<sh Size of reply headers sent to client (after adaptation)
4843 [http::]<sH Reply high offset sent
4844 [http::]<sS Upstream object size
4846 [http::]<bs Number of HTTP-equivalent message body bytes
4847 received from the next hop, excluding chunked
4848 transfer encoding and control messages.
4849 Generated FTP/Gopher listings are treated as
4854 [http::]<pt Peer response time in milliseconds. The timer starts
4855 when the last request byte is sent to the next hop
4856 and stops when the last response byte is received.
4857 [http::]<tt Total time in milliseconds. The timer
4858 starts with the first connect request (or write I/O)
4859 sent to the first selected peer. The timer stops
4860 with the last I/O with the last peer.
4862 Squid handling related format codes:
4864 Ss Squid request status (TCP_MISS etc)
4865 Sh Squid hierarchy status (DEFAULT_PARENT etc)
4867 SSL-related format codes:
4869 ssl::bump_mode SslBump decision for the transaction:
4871 For CONNECT requests that initiated bumping of
4872 a connection and for any request received on
4873 an already bumped connection, Squid logs the
4874 corresponding SslBump mode ("splice", "bump",
4875 "peek", "stare", "terminate", "server-first"
4876 or "client-first"). See the ssl_bump option
4877 for more information about these modes.
4879 A "none" token is logged for requests that
4880 triggered "ssl_bump" ACL evaluation matching
4883 In all other cases, a single dash ("-") is
4886 ssl::>sni SSL client SNI sent to Squid.
4889 The Subject field of the received client
4890 SSL certificate or a dash ('-') if Squid has
4891 received an invalid/malformed certificate or
4892 no certificate at all. Consider encoding the
4893 logged value because Subject often has spaces.
4896 The Issuer field of the received client
4897 SSL certificate or a dash ('-') if Squid has
4898 received an invalid/malformed certificate or
4899 no certificate at all. Consider encoding the
4900 logged value because Issuer often has spaces.
4903 The Subject field of the received server
4904 TLS certificate or a dash ('-') if this is
4905 not available. Consider encoding the logged
4906 value because Subject often has spaces.
4909 The Issuer field of the received server
4910 TLS certificate or a dash ('-') if this is
4911 not available. Consider encoding the logged
4912 value because Issuer often has spaces.
4915 The received server x509 certificate in PEM
4916 format, including BEGIN and END lines (or a
4917 dash ('-') if the certificate is unavailable).
4919 WARNING: Large certificates will exceed the
4920 current 8KB access.log record limit, resulting
4921 in truncated records. Such truncation usually
4922 happens in the middle of a record field. The
4923 limit applies to all access logging modules.
4925 The logged certificate may have failed
4926 validation and may not be trusted by Squid.
4927 This field does not include any intermediate
4928 certificates that may have been received from
4929 the server or fetched during certificate
4932 Currently, Squid only collects server
4933 certificates during step3 of SslBump
4934 processing; connections that were not subject
4935 to ssl_bump rules or that did not match a peek
4936 or stare rule at step2 will not have the
4937 server certificate information.
4939 This field is using pass-through URL encoding
4943 The list of certificate validation errors
4944 detected by Squid (including OpenSSL and
4945 certificate validation helper components). The
4946 errors are listed in the discovery order. By
4947 default, the error codes are separated by ':'.
4948 Accepts an optional separator argument.
4950 %ssl::>negotiated_version The negotiated TLS version of the
4953 %ssl::<negotiated_version The negotiated TLS version of the
4954 last server or peer connection.
4956 %ssl::>received_hello_version The TLS version of the Hello
4957 message received from TLS client.
4959 %ssl::<received_hello_version The TLS version of the Hello
4960 message received from TLS server.
4962 %ssl::>received_supported_version The maximum TLS version
4963 supported by the TLS client.
4965 %ssl::<received_supported_version The maximum TLS version
4966 supported by the TLS server.
4968 %ssl::>negotiated_cipher The negotiated cipher of the
4971 %ssl::<negotiated_cipher The negotiated cipher of the
4972 last server or peer connection.
4974 If ICAP is enabled, the following code becomes available (as
4975 well as ICAP log codes documented with the icap_log option):
4977 icap::tt Total ICAP processing time for the HTTP
4978 transaction. The timer ticks when ICAP
4979 ACLs are checked and when ICAP
4980 transaction is in progress.
4982 If adaptation is enabled the following codes become available:
4984 adapt::<last_h The header of the last ICAP response or
4985 meta-information from the last eCAP
4986 transaction related to the HTTP transaction.
4987 Like <h, accepts an optional header name
4990 adapt::sum_trs Summed adaptation transaction response
4991 times recorded as a comma-separated list in
4992 the order of transaction start time. Each time
4993 value is recorded as an integer number,
4994 representing response time of one or more
4995 adaptation (ICAP or eCAP) transaction in
4996 milliseconds. When a failed transaction is
4997 being retried or repeated, its time is not
4998 logged individually but added to the
4999 replacement (next) transaction. See also:
5002 adapt::all_trs All adaptation transaction response times.
5003 Same as adaptation_strs but response times of
5004 individual transactions are never added
5005 together. Instead, all transaction response
5006 times are recorded individually.
5008 You can prefix adapt::*_trs format codes with adaptation
5009 service name in curly braces to record response time(s) specific
5010 to that service. For example: %{my_service}adapt::sum_trs
5012 Format codes related to the PROXY protocol:
5014 proxy_protocol::>h PROXY protocol header, including optional TLVs.
5016 Supports the same field and element reporting/extraction logic
5017 as %http::>h. For configuration and reporting purposes, Squid
5018 maps each PROXY TLV to an HTTP header field: the TLV type
5019 (configured as a decimal integer) is the field name, and the
5020 TLV value is the field value. All TLVs of "LOCAL" connections
5021 (in PROXY protocol terminology) are currently skipped/ignored.
5023 Squid also maps the following standard PROXY protocol header
5024 blocks to pseudo HTTP headers (their names use PROXY
5025 terminology and start with a colon, following HTTP tradition
5026 for pseudo headers): :command, :version, :src_addr, :dst_addr,
5027 :src_port, and :dst_port.
5029 Without optional parameters, this logformat code logs
5030 pseudo headers and TLVs.
5032 This format code uses pass-through URL encoding by default.
5035 # relay custom PROXY TLV #224 to adaptation services
5036 adaptation_meta Client-Foo "%proxy_protocol::>h{224}"
5040 The default formats available (which do not need re-defining) are:
5042 logformat squid %ts.%03tu %6tr %>a %Ss/%03>Hs %<st %rm %ru %[un %Sh/%<a %mt
5043 logformat common %>a %[ui %[un [%tl] "%rm %ru HTTP/%rv" %>Hs %<st %Ss:%Sh
5044 logformat combined %>a %[ui %[un [%tl] "%rm %ru HTTP/%rv" %>Hs %<st "%{Referer}>h" "%{User-Agent}>h" %Ss:%Sh
5045 logformat referrer %ts.%03tu %>a %{Referer}>h %ru
5046 logformat useragent %>a [%tl] "%{User-Agent}>h"
5048 NOTE: When the log_mime_hdrs directive is set to ON.
5049 The squid, common and combined formats have a safely encoded copy
5050 of the mime headers appended to each line within a pair of brackets.
5052 NOTE: The common and combined formats are not quite true to the Apache definition.
5053 The logs from Squid contain an extra status and hierarchy code appended.
5057 NAME: access_log cache_access_log
5059 LOC: Config.Log.accesslogs
5060 DEFAULT_IF_NONE: daemon:@DEFAULT_ACCESS_LOG@ squid
5062 Configures whether and how Squid logs HTTP and ICP transactions.
5063 If access logging is enabled, a single line is logged for every
5064 matching HTTP or ICP request. The recommended directive formats are:
5066 access_log <module>:<place> [option ...] [acl acl ...]
5067 access_log none [acl acl ...]
5069 The following directive format is accepted but may be deprecated:
5070 access_log <module>:<place> [<logformat name> [acl acl ...]]
5072 In most cases, the first ACL name must not contain the '=' character
5073 and should not be equal to an existing logformat name. You can always
5074 start with an 'all' ACL to work around those restrictions.
5076 Will log to the specified module:place using the specified format (which
5077 must be defined in a logformat directive) those entries which match
5078 ALL the acl's specified (which must be defined in acl clauses).
5079 If no acl is specified, all requests will be logged to this destination.
5081 ===== Available options for the recommended directive format =====
5083 logformat=name Names log line format (either built-in or
5084 defined by a logformat directive). Defaults
5087 buffer-size=64KB Defines approximate buffering limit for log
5088 records (see buffered_logs). Squid should not
5089 keep more than the specified size and, hence,
5090 should flush records before the buffer becomes
5091 full to avoid overflows under normal
5092 conditions (the exact flushing algorithm is
5093 module-dependent though). The on-error option
5094 controls overflow handling.
5096 on-error=die|drop Defines action on unrecoverable errors. The
5097 'drop' action ignores (i.e., does not log)
5098 affected log records. The default 'die' action
5099 kills the affected worker. The drop action
5100 support has not been tested for modules other
5103 rotate=N Specifies the number of log file rotations to
5104 make when you run 'squid -k rotate'. The default
5105 is to obey the logfile_rotate directive. Setting
5106 rotate=0 will disable the file name rotation,
5107 but the log files are still closed and re-opened.
5108 This will enable you to rename the logfiles
5109 yourself just before sending the rotate signal.
5110 Only supported by the stdio module.
5112 ===== Modules Currently available =====
5114 none Do not log any requests matching these ACL.
5115 Do not specify Place or logformat name.
5117 stdio Write each log line to disk immediately at the completion of
5119 Place: the filename and path to be written.
5121 daemon Very similar to stdio. But instead of writing to disk the log
5122 line is passed to a daemon helper for asychronous handling instead.
5123 Place: varies depending on the daemon.
5125 log_file_daemon Place: the file name and path to be written.
5127 syslog To log each request via syslog facility.
5128 Place: The syslog facility and priority level for these entries.
5129 Place Format: facility.priority
5131 where facility could be any of:
5132 authpriv, daemon, local0 ... local7 or user.
5134 And priority could be any of:
5135 err, warning, notice, info, debug.
5137 udp To send each log line as text data to a UDP receiver.
5138 Place: The destination host name or IP and port.
5139 Place Format: //host:port
5141 tcp To send each log line as text data to a TCP receiver.
5142 Lines may be accumulated before sending (see buffered_logs).
5143 Place: The destination host name or IP and port.
5144 Place Format: //host:port
5147 access_log daemon:@DEFAULT_ACCESS_LOG@ squid
5153 LOC: Config.Log.icaplogs
5156 ICAP log files record ICAP transaction summaries, one line per
5159 The icap_log option format is:
5160 icap_log <filepath> [<logformat name> [acl acl ...]]
5161 icap_log none [acl acl ...]]
5163 Please see access_log option documentation for details. The two
5164 kinds of logs share the overall configuration approach and many
5167 ICAP processing of a single HTTP message or transaction may
5168 require multiple ICAP transactions. In such cases, multiple
5169 ICAP transaction log lines will correspond to a single access
5172 ICAP log supports many access.log logformat %codes. In ICAP context,
5173 HTTP message-related %codes are applied to the HTTP message embedded
5174 in an ICAP message. Logformat "%http::>..." codes are used for HTTP
5175 messages embedded in ICAP requests while "%http::<..." codes are used
5176 for HTTP messages embedded in ICAP responses. For example:
5178 http::>h To-be-adapted HTTP message headers sent by Squid to
5179 the ICAP service. For REQMOD transactions, these are
5180 HTTP request headers. For RESPMOD, these are HTTP
5181 response headers, but Squid currently cannot log them
5182 (i.e., %http::>h will expand to "-" for RESPMOD).
5184 http::<h Adapted HTTP message headers sent by the ICAP
5185 service to Squid (i.e., HTTP request headers in regular
5186 REQMOD; HTTP response headers in RESPMOD and during
5187 request satisfaction in REQMOD).
5189 ICAP OPTIONS transactions do not embed HTTP messages.
5191 Several logformat codes below deal with ICAP message bodies. An ICAP
5192 message body, if any, typically includes a complete HTTP message
5193 (required HTTP headers plus optional HTTP message body). When
5194 computing HTTP message body size for these logformat codes, Squid
5195 either includes or excludes chunked encoding overheads; see
5196 code-specific documentation for details.
5198 For Secure ICAP services, all size-related information is currently
5199 computed before/after TLS encryption/decryption, as if TLS was not
5202 The following format codes are also available for ICAP logs:
5204 icap::<A ICAP server IP address. Similar to <A.
5206 icap::<service_name ICAP service name from the icap_service
5207 option in Squid configuration file.
5209 icap::ru ICAP Request-URI. Similar to ru.
5211 icap::rm ICAP request method (REQMOD, RESPMOD, or
5212 OPTIONS). Similar to existing rm.
5214 icap::>st The total size of the ICAP request sent to the ICAP
5215 server (ICAP headers + ICAP body), including chunking
5218 icap::<st The total size of the ICAP response received from the
5219 ICAP server (ICAP headers + ICAP body), including
5220 chunking metadata (if any).
5222 icap::<bs The size of the ICAP response body received from the
5223 ICAP server, excluding chunking metadata (if any).
5225 icap::tr Transaction response time (in
5226 milliseconds). The timer starts when
5227 the ICAP transaction is created and
5228 stops when the transaction is completed.
5231 icap::tio Transaction I/O time (in milliseconds). The
5232 timer starts when the first ICAP request
5233 byte is scheduled for sending. The timers
5234 stops when the last byte of the ICAP response
5237 icap::to Transaction outcome: ICAP_ERR* for all
5238 transaction errors, ICAP_OPT for OPTION
5239 transactions, ICAP_ECHO for 204
5240 responses, ICAP_MOD for message
5241 modification, and ICAP_SAT for request
5242 satisfaction. Similar to Ss.
5244 icap::Hs ICAP response status code. Similar to Hs.
5246 icap::>h ICAP request header(s). Similar to >h.
5248 icap::<h ICAP response header(s). Similar to <h.
5250 The default ICAP log format, which can be used without an explicit
5251 definition, is called icap_squid:
5253 logformat icap_squid %ts.%03tu %6icap::tr %>A %icap::to/%03icap::Hs %icap::<st %icap::rm %icap::ru %un -/%icap::<A -
5255 See also: logformat and %adapt::<last_h
5258 NAME: logfile_daemon
5260 DEFAULT: @DEFAULT_LOGFILED@
5261 LOC: Log::TheConfig.logfile_daemon
5263 Specify the path to the logfile-writing daemon. This daemon is
5264 used to write the access and store logs, if configured.
5266 Squid sends a number of commands to the log daemon:
5267 L<data>\n - logfile data
5272 r<n>\n - set rotate count to <n>
5273 b<n>\n - 1 = buffer output, 0 = don't buffer output
5275 No responses is expected.
5278 NAME: stats_collection
5280 LOC: Config.accessList.stats_collection
5282 DEFAULT_DOC: Allow logging for all transactions.
5283 COMMENT: allow|deny acl acl...
5285 This options allows you to control which requests gets accounted
5286 in performance counters.
5288 This clause only supports fast acl types.
5289 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
5292 NAME: cache_store_log
5295 LOC: Config.Log.store
5297 Logs the activities of the storage manager. Shows which
5298 objects are ejected from the cache, and which objects are
5299 saved and for how long.
5300 There are not really utilities to analyze this data, so you can safely
5301 disable it (the default).
5303 Store log uses modular logging outputs. See access_log for the list
5304 of modules supported.
5307 cache_store_log stdio:@DEFAULT_STORE_LOG@
5308 cache_store_log daemon:@DEFAULT_STORE_LOG@
5311 NAME: cache_swap_state cache_swap_log
5313 LOC: Config.Log.swap
5315 DEFAULT_DOC: Store the journal inside its cache_dir
5317 Location for the cache "swap.state" file. This index file holds
5318 the metadata of objects saved on disk. It is used to rebuild
5319 the cache during startup. Normally this file resides in each
5320 'cache_dir' directory, but you may specify an alternate
5321 pathname here. Note you must give a full filename, not just
5322 a directory. Since this is the index for the whole object
5323 list you CANNOT periodically rotate it!
5325 If %s can be used in the file name it will be replaced with a
5326 a representation of the cache_dir name where each / is replaced
5327 with '.'. This is needed to allow adding/removing cache_dir
5328 lines when cache_swap_log is being used.
5330 If have more than one 'cache_dir', and %s is not used in the name
5331 these swap logs will have names such as:
5337 The numbered extension (which is added automatically)
5338 corresponds to the order of the 'cache_dir' lines in this
5339 configuration file. If you change the order of the 'cache_dir'
5340 lines in this file, these index files will NOT correspond to
5341 the correct 'cache_dir' entry (unless you manually rename
5342 them). We recommend you do NOT use this option. It is
5343 better to keep these index files in each 'cache_dir' directory.
5346 NAME: logfile_rotate
5349 LOC: Config.Log.rotateNumber
5351 Specifies the default number of logfile rotations to make when you
5352 type 'squid -k rotate'. The default is 10, which will rotate
5353 with extensions 0 through 9. Setting logfile_rotate to 0 will
5354 disable the file name rotation, but the logfiles are still closed
5355 and re-opened. This will enable you to rename the logfiles
5356 yourself just before sending the rotate signal.
5358 Note, from Squid-3.1 this option is only a default for cache.log,
5359 that log can be rotated separately by using debug_options.
5361 Note, from Squid-4 this option is only a default for access.log
5362 recorded by stdio: module. Those logs can be rotated separately by
5363 using the rotate=N option on their access_log directive.
5365 Note, the 'squid -k rotate' command normally sends a USR1
5366 signal to the running squid process. In certain situations
5367 (e.g. on Linux with Async I/O), USR1 is used for other
5368 purposes, so -k rotate uses another signal. It is best to get
5369 in the habit of using 'squid -k rotate' instead of 'kill -USR1
5376 DEFAULT: @DEFAULT_MIME_TABLE@
5377 LOC: Config.mimeTablePathname
5379 Path to Squid's icon configuration file.
5381 You shouldn't need to change this, but the default file contains
5382 examples and formatting information if you do.
5388 LOC: Config.onoff.log_mime_hdrs
5391 The Cache can record both the request and the response MIME
5392 headers for each HTTP transaction. The headers are encoded
5393 safely and will appear as two bracketed fields at the end of
5394 the access log (for either the native or httpd-emulated log
5395 formats). To enable this logging set log_mime_hdrs to 'on'.
5400 DEFAULT: @DEFAULT_PID_FILE@
5401 LOC: Config.pidFilename
5403 A filename to write the process-id to. To disable, enter "none".
5406 NAME: client_netmask
5408 LOC: Config.Addrs.client_netmask
5410 DEFAULT_DOC: Log full client IP address
5412 A netmask for client addresses in logfiles and cachemgr output.
5413 Change this to protect the privacy of your cache clients.
5414 A netmask of 255.255.255.0 will log all IP's in that range with
5415 the last digit set to '0'.
5418 NAME: strip_query_terms
5420 LOC: Config.onoff.strip_query_terms
5423 By default, Squid strips query terms from requested URLs before
5424 logging. This protects your user's privacy and reduces log size.
5426 When investigating HIT/MISS or other caching behaviour you
5427 will need to disable this to see the full URL used by Squid.
5434 LOC: Config.onoff.buffered_logs
5436 Whether to write/send access_log records ASAP or accumulate them and
5437 then write/send them in larger chunks. Buffering may improve
5438 performance because it decreases the number of I/Os. However,
5439 buffering increases the delay before log records become available to
5440 the final recipient (e.g., a disk file or logging daemon) and,
5441 hence, increases the risk of log records loss.
5443 Note that even when buffered_logs are off, Squid may have to buffer
5444 records if it cannot write/send them immediately due to pending I/Os
5445 (e.g., the I/O writing the previous log record) or connectivity loss.
5447 Currently honored by 'daemon' and 'tcp' access_log modules only.
5450 NAME: netdb_filename
5452 DEFAULT: stdio:@DEFAULT_NETDB_FILE@
5453 LOC: Config.netdbFilename
5456 Where Squid stores it's netdb journal.
5457 When enabled this journal preserves netdb state between restarts.
5459 To disable, enter "none".
5463 TYPE: Security::KeyLog*
5465 LOC: Config.Log.tlsKeys
5468 Configures whether and where Squid records pre-master secret and
5469 related encryption details for TLS connections accepted or established
5470 by Squid. These connections include connections accepted at
5471 https_port, TLS connections opened to origin servers/cache_peers/ICAP
5472 services, and TLS tunnels bumped by Squid using the SslBump feature.
5473 This log (a.k.a. SSLKEYLOGFILE) is meant for triage with traffic
5474 inspection tools like Wireshark.
5476 tls_key_log <destination> [options] [if [!]<acl>...]
5478 WARNING: This log allows anybody to decrypt the corresponding
5479 encrypted TLS connections, both in-flight and postmortem.
5481 At most one log file is supported at this time. Repeated tls_key_log
5482 directives are treated as fatal configuration errors. By default, no
5483 log is created or updated.
5485 If the log file does not exist, Squid creates it. Otherwise, Squid
5486 appends an existing log file.
5488 The directive is consulted whenever a TLS connection is accepted or
5489 established by Squid. TLS connections that fail the handshake may be
5490 logged if Squid got enough information to form a log record. A record
5491 is logged only if all of the configured ACLs match.
5493 While transport-related ACLs like src and dst should work, Squid may
5494 not have access to higher-level information. For example, when logging
5495 accepted https_port connections, Squid does not yet have access to the
5496 expected HTTPS request. Similarly, an HTTPS response is not available
5497 when logging most TLS connections established by Squid.
5499 The log record format is meant to be compatible with TLS deciphering
5500 features of Wireshark which relies on fields like CLIENT_RANDOM and
5501 RSA Master-Key. A single log record usually spans multiple lines.
5502 Technical documentation for that format is maintained inside the
5503 Wireshark code (e.g., see tls_keylog_process_lines() comments as of
5504 Wireshark commit e3d44136f0f0026c5e893fa249f458073f3b7328). TLS key
5505 log does not support custom record formats.
5507 This clause only supports fast acl types.
5508 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
5510 See access_log's <module>:<place> parameter for a list of supported
5511 logging destinations.
5513 TLS key log supports all access_log key=value options with the
5514 exception of logformat=name.
5516 Requires Squid built with OpenSSL support.
5521 OPTIONS FOR TROUBLESHOOTING
5522 -----------------------------------------------------------------------------
5527 DEFAULT_IF_NONE: @DEFAULT_CACHE_LOG@
5528 LOC: Debug::cache_log
5530 Squid administrative logging file.
5532 This is where general information about Squid behavior goes. You can
5533 increase the amount of data logged to this file and how often it is
5534 rotated with "debug_options"
5540 DEFAULT_DOC: Log all critical and important messages.
5541 LOC: Debug::debugOptions
5543 Logging options are set as section,level where each source file
5544 is assigned a unique section. Lower levels result in less
5545 output, Full debugging (level 9) can result in a very large
5546 log file, so be careful.
5548 The magic word "ALL" sets debugging levels for all sections.
5549 The default is to run with "ALL,1" to record important warnings.
5551 The rotate=N option can be used to keep more or less of these logs
5552 than would otherwise be kept by logfile_rotate.
5553 For most uses a single log should be enough to monitor current
5554 events affecting Squid.
5559 LOC: Config.coredump_dir
5560 DEFAULT_IF_NONE: none
5561 DEFAULT_DOC: Use the directory from where Squid was started.
5563 By default Squid leaves core files in the directory from where
5564 it was started. If you set 'coredump_dir' to a directory
5565 that exists, Squid will chdir() to that directory at startup
5566 and coredump files will be left there.
5570 # Leave coredumps in the first cache dir
5571 coredump_dir @DEFAULT_SWAP_DIR@
5577 OPTIONS FOR FTP GATEWAYING
5578 -----------------------------------------------------------------------------
5584 LOC: Config.Ftp.anon_user
5586 If you want the anonymous login password to be more informative
5587 (and enable the use of picky FTP servers), set this to something
5588 reasonable for your domain, like wwwuser@somewhere.net
5590 The reason why this is domainless by default is the
5591 request can be made on the behalf of a user in any domain,
5592 depending on how the cache is used.
5593 Some FTP server also validate the email address is valid
5594 (for example perl.com).
5600 LOC: Config.Ftp.passive
5602 If your firewall does not allow Squid to use passive
5603 connections, turn off this option.
5605 Use of ftp_epsv_all option requires this to be ON.
5611 LOC: Config.Ftp.epsv_all
5613 FTP Protocol extensions permit the use of a special "EPSV ALL" command.
5615 NATs may be able to put the connection on a "fast path" through the
5616 translator, as the EPRT command will never be used and therefore,
5617 translation of the data portion of the segments will never be needed.
5619 When a client only expects to do two-way FTP transfers this may be
5621 If squid finds that it must do a three-way FTP transfer after issuing
5622 an EPSV ALL command, the FTP session will fail.
5624 If you have any doubts about this option do not use it.
5625 Squid will nicely attempt all other connection methods.
5627 Requires ftp_passive to be ON (default) for any effect.
5633 LOC: Config.accessList.ftp_epsv
5635 FTP Protocol extensions permit the use of a special "EPSV" command.
5637 NATs may be able to put the connection on a "fast path" through the
5638 translator using EPSV, as the EPRT command will never be used
5639 and therefore, translation of the data portion of the segments
5640 will never be needed.
5642 EPSV is often required to interoperate with FTP servers on IPv6
5643 networks. On the other hand, it may break some IPv4 servers.
5645 By default, EPSV may try EPSV with any FTP server. To fine tune
5646 that decision, you may restrict EPSV to certain clients or servers
5649 ftp_epsv allow|deny al1 acl2 ...
5651 WARNING: Disabling EPSV may cause problems with external NAT and IPv6.
5653 Only fast ACLs are supported.
5654 Requires ftp_passive to be ON (default) for any effect.
5660 LOC: Config.Ftp.eprt
5662 FTP Protocol extensions permit the use of a special "EPRT" command.
5664 This extension provides a protocol neutral alternative to the
5665 IPv4-only PORT command. When supported it enables active FTP data
5666 channels over IPv6 and efficient NAT handling.
5668 Turning this OFF will prevent EPRT being attempted and will skip
5669 straight to using PORT for IPv4 servers.
5671 Some devices are known to not handle this extension correctly and
5672 may result in crashes. Devices which support EPRT enough to fail
5673 cleanly will result in Squid attempting PORT anyway. This directive
5674 should only be disabled when EPRT results in device failures.
5676 WARNING: Doing so will convert Squid back to the old behavior with all
5677 the related problems with external NAT devices/layers and IPv4-only FTP.
5680 NAME: ftp_sanitycheck
5683 LOC: Config.Ftp.sanitycheck
5685 For security and data integrity reasons Squid by default performs
5686 sanity checks of the addresses of FTP data connections ensure the
5687 data connection is to the requested server. If you need to allow
5688 FTP connections to servers using another IP address for the data
5689 connection turn this off.
5692 NAME: ftp_telnet_protocol
5695 LOC: Config.Ftp.telnet
5697 The FTP protocol is officially defined to use the telnet protocol
5698 as transport channel for the control connection. However, many
5699 implementations are broken and does not respect this aspect of
5702 If you have trouble accessing files with ASCII code 255 in the
5703 path or similar problems involving this ASCII code you can
5704 try setting this directive to off. If that helps, report to the
5705 operator of the FTP server in question that their FTP server
5706 is broken and does not follow the FTP standard.
5710 OPTIONS FOR EXTERNAL SUPPORT PROGRAMS
5711 -----------------------------------------------------------------------------
5716 DEFAULT: @DEFAULT_DISKD@
5717 LOC: Config.Program.diskd
5719 Specify the location of the diskd executable.
5720 Note this is only useful if you have compiled in
5721 diskd as one of the store io modules.
5724 NAME: unlinkd_program
5727 DEFAULT: @DEFAULT_UNLINKD@
5728 LOC: Config.Program.unlinkd
5730 Specify the location of the executable for file deletion process.
5733 NAME: pinger_program
5736 DEFAULT: @DEFAULT_PINGER@
5739 Specify the location of the executable for the pinger process.
5748 Control whether the pinger is active at run-time.
5749 Enables turning ICMP pinger on and off with a simple
5750 squid -k reconfigure.
5755 OPTIONS FOR URL REWRITING
5756 -----------------------------------------------------------------------------
5759 NAME: url_rewrite_program redirect_program
5761 LOC: Config.Program.redirect
5764 The name and command line parameters of an admin-provided executable
5765 for redirecting clients or adjusting/replacing client request URLs.
5767 This helper is consulted after the received request is cleared by
5768 http_access and adapted using eICAP/ICAP services (if any). If the
5769 helper does not redirect the client, Squid checks adapted_http_access
5770 and may consult the cache or forward the request to the next hop.
5773 For each request, the helper gets one line in the following format:
5775 [channel-ID <SP>] request-URL [<SP> extras] <NL>
5777 Use url_rewrite_extras to configure what Squid sends as 'extras'.
5780 The helper must reply to each query using a single line:
5782 [channel-ID <SP>] result [<SP> kv-pairs] <NL>
5784 The result section must match exactly one of the following outcomes:
5786 OK [status=30N] url="..."
5788 Redirect the client to a URL supplied in the 'url' parameter.
5789 Optional 'status' specifies the status code to send to the
5790 client in Squid's HTTP redirect response. It must be one of
5791 the standard HTTP redirect status codes: 301, 302, 303, 307,
5792 or 308. When no specific status is requested, Squid uses 302.
5794 OK rewrite-url="..."
5796 Replace the current request URL with the one supplied in the
5797 'rewrite-url' parameter. Squid fetches the resource specified
5798 by the new URL and forwards the received response (or its
5799 cached copy) to the client.
5801 WARNING: Avoid rewriting URLs! When possible, redirect the
5802 client using an "OK url=..." helper response instead.
5803 Rewriting URLs may create inconsistent requests and/or break
5804 synchronization between internal client and origin server
5805 states, especially when URLs or other message parts contain
5806 snippets of that state. For example, Squid does not adjust
5807 Location headers and embedded URLs after the helper rewrites
5811 Keep the client request intact.
5814 Keep the client request intact.
5817 A helper problem that should be reported to the Squid admin
5818 via a level-1 cache.log message. The 'message' parameter is
5819 reserved for specifying the log message.
5821 In addition to the kv-pairs mentioned above, Squid also understands
5822 the following optional kv-pairs in URL rewriter responses:
5825 Associates a TAG with the client TCP connection.
5827 The clt_conn_tag=TAG pair is treated as a regular transaction
5828 annotation for the current request and also annotates future
5829 requests on the same client connection. A helper may update
5830 the TAG during subsequent requests by returning a new kv-pair.
5833 Helper messages contain the channel-ID part if and only if the
5834 url_rewrite_children directive specifies positive concurrency. As a
5835 channel-ID value, Squid sends a number between 0 and concurrency-1.
5836 The helper must echo back the received channel-ID in its response.
5838 By default, Squid does not use a URL rewriter.
5841 NAME: url_rewrite_children redirect_children
5842 TYPE: HelperChildConfig
5843 DEFAULT: 20 startup=0 idle=1 concurrency=0
5844 LOC: Config.redirectChildren
5846 Specifies the maximum number of redirector processes that Squid may
5847 spawn (numberofchildren) and several related options. Using too few of
5848 these helper processes (a.k.a. "helpers") creates request queues.
5849 Using too many helpers wastes your system resources.
5851 Usage: numberofchildren [option]...
5853 The startup= and idle= options allow some measure of skew in your
5858 Sets a minimum of how many processes are to be spawned when Squid
5859 starts or reconfigures. When set to zero the first request will
5860 cause spawning of the first child process to handle it.
5862 Starting too few will cause an initial slowdown in traffic as Squid
5863 attempts to simultaneously spawn enough processes to cope.
5867 Sets a minimum of how many processes Squid is to try and keep available
5868 at all times. When traffic begins to rise above what the existing
5869 processes can handle this many more will be spawned up to the maximum
5870 configured. A minimum setting of 1 is required.
5874 The number of requests each redirector helper can handle in
5875 parallel. Defaults to 0 which indicates the redirector
5876 is a old-style single threaded redirector.
5878 When this directive is set to a value >= 1 then the protocol
5879 used to communicate with the helper is modified to include
5880 an ID in front of the request/response. The ID from the request
5881 must be echoed back with the response to that request.
5885 Sets the maximum number of queued requests. A request is queued when
5886 no existing child can accept it due to concurrency limit and no new
5887 child can be started due to numberofchildren limit. The default
5888 maximum is zero if url_rewrite_bypass is enabled and
5889 2*numberofchildren otherwise. If the queued requests exceed queue size
5890 and redirector_bypass configuration option is set, then redirector is
5891 bypassed. Otherwise, Squid is allowed to temporarily exceed the
5892 configured maximum, marking the affected helper as "overloaded". If
5893 the helper overload lasts more than 3 minutes, the action prescribed
5894 by the on-persistent-overload option applies.
5896 on-persistent-overload=action
5898 Specifies Squid reaction to a new helper request arriving when the helper
5899 has been overloaded for more that 3 minutes already. The number of queued
5900 requests determines whether the helper is overloaded (see the queue-size
5903 Two actions are supported:
5905 die Squid worker quits. This is the default behavior.
5907 ERR Squid treats the helper request as if it was
5908 immediately submitted, and the helper immediately
5909 replied with an ERR response. This action has no effect
5910 on the already queued and in-progress helper requests.
5913 NAME: url_rewrite_host_header redirect_rewrites_host_header
5916 LOC: Config.onoff.redir_rewrites_host
5918 To preserve same-origin security policies in browsers and
5919 prevent Host: header forgery by redirectors Squid rewrites
5920 any Host: header in redirected requests.
5922 If you are running an accelerator this may not be a wanted
5923 effect of a redirector. This directive enables you disable
5924 Host: alteration in reverse-proxy traffic.
5926 WARNING: Entries are cached on the result of the URL rewriting
5927 process, so be careful if you have domain-virtual hosts.
5929 WARNING: Squid and other software verifies the URL and Host
5930 are matching, so be careful not to relay through other proxies
5931 or inspecting firewalls with this disabled.
5934 NAME: url_rewrite_access redirector_access
5937 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
5938 LOC: Config.accessList.redirector
5940 If defined, this access list specifies which requests are
5941 sent to the redirector processes.
5943 This clause supports both fast and slow acl types.
5944 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
5947 NAME: url_rewrite_bypass redirector_bypass
5949 LOC: Config.onoff.redirector_bypass
5952 When this is 'on', a request will not go through the
5953 redirector if all the helpers are busy. If this is 'off' and the
5954 redirector queue grows too large, the action is prescribed by the
5955 on-persistent-overload option. You should only enable this if the
5956 redirectors are not critical to your caching system. If you use
5957 redirectors for access control, and you enable this option,
5958 users may have access to pages they should not
5959 be allowed to request.
5961 Enabling this option sets the default url_rewrite_children queue-size
5965 NAME: url_rewrite_extras
5966 TYPE: TokenOrQuotedString
5967 LOC: Config.redirector_extras
5968 DEFAULT: "%>a/%>A %un %>rm myip=%la myport=%lp"
5970 Specifies a string to be append to request line format for the
5971 rewriter helper. "Quoted" format values may contain spaces and
5972 logformat %macros. In theory, any logformat %macro can be used.
5973 In practice, a %macro expands as a dash (-) if the helper request is
5974 sent before the required macro information is available to Squid.
5977 NAME: url_rewrite_timeout
5978 TYPE: UrlHelperTimeout
5979 LOC: Config.onUrlRewriteTimeout
5981 DEFAULT_DOC: Squid waits for the helper response forever
5983 Squid times active requests to redirector. The timeout value and Squid
5984 reaction to a timed out request are configurable using the following
5987 url_rewrite_timeout timeout time-units on_timeout=<action> [response=<quoted-response>]
5989 supported timeout actions:
5990 fail Squid return a ERR_GATEWAY_FAILURE error page
5992 bypass Do not re-write the URL
5994 retry Send the lookup to the helper again
5996 use_configured_response
5997 Use the <quoted-response> as helper response
6001 OPTIONS FOR STORE ID
6002 -----------------------------------------------------------------------------
6005 NAME: store_id_program storeurl_rewrite_program
6007 LOC: Config.Program.store_id
6010 Specify the location of the executable StoreID helper to use.
6011 Since they can perform almost any function there isn't one included.
6013 For each requested URL, the helper will receive one line with the format
6015 [channel-ID <SP>] URL [<SP> extras]<NL>
6018 After processing the request the helper must reply using the following format:
6020 [channel-ID <SP>] result [<SP> kv-pairs]
6022 The result code can be:
6025 Use the StoreID supplied in 'store-id='.
6028 The default is to use HTTP request URL as the store ID.
6031 An internal error occurred in the helper, preventing
6032 a result being identified.
6034 In addition to the above kv-pairs Squid also understands the following
6035 optional kv-pairs received from URL rewriters:
6037 Associates a TAG with the client TCP connection.
6038 Please see url_rewrite_program related documentation for this
6041 Helper programs should be prepared to receive and possibly ignore
6042 additional whitespace-separated tokens on each input line.
6044 When using the concurrency= option the protocol is changed by
6045 introducing a query channel tag in front of the request/response.
6046 The query channel tag is a number between 0 and concurrency-1.
6047 This value must be echoed back unchanged to Squid as the first part
6048 of the response relating to its request.
6050 NOTE: when using StoreID refresh_pattern will apply to the StoreID
6051 returned from the helper and not the URL.
6053 WARNING: Wrong StoreID value returned by a careless helper may result
6054 in the wrong cached response returned to the user.
6056 By default, a StoreID helper is not used.
6059 NAME: store_id_extras
6060 TYPE: TokenOrQuotedString
6061 LOC: Config.storeId_extras
6062 DEFAULT: "%>a/%>A %un %>rm myip=%la myport=%lp"
6064 Specifies a string to be append to request line format for the
6065 StoreId helper. "Quoted" format values may contain spaces and
6066 logformat %macros. In theory, any logformat %macro can be used.
6067 In practice, a %macro expands as a dash (-) if the helper request is
6068 sent before the required macro information is available to Squid.
6071 NAME: store_id_children storeurl_rewrite_children
6072 TYPE: HelperChildConfig
6073 DEFAULT: 20 startup=0 idle=1 concurrency=0
6074 LOC: Config.storeIdChildren
6076 Specifies the maximum number of StoreID helper processes that Squid
6077 may spawn (numberofchildren) and several related options. Using
6078 too few of these helper processes (a.k.a. "helpers") creates request
6079 queues. Using too many helpers wastes your system resources.
6081 Usage: numberofchildren [option]...
6083 The startup= and idle= options allow some measure of skew in your
6088 Sets a minimum of how many processes are to be spawned when Squid
6089 starts or reconfigures. When set to zero the first request will
6090 cause spawning of the first child process to handle it.
6092 Starting too few will cause an initial slowdown in traffic as Squid
6093 attempts to simultaneously spawn enough processes to cope.
6097 Sets a minimum of how many processes Squid is to try and keep available
6098 at all times. When traffic begins to rise above what the existing
6099 processes can handle this many more will be spawned up to the maximum
6100 configured. A minimum setting of 1 is required.
6104 The number of requests each storeID helper can handle in
6105 parallel. Defaults to 0 which indicates the helper
6106 is a old-style single threaded program.
6108 When this directive is set to a value >= 1 then the protocol
6109 used to communicate with the helper is modified to include
6110 an ID in front of the request/response. The ID from the request
6111 must be echoed back with the response to that request.
6115 Sets the maximum number of queued requests to N. A request is queued
6116 when no existing child can accept it due to concurrency limit and no
6117 new child can be started due to numberofchildren limit. The default
6118 maximum is 2*numberofchildren. If the queued requests exceed queue
6119 size and redirector_bypass configuration option is set, then
6120 redirector is bypassed. Otherwise, Squid is allowed to temporarily
6121 exceed the configured maximum, marking the affected helper as
6122 "overloaded". If the helper overload lasts more than 3 minutes, the
6123 action prescribed by the on-persistent-overload option applies.
6125 on-persistent-overload=action
6127 Specifies Squid reaction to a new helper request arriving when the helper
6128 has been overloaded for more that 3 minutes already. The number of queued
6129 requests determines whether the helper is overloaded (see the queue-size
6132 Two actions are supported:
6134 die Squid worker quits. This is the default behavior.
6136 ERR Squid treats the helper request as if it was
6137 immediately submitted, and the helper immediately
6138 replied with an ERR response. This action has no effect
6139 on the already queued and in-progress helper requests.
6142 NAME: store_id_access storeurl_rewrite_access
6145 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
6146 LOC: Config.accessList.store_id
6148 If defined, this access list specifies which requests are
6149 sent to the StoreID processes. By default all requests
6152 This clause supports both fast and slow acl types.
6153 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6156 NAME: store_id_bypass storeurl_rewrite_bypass
6158 LOC: Config.onoff.store_id_bypass
6161 When this is 'on', a request will not go through the
6162 helper if all helpers are busy. If this is 'off' and the helper
6163 queue grows too large, the action is prescribed by the
6164 on-persistent-overload option. You should only enable this if the
6165 helpers are not critical to your caching system. If you use
6166 helpers for critical caching components, and you enable this
6167 option, users may not get objects from cache.
6168 This options sets default queue-size option of the store_id_children
6173 OPTIONS FOR TUNING THE CACHE
6174 -----------------------------------------------------------------------------
6177 NAME: cache no_cache
6180 DEFAULT_DOC: By default, this directive is unused and has no effect.
6181 LOC: Config.accessList.noCache
6183 Requests denied by this directive will not be served from the cache
6184 and their responses will not be stored in the cache. This directive
6185 has no effect on other transactions and on already cached responses.
6187 This clause supports both fast and slow acl types.
6188 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6190 This and the two other similar caching directives listed below are
6191 checked at different transaction processing stages, have different
6192 access to response information, affect different cache operations,
6193 and differ in slow ACLs support:
6195 * cache: Checked before Squid makes a hit/miss determination.
6196 No access to reply information!
6197 Denies both serving a hit and storing a miss.
6198 Supports both fast and slow ACLs.
6199 * send_hit: Checked after a hit was detected.
6200 Has access to reply (hit) information.
6201 Denies serving a hit only.
6202 Supports fast ACLs only.
6203 * store_miss: Checked before storing a cachable miss.
6204 Has access to reply (miss) information.
6205 Denies storing a miss only.
6206 Supports fast ACLs only.
6208 If you are not sure which of the three directives to use, apply the
6209 following decision logic:
6211 * If your ACL(s) are of slow type _and_ need response info, redesign.
6212 Squid does not support that particular combination at this time.
6214 * If your directive ACL(s) are of slow type, use "cache"; and/or
6215 * if your directive ACL(s) need no response info, use "cache".
6217 * If you do not want the response cached, use store_miss; and/or
6218 * if you do not want a hit on a cached response, use send_hit.
6224 DEFAULT_DOC: By default, this directive is unused and has no effect.
6225 LOC: Config.accessList.sendHit
6227 Responses denied by this directive will not be served from the cache
6228 (but may still be cached, see store_miss). This directive has no
6229 effect on the responses it allows and on the cached objects.
6231 Please see the "cache" directive for a summary of differences among
6232 store_miss, send_hit, and cache directives.
6234 Unlike the "cache" directive, send_hit only supports fast acl
6235 types. See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6239 # apply custom Store ID mapping to some URLs
6240 acl MapMe dstdomain .c.example.com
6241 store_id_program ...
6242 store_id_access allow MapMe
6244 # but prevent caching of special responses
6245 # such as 302 redirects that cause StoreID loops
6246 acl Ordinary http_status 200-299
6247 store_miss deny MapMe !Ordinary
6249 # and do not serve any previously stored special responses
6250 # from the cache (in case they were already cached before
6251 # the above store_miss rule was in effect).
6252 send_hit deny MapMe !Ordinary
6258 DEFAULT_DOC: By default, this directive is unused and has no effect.
6259 LOC: Config.accessList.storeMiss
6261 Responses denied by this directive will not be cached (but may still
6262 be served from the cache, see send_hit). This directive has no
6263 effect on the responses it allows and on the already cached responses.
6265 Please see the "cache" directive for a summary of differences among
6266 store_miss, send_hit, and cache directives. See the
6267 send_hit directive for a usage example.
6269 Unlike the "cache" directive, store_miss only supports fast acl
6270 types. See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6276 LOC: Config.maxStale
6279 This option puts an upper limit on how stale content Squid
6280 will serve from the cache if cache validation fails.
6281 Can be overridden by the refresh_pattern max-stale option.
6284 NAME: refresh_pattern
6285 TYPE: refreshpattern
6289 usage: refresh_pattern [-i] regex min percent max [options]
6291 By default, regular expressions are CASE-SENSITIVE. To make
6292 them case-insensitive, use the -i option.
6294 'Min' is the time (in minutes) an object without an explicit
6295 expiry time should be considered fresh. The recommended
6296 value is 0, any higher values may cause dynamic applications
6297 to be erroneously cached unless the application designer
6298 has taken the appropriate actions.
6300 'Percent' is a percentage of the objects age (time since last
6301 modification age) an object without explicit expiry time
6302 will be considered fresh.
6304 'Max' is an upper limit on how long objects without an explicit
6305 expiry time will be considered fresh. The value is also used
6306 to form Cache-Control: max-age header for a request sent from
6307 Squid to origin/parent.
6309 options: override-expire
6319 override-expire enforces min age even if the server
6320 sent an explicit expiry time (e.g., with the
6321 Expires: header or Cache-Control: max-age). Doing this
6322 VIOLATES the HTTP standard. Enabling this feature
6323 could make you liable for problems which it causes.
6325 Note: override-expire does not enforce staleness - it only extends
6326 freshness / min. If the server returns a Expires time which
6327 is longer than your max time, Squid will still consider
6328 the object fresh for that period of time.
6330 override-lastmod enforces min age even on objects
6331 that were modified recently.
6333 reload-into-ims changes a client no-cache or ``reload''
6334 request for a cached entry into a conditional request using
6335 If-Modified-Since and/or If-None-Match headers, provided the
6336 cached entry has a Last-Modified and/or a strong ETag header.
6337 Doing this VIOLATES the HTTP standard. Enabling this feature
6338 could make you liable for problems which it causes.
6340 ignore-reload ignores a client no-cache or ``reload''
6341 header. Doing this VIOLATES the HTTP standard. Enabling
6342 this feature could make you liable for problems which
6345 ignore-no-store ignores any ``Cache-control: no-store''
6346 headers received from a server. Doing this VIOLATES
6347 the HTTP standard. Enabling this feature could make you
6348 liable for problems which it causes.
6350 ignore-private ignores any ``Cache-control: private''
6351 headers received from a server. Doing this VIOLATES
6352 the HTTP standard. Enabling this feature could make you
6353 liable for problems which it causes.
6355 refresh-ims causes squid to contact the origin server
6356 when a client issues an If-Modified-Since request. This
6357 ensures that the client will receive an updated version
6358 if one is available.
6360 store-stale stores responses even if they don't have explicit
6361 freshness or a validator (i.e., Last-Modified or an ETag)
6362 present, or if they're already stale. By default, Squid will
6363 not cache such responses because they usually can't be
6364 reused. Note that such responses will be stale by default.
6366 max-stale=NN provide a maximum staleness factor. Squid won't
6367 serve objects more stale than this even if it failed to
6368 validate the object. Default: use the max_stale global limit.
6370 Basically a cached object is:
6372 FRESH if expire > now, else STALE
6374 FRESH if lm-factor < percent, else STALE
6378 The refresh_pattern lines are checked in the order listed here.
6379 The first entry which matches is used. If none of the entries
6380 match the default will be used.
6382 Note, you must uncomment all the default lines if you want
6383 to change one. The default setting is only active if none is
6389 # Add any of your own refresh_pattern entries above these.
6391 refresh_pattern ^ftp: 1440 20% 10080
6392 refresh_pattern ^gopher: 1440 0% 1440
6393 refresh_pattern -i (/cgi-bin/|\?) 0 0% 0
6394 refresh_pattern . 0 20% 4320
6398 NAME: quick_abort_min
6402 LOC: Config.quickAbort.min
6405 NAME: quick_abort_max
6409 LOC: Config.quickAbort.max
6412 NAME: quick_abort_pct
6416 LOC: Config.quickAbort.pct
6418 The cache by default continues downloading aborted requests
6419 which are almost completed (less than 16 KB remaining). This
6420 may be undesirable on slow (e.g. SLIP) links and/or very busy
6421 caches. Impatient users may tie up file descriptors and
6422 bandwidth by repeatedly requesting and immediately aborting
6425 When the user aborts a request, Squid will check the
6426 quick_abort values to the amount of data transferred until
6429 If the transfer has less than 'quick_abort_min' KB remaining,
6430 it will finish the retrieval.
6432 If the transfer has more than 'quick_abort_max' KB remaining,
6433 it will abort the retrieval.
6435 If more than 'quick_abort_pct' of the transfer has completed,
6436 it will finish the retrieval.
6438 If you do not want any retrieval to continue after the client
6439 has aborted, set both 'quick_abort_min' and 'quick_abort_max'
6442 If you want retrievals to always continue if they are being
6443 cached set 'quick_abort_min' to '-1 KB'.
6446 NAME: read_ahead_gap
6447 COMMENT: buffer-size
6449 LOC: Config.readAheadGap
6452 The amount of data the cache will buffer ahead of what has been
6453 sent to the client when retrieving an object from another server.
6457 IFDEF: USE_HTTP_VIOLATIONS
6460 LOC: Config.negativeTtl
6463 Set the Default Time-to-Live (TTL) for failed requests.
6464 Certain types of failures (such as "connection refused" and
6465 "404 Not Found") are able to be negatively-cached for a short time.
6466 Modern web servers should provide Expires: header, however if they
6467 do not this can provide a minimum TTL.
6468 The default is not to cache errors with unknown expiry details.
6470 Note that this is different from negative caching of DNS lookups.
6472 WARNING: Doing this VIOLATES the HTTP standard. Enabling
6473 this feature could make you liable for problems which it
6477 NAME: positive_dns_ttl
6480 LOC: Config.positiveDnsTtl
6483 Upper limit on how long Squid will cache positive DNS responses.
6484 Default is 6 hours (360 minutes). This directive must be set
6485 larger than negative_dns_ttl.
6488 NAME: negative_dns_ttl
6491 LOC: Config.negativeDnsTtl
6494 Time-to-Live (TTL) for negative caching of failed DNS lookups.
6495 This also sets the lower cache limit on positive lookups.
6496 Minimum value is 1 second, and it is not recommendable to go
6497 much below 10 seconds.
6500 NAME: range_offset_limit
6501 COMMENT: size [acl acl...]
6503 LOC: Config.rangeOffsetLimit
6506 usage: (size) [units] [[!]aclname]
6508 Sets an upper limit on how far (number of bytes) into the file
6509 a Range request may be to cause Squid to prefetch the whole file.
6510 If beyond this limit, Squid forwards the Range request as it is and
6511 the result is NOT cached.
6513 This is to stop a far ahead range request (lets say start at 17MB)
6514 from making Squid fetch the whole object up to that point before
6515 sending anything to the client.
6517 Multiple range_offset_limit lines may be specified, and they will
6518 be searched from top to bottom on each request until a match is found.
6519 The first match found will be used. If no line matches a request, the
6520 default limit of 0 bytes will be used.
6522 'size' is the limit specified as a number of units.
6524 'units' specifies whether to use bytes, KB, MB, etc.
6525 If no units are specified bytes are assumed.
6527 A size of 0 causes Squid to never fetch more than the
6528 client requested. (default)
6530 A size of 'none' causes Squid to always fetch the object from the
6531 beginning so it may cache the result. (2.0 style)
6533 'aclname' is the name of a defined ACL.
6535 NP: Using 'none' as the byte value here will override any quick_abort settings
6536 that may otherwise apply to the range request. The range request will
6537 be fully fetched from start to finish regardless of the client
6538 actions. This affects bandwidth usage.
6541 NAME: minimum_expiry_time
6544 LOC: Config.minimum_expiry_time
6547 The minimum caching time according to (Expires - Date)
6548 headers Squid honors if the object can't be revalidated.
6549 The default is 60 seconds.
6551 In reverse proxy environments it might be desirable to honor
6552 shorter object lifetimes. It is most likely better to make
6553 your server return a meaningful Last-Modified header however.
6555 In ESI environments where page fragments often have short
6556 lifetimes, this will often be best set to 0.
6559 NAME: store_avg_object_size
6563 LOC: Config.Store.avgObjectSize
6565 Average object size, used to estimate number of objects your
6566 cache can hold. The default is 13 KB.
6568 This is used to pre-seed the cache index memory allocation to
6569 reduce expensive reallocate operations while handling clients
6570 traffic. Too-large values may result in memory allocation during
6571 peak traffic, too-small values will result in wasted memory.
6573 Check the cache manager 'info' report metrics for the real
6574 object sizes seen by your Squid before tuning this.
6577 NAME: store_objects_per_bucket
6580 LOC: Config.Store.objectsPerBucket
6582 Target number of objects per bucket in the store hash table.
6583 Lowering this value increases the total number of buckets and
6584 also the storage maintenance rate. The default is 20.
6589 -----------------------------------------------------------------------------
6592 NAME: request_header_max_size
6596 LOC: Config.maxRequestHeaderSize
6598 This specifies the maximum size for HTTP headers in a request.
6599 Request headers are usually relatively small (about 512 bytes).
6600 Placing a limit on the request header size will catch certain
6601 bugs (for example with persistent connections) and possibly
6602 buffer-overflow or denial-of-service attacks.
6605 NAME: reply_header_max_size
6609 LOC: Config.maxReplyHeaderSize
6611 This specifies the maximum size for HTTP headers in a reply.
6612 Reply headers are usually relatively small (about 512 bytes).
6613 Placing a limit on the reply header size will catch certain
6614 bugs (for example with persistent connections) and possibly
6615 buffer-overflow or denial-of-service attacks.
6618 NAME: request_body_max_size
6622 DEFAULT_DOC: No limit.
6623 LOC: Config.maxRequestBodySize
6625 This specifies the maximum size for an HTTP request body.
6626 In other words, the maximum size of a PUT/POST request.
6627 A user who attempts to send a request with a body larger
6628 than this limit receives an "Invalid Request" error message.
6629 If you set this parameter to a zero (the default), there will
6630 be no limit imposed.
6632 See also client_request_buffer_max_size for an alternative
6633 limitation on client uploads which can be configured.
6636 NAME: client_request_buffer_max_size
6640 LOC: Config.maxRequestBufferSize
6642 This specifies the maximum buffer size of a client request.
6643 It prevents squid eating too much memory when somebody uploads
6648 IFDEF: USE_HTTP_VIOLATIONS
6651 DEFAULT_DOC: Obey RFC 2616.
6652 LOC: Config.accessList.brokenPosts
6654 A list of ACL elements which, if matched, causes Squid to send
6655 an extra CRLF pair after the body of a PUT/POST request.
6657 Some HTTP servers has broken implementations of PUT/POST,
6658 and rely on an extra CRLF pair sent by some WWW clients.
6660 Quote from RFC2616 section 4.1 on this matter:
6662 Note: certain buggy HTTP/1.0 client implementations generate an
6663 extra CRLF's after a POST request. To restate what is explicitly
6664 forbidden by the BNF, an HTTP/1.1 client must not preface or follow
6665 a request with an extra CRLF.
6667 This clause only supports fast acl types.
6668 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6671 acl buggy_server url_regex ^http://....
6672 broken_posts allow buggy_server
6675 NAME: adaptation_uses_indirect_client icap_uses_indirect_client
6678 IFDEF: FOLLOW_X_FORWARDED_FOR&&USE_ADAPTATION
6680 LOC: Adaptation::Config::use_indirect_client
6682 Controls whether the indirect client IP address (instead of the direct
6683 client IP address) is passed to adaptation services.
6685 See also: follow_x_forwarded_for adaptation_send_client_ip
6689 IFDEF: USE_HTTP_VIOLATIONS
6693 LOC: Config.onoff.via
6695 If set (default), Squid will include a Via header in requests and
6696 replies as required by RFC2616.
6699 NAME: vary_ignore_expire
6702 LOC: Config.onoff.vary_ignore_expire
6705 Many HTTP servers supporting Vary gives such objects
6706 immediate expiry time with no cache-control header
6707 when requested by a HTTP/1.0 client. This option
6708 enables Squid to ignore such expiry times until
6709 HTTP/1.1 is fully implemented.
6711 WARNING: If turned on this may eventually cause some
6712 varying objects not intended for caching to get cached.
6715 NAME: request_entities
6717 LOC: Config.onoff.request_entities
6720 Squid defaults to deny GET and HEAD requests with request entities,
6721 as the meaning of such requests are undefined in the HTTP standard
6722 even if not explicitly forbidden.
6724 Set this directive to on if you have clients which insists
6725 on sending request entities in GET or HEAD requests. But be warned
6726 that there is server software (both proxies and web servers) which
6727 can fail to properly process this kind of request which may make you
6728 vulnerable to cache pollution attacks if enabled.
6731 NAME: request_header_access
6732 IFDEF: USE_HTTP_VIOLATIONS
6733 TYPE: http_header_access
6734 LOC: Config.request_header_access
6736 DEFAULT_DOC: No limits.
6738 Usage: request_header_access header_name allow|deny [!]aclname ...
6740 WARNING: Doing this VIOLATES the HTTP standard. Enabling
6741 this feature could make you liable for problems which it
6744 This option replaces the old 'anonymize_headers' and the
6745 older 'http_anonymizer' option with something that is much
6746 more configurable. A list of ACLs for each header name allows
6747 removal of specific header fields under specific conditions.
6749 This option only applies to outgoing HTTP request headers (i.e.,
6750 headers sent by Squid to the next HTTP hop such as a cache peer
6751 or an origin server). The option has no effect during cache hit
6752 detection. The equivalent adaptation vectoring point in ICAP
6753 terminology is post-cache REQMOD.
6755 The option is applied to individual outgoing request header
6756 fields. For each request header field F, Squid uses the first
6757 qualifying sets of request_header_access rules:
6759 1. Rules with header_name equal to F's name.
6760 2. Rules with header_name 'Other', provided F's name is not
6761 on the hard-coded list of commonly used HTTP header names.
6762 3. Rules with header_name 'All'.
6764 Within that qualifying rule set, rule ACLs are checked as usual.
6765 If ACLs of an "allow" rule match, the header field is allowed to
6766 go through as is. If ACLs of a "deny" rule match, the header is
6767 removed and request_header_replace is then checked to identify
6768 if the removed header has a replacement. If no rules within the
6769 set have matching ACLs, the header field is left as is.
6771 For example, to achieve the same behavior as the old
6772 'http_anonymizer standard' option, you should use:
6774 request_header_access From deny all
6775 request_header_access Referer deny all
6776 request_header_access User-Agent deny all
6778 Or, to reproduce the old 'http_anonymizer paranoid' feature
6781 request_header_access Authorization allow all
6782 request_header_access Proxy-Authorization allow all
6783 request_header_access Cache-Control allow all
6784 request_header_access Content-Length allow all
6785 request_header_access Content-Type allow all
6786 request_header_access Date allow all
6787 request_header_access Host allow all
6788 request_header_access If-Modified-Since allow all
6789 request_header_access Pragma allow all
6790 request_header_access Accept allow all
6791 request_header_access Accept-Charset allow all
6792 request_header_access Accept-Encoding allow all
6793 request_header_access Accept-Language allow all
6794 request_header_access Connection allow all
6795 request_header_access All deny all
6797 HTTP reply headers are controlled with the reply_header_access directive.
6799 By default, all headers are allowed (no anonymizing is performed).
6802 NAME: reply_header_access
6803 IFDEF: USE_HTTP_VIOLATIONS
6804 TYPE: http_header_access
6805 LOC: Config.reply_header_access
6807 DEFAULT_DOC: No limits.
6809 Usage: reply_header_access header_name allow|deny [!]aclname ...
6811 WARNING: Doing this VIOLATES the HTTP standard. Enabling
6812 this feature could make you liable for problems which it
6815 This option only applies to reply headers, i.e., from the
6816 server to the client.
6818 This is the same as request_header_access, but in the other
6819 direction. Please see request_header_access for detailed
6822 For example, to achieve the same behavior as the old
6823 'http_anonymizer standard' option, you should use:
6825 reply_header_access Server deny all
6826 reply_header_access WWW-Authenticate deny all
6827 reply_header_access Link deny all
6829 Or, to reproduce the old 'http_anonymizer paranoid' feature
6832 reply_header_access Allow allow all
6833 reply_header_access WWW-Authenticate allow all
6834 reply_header_access Proxy-Authenticate allow all
6835 reply_header_access Cache-Control allow all
6836 reply_header_access Content-Encoding allow all
6837 reply_header_access Content-Length allow all
6838 reply_header_access Content-Type allow all
6839 reply_header_access Date allow all
6840 reply_header_access Expires allow all
6841 reply_header_access Last-Modified allow all
6842 reply_header_access Location allow all
6843 reply_header_access Pragma allow all
6844 reply_header_access Content-Language allow all
6845 reply_header_access Retry-After allow all
6846 reply_header_access Title allow all
6847 reply_header_access Content-Disposition allow all
6848 reply_header_access Connection allow all
6849 reply_header_access All deny all
6851 HTTP request headers are controlled with the request_header_access directive.
6853 By default, all headers are allowed (no anonymizing is
6857 NAME: request_header_replace header_replace
6858 IFDEF: USE_HTTP_VIOLATIONS
6859 TYPE: http_header_replace
6860 LOC: Config.request_header_access
6863 Usage: request_header_replace header_name message
6864 Example: request_header_replace User-Agent Nutscrape/1.0 (CP/M; 8-bit)
6866 This option allows you to change the contents of headers
6867 denied with request_header_access above, by replacing them
6868 with some fixed string.
6870 This only applies to request headers, not reply headers.
6872 By default, headers are removed if denied.
6875 NAME: reply_header_replace
6876 IFDEF: USE_HTTP_VIOLATIONS
6877 TYPE: http_header_replace
6878 LOC: Config.reply_header_access
6881 Usage: reply_header_replace header_name message
6882 Example: reply_header_replace Server Foo/1.0
6884 This option allows you to change the contents of headers
6885 denied with reply_header_access above, by replacing them
6886 with some fixed string.
6888 This only applies to reply headers, not request headers.
6890 By default, headers are removed if denied.
6893 NAME: request_header_add
6894 TYPE: HeaderWithAclList
6895 LOC: Config.request_header_add
6898 Usage: request_header_add field-name field-value [ acl ... ]
6899 Example: request_header_add X-Client-CA "CA=%ssl::>cert_issuer" all
6901 This option adds header fields to outgoing HTTP requests (i.e.,
6902 request headers sent by Squid to the next HTTP hop such as a
6903 cache peer or an origin server). The option has no effect during
6904 cache hit detection. The equivalent adaptation vectoring point
6905 in ICAP terminology is post-cache REQMOD.
6907 Field-name is a token specifying an HTTP header name. If a
6908 standard HTTP header name is used, Squid does not check whether
6909 the new header conflicts with any existing headers or violates
6910 HTTP rules. If the request to be modified already contains a
6911 field with the same name, the old field is preserved but the
6912 header field values are not merged.
6914 Field-value is either a token or a quoted string. If quoted
6915 string format is used, then the surrounding quotes are removed
6916 while escape sequences and %macros are processed.
6918 One or more Squid ACLs may be specified to restrict header
6919 injection to matching requests. As always in squid.conf, all
6920 ACLs in the ACL list must be satisfied for the insertion to
6921 happen. The request_header_add supports fast ACLs only.
6923 See also: reply_header_add.
6926 NAME: reply_header_add
6927 TYPE: HeaderWithAclList
6928 LOC: Config.reply_header_add
6931 Usage: reply_header_add field-name field-value [ acl ... ]
6932 Example: reply_header_add X-Client-CA "CA=%ssl::>cert_issuer" all
6934 This option adds header fields to outgoing HTTP responses (i.e., response
6935 headers delivered by Squid to the client). This option has no effect on
6936 cache hit detection. The equivalent adaptation vectoring point in
6937 ICAP terminology is post-cache RESPMOD. This option does not apply to
6938 successful CONNECT replies.
6940 Field-name is a token specifying an HTTP header name. If a
6941 standard HTTP header name is used, Squid does not check whether
6942 the new header conflicts with any existing headers or violates
6943 HTTP rules. If the response to be modified already contains a
6944 field with the same name, the old field is preserved but the
6945 header field values are not merged.
6947 Field-value is either a token or a quoted string. If quoted
6948 string format is used, then the surrounding quotes are removed
6949 while escape sequences and %macros are processed.
6951 One or more Squid ACLs may be specified to restrict header
6952 injection to matching responses. As always in squid.conf, all
6953 ACLs in the ACL list must be satisfied for the insertion to
6954 happen. The reply_header_add option supports fast ACLs only.
6956 See also: request_header_add.
6964 This option used to log custom information about the master
6965 transaction. For example, an admin may configure Squid to log
6966 which "user group" the transaction belongs to, where "user group"
6967 will be determined based on a set of ACLs and not [just]
6968 authentication information.
6969 Values of key/value pairs can be logged using %{key}note macros:
6971 note key value acl ...
6972 logformat myFormat ... %{key}note ...
6974 This clause only supports fast acl types.
6975 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6978 NAME: relaxed_header_parser
6979 COMMENT: on|off|warn
6981 LOC: Config.onoff.relaxed_header_parser
6984 In the default "on" setting Squid accepts certain forms
6985 of non-compliant HTTP messages where it is unambiguous
6986 what the sending application intended even if the message
6987 is not correctly formatted. The messages is then normalized
6988 to the correct form when forwarded by Squid.
6990 If set to "warn" then a warning will be emitted in cache.log
6991 each time such HTTP error is encountered.
6993 If set to "off" then such HTTP errors will cause the request
6994 or response to be rejected.
6997 NAME: collapsed_forwarding
7000 LOC: Config.onoff.collapsed_forwarding
7003 This option controls whether Squid is allowed to merge multiple
7004 potentially cachable requests for the same URI before Squid knows
7005 whether the response is going to be cachable.
7007 When enabled, instead of forwarding each concurrent request for
7008 the same URL, Squid just sends the first of them. The other, so
7009 called "collapsed" requests, wait for the response to the first
7010 request and, if it happens to be cachable, use that response.
7011 Here, "concurrent requests" means "received after the first
7012 request headers were parsed and before the corresponding response
7013 headers were parsed".
7015 This feature is disabled by default: enabling collapsed
7016 forwarding needlessly delays forwarding requests that look
7017 cachable (when they are collapsed) but then need to be forwarded
7018 individually anyway because they end up being for uncachable
7019 content. However, in some cases, such as acceleration of highly
7020 cachable content with periodic or grouped expiration times, the
7021 gains from collapsing [large volumes of simultaneous refresh
7022 requests] outweigh losses from such delays.
7024 Squid collapses two kinds of requests: regular client requests
7025 received on one of the listening ports and internal "cache
7026 revalidation" requests which are triggered by those regular
7027 requests hitting a stale cached object. Revalidation collapsing
7028 is currently disabled for Squid instances containing SMP-aware
7029 disk or memory caches and for Vary-controlled cached objects.
7032 NAME: collapsed_forwarding_access
7035 DEFAULT_DOC: Requests may be collapsed if collapsed_forwarding is on.
7036 LOC: Config.accessList.collapsedForwardingAccess
7038 Use this directive to restrict collapsed forwarding to a subset of
7039 eligible requests. The directive is checked for regular HTTP
7040 requests, internal revalidation requests, and HTCP/ICP requests.
7042 collapsed_forwarding_access allow|deny [!]aclname ...
7044 This directive cannot force collapsing. It has no effect on
7045 collapsing unless collapsed_forwarding is 'on', and all other
7046 collapsing preconditions are satisfied.
7048 * A denied request will not collapse, and future transactions will
7049 not collapse on it (even if they are allowed to collapse).
7051 * An allowed request may collapse, or future transactions may
7052 collapse on it (provided they are allowed to collapse).
7054 This directive is evaluated before receiving HTTP response headers
7055 and without access to Squid-to-peer connection (if any).
7057 Only fast ACLs are supported.
7059 See also: collapsed_forwarding.
7062 NAME: shared_transient_entries_limit collapsed_forwarding_shared_entries_limit
7063 COMMENT: (number of entries)
7065 LOC: Config.shared_transient_entries_limit
7068 This directive limits the size of a table used for sharing current
7069 transaction information among SMP workers. A table entry stores meta
7070 information about a single cache entry being delivered to Squid
7071 client(s) by one or more SMP workers. A single table entry consumes
7072 less than 128 shared memory bytes.
7074 The limit should be significantly larger than the number of
7075 concurrent non-collapsed cachable responses leaving Squid. For a
7076 cache that handles less than 5000 concurrent requests, the default
7077 setting of 16384 should be plenty.
7079 Using excessively large values wastes shared memory. Limiting the
7080 table size too much results in hash collisions, leading to lower hit
7081 ratio and missed SMP request collapsing opportunities: Transactions
7082 left without a table entry cannot cache their responses and are
7083 invisible to other concurrent requests for the same resource.
7085 A zero limit is allowed but unsupported. A positive small limit
7086 lowers hit ratio, but zero limit disables a lot of essential
7087 synchronization among SMP workers, leading to HTTP violations (e.g.,
7088 stale hit responses). It also disables shared collapsed forwarding:
7089 A worker becomes unable to collapse its requests on transactions in
7090 other workers, resulting in more trips to the origin server and more
7096 -----------------------------------------------------------------------------
7099 NAME: forward_timeout
7102 LOC: Config.Timeout.forward
7105 This parameter specifies how long Squid should at most attempt in
7106 finding a forwarding path for the request before giving up.
7109 NAME: connect_timeout
7112 LOC: Config.Timeout.connect
7115 This parameter specifies how long to wait for the TCP connect to
7116 the requested server or peer to complete before Squid should
7117 attempt to find another path where to forward the request.
7120 NAME: peer_connect_timeout
7123 LOC: Config.Timeout.peer_connect
7126 This parameter specifies how long to wait for a pending TCP
7127 connection to a peer cache. The default is 30 seconds. You
7128 may also set different timeout values for individual neighbors
7129 with the 'connect-timeout' option on a 'cache_peer' line.
7135 LOC: Config.Timeout.read
7138 Applied on peer server connections.
7140 After each successful read(), the timeout will be extended by this
7141 amount. If no data is read again after this amount of time,
7142 the request is aborted and logged with ERR_READ_TIMEOUT.
7144 The default is 15 minutes.
7150 LOC: Config.Timeout.write
7153 This timeout is tracked for all connections that have data
7154 available for writing and are waiting for the socket to become
7155 ready. After each successful write, the timeout is extended by
7156 the configured amount. If Squid has data to write but the
7157 connection is not ready for the configured duration, the
7158 transaction associated with the connection is terminated. The
7159 default is 15 minutes.
7162 NAME: request_timeout
7164 LOC: Config.Timeout.request
7167 How long to wait for complete HTTP request headers after initial
7168 connection establishment.
7171 NAME: request_start_timeout
7173 LOC: Config.Timeout.request_start_timeout
7176 How long to wait for the first request byte after initial
7177 connection establishment.
7180 NAME: client_idle_pconn_timeout persistent_request_timeout
7182 LOC: Config.Timeout.clientIdlePconn
7185 How long to wait for the next HTTP request on a persistent
7186 client connection after the previous request completes.
7189 NAME: ftp_client_idle_timeout
7191 LOC: Config.Timeout.ftpClientIdle
7194 How long to wait for an FTP request on a connection to Squid ftp_port.
7195 Many FTP clients do not deal with idle connection closures well,
7196 necessitating a longer default timeout than client_idle_pconn_timeout
7197 used for incoming HTTP requests.
7200 NAME: client_lifetime
7203 LOC: Config.Timeout.lifetime
7206 The maximum amount of time a client (browser) is allowed to
7207 remain connected to the cache process. This protects the Cache
7208 from having a lot of sockets (and hence file descriptors) tied up
7209 in a CLOSE_WAIT state from remote clients that go away without
7210 properly shutting down (either because of a network failure or
7211 because of a poor client implementation). The default is one
7214 NOTE: The default value is intended to be much larger than any
7215 client would ever need to be connected to your cache. You
7216 should probably change client_lifetime only as a last resort.
7217 If you seem to have many client connections tying up
7218 filedescriptors, we recommend first tuning the read_timeout,
7219 request_timeout, persistent_request_timeout and quick_abort values.
7222 NAME: pconn_lifetime
7225 LOC: Config.Timeout.pconnLifetime
7228 Desired maximum lifetime of a persistent connection.
7229 When set, Squid will close a now-idle persistent connection that
7230 exceeded configured lifetime instead of moving the connection into
7231 the idle connection pool (or equivalent). No effect on ongoing/active
7232 transactions. Connection lifetime is the time period from the
7233 connection acceptance or opening time until "now".
7235 This limit is useful in environments with long-lived connections
7236 where Squid configuration or environmental factors change during a
7237 single connection lifetime. If unrestricted, some connections may
7238 last for hours and even days, ignoring those changes that should
7239 have affected their behavior or their existence.
7241 Currently, a new lifetime value supplied via Squid reconfiguration
7242 has no effect on already idle connections unless they become busy.
7244 When set to '0' this limit is not used.
7247 NAME: half_closed_clients
7249 LOC: Config.onoff.half_closed_clients
7252 Some clients may shutdown the sending side of their TCP
7253 connections, while leaving their receiving sides open. Sometimes,
7254 Squid can not tell the difference between a half-closed and a
7255 fully-closed TCP connection.
7257 By default, Squid will immediately close client connections when
7258 read(2) returns "no more data to read."
7260 Change this option to 'on' and Squid will keep open connections
7261 until a read(2) or write(2) on the socket returns an error.
7262 This may show some benefits for reverse proxies. But if not
7263 it is recommended to leave OFF.
7266 NAME: server_idle_pconn_timeout pconn_timeout
7268 LOC: Config.Timeout.serverIdlePconn
7271 Timeout for idle persistent connections to servers and other
7278 LOC: Ident::TheConfig.timeout
7281 Maximum time to wait for IDENT lookups to complete.
7283 If this is too high, and you enabled IDENT lookups from untrusted
7284 users, you might be susceptible to denial-of-service by having
7285 many ident requests going at once.
7288 NAME: shutdown_lifetime
7291 LOC: Config.shutdownLifetime
7294 When SIGTERM or SIGHUP is received, the cache is put into
7295 "shutdown pending" mode until all active sockets are closed.
7296 This value is the lifetime to set for all open descriptors
7297 during shutdown mode. Any active clients after this many
7298 seconds will receive a 'timeout' message.
7302 ADMINISTRATIVE PARAMETERS
7303 -----------------------------------------------------------------------------
7309 LOC: Config.adminEmail
7311 Email-address of local cache manager who will receive
7312 mail if the cache dies. The default is "webmaster".
7318 LOC: Config.EmailFrom
7320 From: email-address for mail sent when the cache dies.
7321 The default is to use 'squid@unique_hostname'.
7323 See also: unique_hostname directive.
7329 LOC: Config.EmailProgram
7331 Email program used to send mail if the cache dies.
7332 The default is "mail". The specified program must comply
7333 with the standard Unix mail syntax:
7334 mail-program recipient < mailfile
7336 Optional command line options can be specified.
7339 NAME: cache_effective_user
7341 DEFAULT: @DEFAULT_CACHE_EFFECTIVE_USER@
7342 LOC: Config.effectiveUser
7344 If you start Squid as root, it will change its effective/real
7345 UID/GID to the user specified below. The default is to change
7346 to UID of @DEFAULT_CACHE_EFFECTIVE_USER@.
7347 see also; cache_effective_group
7350 NAME: cache_effective_group
7353 DEFAULT_DOC: Use system group memberships of the cache_effective_user account
7354 LOC: Config.effectiveGroup
7356 Squid sets the GID to the effective user's default group ID
7357 (taken from the password file) and supplementary group list
7358 from the groups membership.
7360 If you want Squid to run with a specific GID regardless of
7361 the group memberships of the effective user then set this
7362 to the group (or GID) you want Squid to run as. When set
7363 all other group privileges of the effective user are ignored
7364 and only this GID is effective. If Squid is not started as
7365 root the user starting Squid MUST be member of the specified
7368 This option is not recommended by the Squid Team.
7369 Our preference is for administrators to configure a secure
7370 user account for squid with UID/GID matching system policies.
7373 NAME: httpd_suppress_version_string
7377 LOC: Config.onoff.httpd_suppress_version_string
7379 Suppress Squid version string info in HTTP headers and HTML error pages.
7382 NAME: visible_hostname
7384 LOC: Config.visibleHostname
7386 DEFAULT_DOC: Automatically detect the system host name
7388 If you want to present a special hostname in error messages, etc,
7389 define this. Otherwise, the return value of gethostname()
7390 will be used. If you have multiple caches in a cluster and
7391 get errors about IP-forwarding you must set them to have individual
7392 names with this setting.
7395 NAME: unique_hostname
7397 LOC: Config.uniqueHostname
7399 DEFAULT_DOC: Copy the value from visible_hostname
7401 If you want to have multiple machines with the same
7402 'visible_hostname' you must give each machine a different
7403 'unique_hostname' so forwarding loops can be detected.
7406 NAME: hostname_aliases
7408 LOC: Config.hostnameAliases
7411 A list of other DNS names your cache has.
7419 Minimum umask which should be enforced while the proxy
7420 is running, in addition to the umask set at startup.
7422 For a traditional octal representation of umasks, start
7427 OPTIONS FOR THE CACHE REGISTRATION SERVICE
7428 -----------------------------------------------------------------------------
7430 This section contains parameters for the (optional) cache
7431 announcement service. This service is provided to help
7432 cache administrators locate one another in order to join or
7433 create cache hierarchies.
7435 An 'announcement' message is sent (via UDP) to the registration
7436 service by Squid. By default, the announcement message is NOT
7437 SENT unless you enable it with 'announce_period' below.
7439 The announcement message includes your hostname, plus the
7440 following information from this configuration file:
7446 All current information is processed regularly and made
7447 available on the Web at http://www.ircache.net/Cache/Tracker/.
7450 NAME: announce_period
7452 LOC: Config.Announce.period
7454 DEFAULT_DOC: Announcement messages disabled.
7456 This is how frequently to send cache announcements.
7458 To enable announcing your cache, just set an announce period.
7461 announce_period 1 day
7466 DEFAULT: tracker.ircache.net
7467 LOC: Config.Announce.host
7469 Set the hostname where announce registration messages will be sent.
7471 See also announce_port and announce_file
7477 LOC: Config.Announce.file
7479 The contents of this file will be included in the announce
7480 registration messages.
7486 LOC: Config.Announce.port
7488 Set the port where announce registration messages will be sent.
7490 See also announce_host and announce_file
7494 HTTPD-ACCELERATOR OPTIONS
7495 -----------------------------------------------------------------------------
7498 NAME: httpd_accel_surrogate_id
7501 DEFAULT_DOC: visible_hostname is used if no specific ID is set.
7502 LOC: Config.Accel.surrogate_id
7504 Surrogates (http://www.esi.org/architecture_spec_1.0.html)
7505 need an identification token to allow control targeting. Because
7506 a farm of surrogates may all perform the same tasks, they may share
7507 an identification token.
7509 When the surrogate is a reverse-proxy, this ID is also
7510 used as cdn-id for CDN-Loop detection (RFC 8586).
7513 NAME: http_accel_surrogate_remote
7517 LOC: Config.onoff.surrogate_is_remote
7519 Remote surrogates (such as those in a CDN) honour the header
7520 "Surrogate-Control: no-store-remote".
7522 Set this to on to have squid behave as a remote surrogate.
7526 IFDEF: USE_SQUID_ESI
7527 COMMENT: libxml2|expat
7529 LOC: ESIParser::Type
7531 DEFAULT_DOC: Selects libxml2 if available at ./configure time or libexpat otherwise.
7533 Selects the XML parsing library to use when interpreting responses with
7536 To disable ESI handling completely, ./configure Squid with --disable-esi.
7540 DELAY POOL PARAMETERS
7541 -----------------------------------------------------------------------------
7545 TYPE: delay_pool_count
7547 IFDEF: USE_DELAY_POOLS
7550 This represents the number of delay pools to be used. For example,
7551 if you have one class 2 delay pool and one class 3 delays pool, you
7552 have a total of 2 delay pools.
7554 See also delay_parameters, delay_class, delay_access for pool
7555 configuration details.
7559 TYPE: delay_pool_class
7561 IFDEF: USE_DELAY_POOLS
7564 This defines the class of each delay pool. There must be exactly one
7565 delay_class line for each delay pool. For example, to define two
7566 delay pools, one of class 2 and one of class 3, the settings above
7570 delay_pools 4 # 4 delay pools
7571 delay_class 1 2 # pool 1 is a class 2 pool
7572 delay_class 2 3 # pool 2 is a class 3 pool
7573 delay_class 3 4 # pool 3 is a class 4 pool
7574 delay_class 4 5 # pool 4 is a class 5 pool
7576 The delay pool classes are:
7578 class 1 Everything is limited by a single aggregate
7581 class 2 Everything is limited by a single aggregate
7582 bucket as well as an "individual" bucket chosen
7583 from bits 25 through 32 of the IPv4 address.
7585 class 3 Everything is limited by a single aggregate
7586 bucket as well as a "network" bucket chosen
7587 from bits 17 through 24 of the IP address and a
7588 "individual" bucket chosen from bits 17 through
7589 32 of the IPv4 address.
7591 class 4 Everything in a class 3 delay pool, with an
7592 additional limit on a per user basis. This
7593 only takes effect if the username is established
7594 in advance - by forcing authentication in your
7597 class 5 Requests are grouped according their tag (see
7598 external_acl's tag= reply).
7601 Each pool also requires a delay_parameters directive to configure the pool size
7602 and speed limits used whenever the pool is applied to a request. Along with
7603 a set of delay_access directives to determine when it is used.
7605 NOTE: If an IP address is a.b.c.d
7606 -> bits 25 through 32 are "d"
7607 -> bits 17 through 24 are "c"
7608 -> bits 17 through 32 are "c * 256 + d"
7610 NOTE-2: Due to the use of bitmasks in class 2,3,4 pools they only apply to
7611 IPv4 traffic. Class 1 and 5 pools may be used with IPv6 traffic.
7613 This clause only supports fast acl types.
7614 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
7616 See also delay_parameters and delay_access.
7620 TYPE: delay_pool_access
7622 DEFAULT_DOC: Deny using the pool, unless allow rules exist in squid.conf for the pool.
7623 IFDEF: USE_DELAY_POOLS
7626 This is used to determine which delay pool a request falls into.
7628 delay_access is sorted per pool and the matching starts with pool 1,
7629 then pool 2, ..., and finally pool N. The first delay pool where the
7630 request is allowed is selected for the request. If it does not allow
7631 the request to any pool then the request is not delayed (default).
7633 For example, if you want some_big_clients in delay
7634 pool 1 and lotsa_little_clients in delay pool 2:
7636 delay_access 1 allow some_big_clients
7637 delay_access 1 deny all
7638 delay_access 2 allow lotsa_little_clients
7639 delay_access 2 deny all
7640 delay_access 3 allow authenticated_clients
7642 See also delay_parameters and delay_class.
7646 NAME: delay_parameters
7647 TYPE: delay_pool_rates
7649 IFDEF: USE_DELAY_POOLS
7652 This defines the parameters for a delay pool. Each delay pool has
7653 a number of "buckets" associated with it, as explained in the
7654 description of delay_class.
7656 For a class 1 delay pool, the syntax is:
7658 delay_parameters pool aggregate
7660 For a class 2 delay pool:
7662 delay_parameters pool aggregate individual
7664 For a class 3 delay pool:
7666 delay_parameters pool aggregate network individual
7668 For a class 4 delay pool:
7670 delay_parameters pool aggregate network individual user
7672 For a class 5 delay pool:
7674 delay_parameters pool tagrate
7676 The option variables are:
7678 pool a pool number - ie, a number between 1 and the
7679 number specified in delay_pools as used in
7682 aggregate the speed limit parameters for the aggregate bucket
7685 individual the speed limit parameters for the individual
7686 buckets (class 2, 3).
7688 network the speed limit parameters for the network buckets
7691 user the speed limit parameters for the user buckets
7694 tagrate the speed limit parameters for the tag buckets
7697 A pair of delay parameters is written restore/maximum, where restore is
7698 the number of bytes (not bits - modem and network speeds are usually
7699 quoted in bits) per second placed into the bucket, and maximum is the
7700 maximum number of bytes which can be in the bucket at any time.
7702 There must be one delay_parameters line for each delay pool.
7705 For example, if delay pool number 1 is a class 2 delay pool as in the
7706 above example, and is being used to strictly limit each host to 64Kbit/sec
7707 (plus overheads), with no overall limit, the line is:
7709 delay_parameters 1 none 8000/8000
7711 Note that 8 x 8K Byte/sec -> 64K bit/sec.
7713 Note that the word 'none' is used to represent no limit.
7716 And, if delay pool number 2 is a class 3 delay pool as in the above
7717 example, and you want to limit it to a total of 256Kbit/sec (strict limit)
7718 with each 8-bit network permitted 64Kbit/sec (strict limit) and each
7719 individual host permitted 4800bit/sec with a bucket maximum size of 64Kbits
7720 to permit a decent web page to be downloaded at a decent speed
7721 (if the network is not being limited due to overuse) but slow down
7722 large downloads more significantly:
7724 delay_parameters 2 32000/32000 8000/8000 600/8000
7726 Note that 8 x 32K Byte/sec -> 256K bit/sec.
7727 8 x 8K Byte/sec -> 64K bit/sec.
7728 8 x 600 Byte/sec -> 4800 bit/sec.
7731 Finally, for a class 4 delay pool as in the example - each user will
7732 be limited to 128Kbits/sec no matter how many workstations they are logged into.:
7734 delay_parameters 4 32000/32000 8000/8000 600/64000 16000/16000
7737 See also delay_class and delay_access.
7741 NAME: delay_initial_bucket_level
7742 COMMENT: (percent, 0-100)
7745 IFDEF: USE_DELAY_POOLS
7746 LOC: Config.Delay.initial
7748 The initial bucket percentage is used to determine how much is put
7749 in each bucket when squid starts, is reconfigured, or first notices
7750 a host accessing it (in class 2 and class 3, individual hosts and
7751 networks only have buckets associated with them once they have been
7756 CLIENT DELAY POOL PARAMETERS
7757 -----------------------------------------------------------------------------
7760 NAME: client_delay_pools
7761 TYPE: client_delay_pool_count
7763 IFDEF: USE_DELAY_POOLS
7764 LOC: Config.ClientDelay
7766 This option specifies the number of client delay pools used. It must
7767 preceed other client_delay_* options.
7770 client_delay_pools 2
7772 See also client_delay_parameters and client_delay_access.
7775 NAME: client_delay_initial_bucket_level
7776 COMMENT: (percent, 0-no_limit)
7779 IFDEF: USE_DELAY_POOLS
7780 LOC: Config.ClientDelay.initial
7782 This option determines the initial bucket size as a percentage of
7783 max_bucket_size from client_delay_parameters. Buckets are created
7784 at the time of the "first" connection from the matching IP. Idle
7785 buckets are periodically deleted up.
7787 You can specify more than 100 percent but note that such "oversized"
7788 buckets are not refilled until their size goes down to max_bucket_size
7789 from client_delay_parameters.
7792 client_delay_initial_bucket_level 50
7795 NAME: client_delay_parameters
7796 TYPE: client_delay_pool_rates
7798 IFDEF: USE_DELAY_POOLS
7799 LOC: Config.ClientDelay
7802 This option configures client-side bandwidth limits using the
7805 client_delay_parameters pool speed_limit max_bucket_size
7807 pool is an integer ID used for client_delay_access matching.
7809 speed_limit is bytes added to the bucket per second.
7811 max_bucket_size is the maximum size of a bucket, enforced after any
7812 speed_limit additions.
7814 Please see the delay_parameters option for more information and
7818 client_delay_parameters 1 1024 2048
7819 client_delay_parameters 2 51200 16384
7821 See also client_delay_access.
7825 NAME: client_delay_access
7826 TYPE: client_delay_pool_access
7828 DEFAULT_DOC: Deny use of the pool, unless allow rules exist in squid.conf for the pool.
7829 IFDEF: USE_DELAY_POOLS
7830 LOC: Config.ClientDelay
7832 This option determines the client-side delay pool for the
7835 client_delay_access pool_ID allow|deny acl_name
7837 All client_delay_access options are checked in their pool ID
7838 order, starting with pool 1. The first checked pool with allowed
7839 request is selected for the request. If no ACL matches or there
7840 are no client_delay_access options, the request bandwidth is not
7843 The ACL-selected pool is then used to find the
7844 client_delay_parameters for the request. Client-side pools are
7845 not used to aggregate clients. Clients are always aggregated
7846 based on their source IP addresses (one bucket per source IP).
7848 This clause only supports fast acl types.
7849 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
7850 Additionally, only the client TCP connection details are available.
7851 ACLs testing HTTP properties will not work.
7853 Please see delay_access for more examples.
7856 client_delay_access 1 allow low_rate_network
7857 client_delay_access 2 allow vips_network
7860 See also client_delay_parameters and client_delay_pools.
7863 NAME: response_delay_pool
7864 TYPE: response_delay_pool_parameters
7866 IFDEF: USE_DELAY_POOLS
7867 LOC: Config.MessageDelay
7869 This option configures client response bandwidth limits using the
7872 response_delay_pool name [option=value] ...
7874 name the response delay pool name
7878 individual-restore The speed limit of an individual
7879 bucket(bytes/s). To be used in conjunction
7880 with 'individual-maximum'.
7882 individual-maximum The maximum number of bytes which can
7883 be placed into the individual bucket. To be used
7884 in conjunction with 'individual-restore'.
7886 aggregate-restore The speed limit for the aggregate
7887 bucket(bytes/s). To be used in conjunction with
7888 'aggregate-maximum'.
7890 aggregate-maximum The maximum number of bytes which can
7891 be placed into the aggregate bucket. To be used
7892 in conjunction with 'aggregate-restore'.
7894 initial-bucket-level The initial bucket size as a percentage
7895 of individual-maximum.
7897 Individual and(or) aggregate bucket options may not be specified,
7898 meaning no individual and(or) aggregate speed limitation.
7899 See also response_delay_pool_access and delay_parameters for
7900 terminology details.
7903 NAME: response_delay_pool_access
7904 TYPE: response_delay_pool_access
7906 DEFAULT_DOC: Deny use of the pool, unless allow rules exist in squid.conf for the pool.
7907 IFDEF: USE_DELAY_POOLS
7908 LOC: Config.MessageDelay
7910 Determines whether a specific named response delay pool is used
7911 for the transaction. The syntax for this directive is:
7913 response_delay_pool_access pool_name allow|deny acl_name
7915 All response_delay_pool_access options are checked in the order
7916 they appear in this configuration file. The first rule with a
7917 matching ACL wins. If (and only if) an "allow" rule won, Squid
7918 assigns the response to the corresponding named delay pool.
7922 WCCPv1 AND WCCPv2 CONFIGURATION OPTIONS
7923 -----------------------------------------------------------------------------
7928 LOC: Config.Wccp.router
7930 DEFAULT_DOC: WCCP disabled.
7933 Use this option to define your WCCP ``home'' router for
7936 wccp_router supports a single WCCP(v1) router
7938 wccp2_router supports multiple WCCPv2 routers
7940 only one of the two may be used at the same time and defines
7941 which version of WCCP to use.
7945 TYPE: IpAddress_list
7946 LOC: Config.Wccp2.router
7948 DEFAULT_DOC: WCCPv2 disabled.
7951 Use this option to define your WCCP ``home'' router for
7954 wccp_router supports a single WCCP(v1) router
7956 wccp2_router supports multiple WCCPv2 routers
7958 only one of the two may be used at the same time and defines
7959 which version of WCCP to use.
7964 LOC: Config.Wccp.version
7968 This directive is only relevant if you need to set up WCCP(v1)
7969 to some very old and end-of-life Cisco routers. In all other
7970 setups it must be left unset or at the default setting.
7971 It defines an internal version in the WCCP(v1) protocol,
7972 with version 4 being the officially documented protocol.
7974 According to some users, Cisco IOS 11.2 and earlier only
7975 support WCCP version 3. If you're using that or an earlier
7976 version of IOS, you may need to change this value to 3, otherwise
7977 do not specify this parameter.
7980 NAME: wccp2_rebuild_wait
7982 LOC: Config.Wccp2.rebuildwait
7986 If this is enabled Squid will wait for the cache dir rebuild to finish
7987 before sending the first wccp2 HereIAm packet
7990 NAME: wccp2_forwarding_method
7992 LOC: Config.Wccp2.forwarding_method
7996 WCCP2 allows the setting of forwarding methods between the
7997 router/switch and the cache. Valid values are as follows:
7999 gre - GRE encapsulation (forward the packet in a GRE/WCCP tunnel)
8000 l2 - L2 redirect (forward the packet using Layer 2/MAC rewriting)
8002 Currently (as of IOS 12.4) cisco routers only support GRE.
8003 Cisco switches only support the L2 redirect assignment method.
8006 NAME: wccp2_return_method
8008 LOC: Config.Wccp2.return_method
8012 WCCP2 allows the setting of return methods between the
8013 router/switch and the cache for packets that the cache
8014 decides not to handle. Valid values are as follows:
8016 gre - GRE encapsulation (forward the packet in a GRE/WCCP tunnel)
8017 l2 - L2 redirect (forward the packet using Layer 2/MAC rewriting)
8019 Currently (as of IOS 12.4) cisco routers only support GRE.
8020 Cisco switches only support the L2 redirect assignment.
8022 If the "ip wccp redirect exclude in" command has been
8023 enabled on the cache interface, then it is still safe for
8024 the proxy server to use a l2 redirect method even if this
8025 option is set to GRE.
8028 NAME: wccp2_assignment_method
8030 LOC: Config.Wccp2.assignment_method
8034 WCCP2 allows the setting of methods to assign the WCCP hash
8035 Valid values are as follows:
8037 hash - Hash assignment
8038 mask - Mask assignment
8040 As a general rule, cisco routers support the hash assignment method
8041 and cisco switches support the mask assignment method.
8046 LOC: Config.Wccp2.info
8047 DEFAULT_IF_NONE: standard 0
8048 DEFAULT_DOC: Use the 'web-cache' standard service.
8051 WCCP2 allows for multiple traffic services. There are two
8052 types: "standard" and "dynamic". The standard type defines
8053 one service id - http (id 0). The dynamic service ids can be from
8054 51 to 255 inclusive. In order to use a dynamic service id
8055 one must define the type of traffic to be redirected; this is done
8056 using the wccp2_service_info option.
8058 The "standard" type does not require a wccp2_service_info option,
8059 just specifying the service id will suffice.
8061 MD5 service authentication can be enabled by adding
8062 "password=<password>" to the end of this service declaration.
8066 wccp2_service standard 0 # for the 'web-cache' standard service
8067 wccp2_service dynamic 80 # a dynamic service type which will be
8068 # fleshed out with subsequent options.
8069 wccp2_service standard 0 password=foo
8072 NAME: wccp2_service_info
8073 TYPE: wccp2_service_info
8074 LOC: Config.Wccp2.info
8078 Dynamic WCCPv2 services require further information to define the
8079 traffic you wish to have diverted.
8083 wccp2_service_info <id> protocol=<protocol> flags=<flag>,<flag>..
8084 priority=<priority> ports=<port>,<port>..
8086 The relevant WCCPv2 flags:
8087 + src_ip_hash, dst_ip_hash
8088 + source_port_hash, dst_port_hash
8089 + src_ip_alt_hash, dst_ip_alt_hash
8090 + src_port_alt_hash, dst_port_alt_hash
8093 The port list can be one to eight entries.
8097 wccp2_service_info 80 protocol=tcp flags=src_ip_hash,ports_source
8098 priority=240 ports=80
8100 Note: the service id must have been defined by a previous
8101 'wccp2_service dynamic <id>' entry.
8106 LOC: Config.Wccp2.weight
8110 Each cache server gets assigned a set of the destination
8111 hash proportional to their weight.
8116 LOC: Config.Wccp.address
8118 DEFAULT_DOC: Address selected by the operating system.
8121 Use this option if you require WCCP(v1) to use a specific
8124 The default behavior is to not bind to any specific address.
8129 LOC: Config.Wccp2.address
8131 DEFAULT_DOC: Address selected by the operating system.
8134 Use this option if you require WCCPv2 to use a specific
8137 The default behavior is to not bind to any specific address.
8141 PERSISTENT CONNECTION HANDLING
8142 -----------------------------------------------------------------------------
8144 Also see "pconn_timeout" in the TIMEOUTS section
8147 NAME: client_persistent_connections
8149 LOC: Config.onoff.client_pconns
8152 Persistent connection support for clients.
8153 Squid uses persistent connections (when allowed). You can use
8154 this option to disable persistent connections with clients.
8157 NAME: server_persistent_connections
8159 LOC: Config.onoff.server_pconns
8162 Persistent connection support for servers.
8163 Squid uses persistent connections (when allowed). You can use
8164 this option to disable persistent connections with servers.
8167 NAME: persistent_connection_after_error
8169 LOC: Config.onoff.error_pconns
8172 With this directive the use of persistent connections after
8173 HTTP errors can be disabled. Useful if you have clients
8174 who fail to handle errors on persistent connections proper.
8177 NAME: detect_broken_pconn
8179 LOC: Config.onoff.detect_broken_server_pconns
8182 Some servers have been found to incorrectly signal the use
8183 of HTTP/1.0 persistent connections even on replies not
8184 compatible, causing significant delays. This server problem
8185 has mostly been seen on redirects.
8187 By enabling this directive Squid attempts to detect such
8188 broken replies and automatically assume the reply is finished
8189 after 10 seconds timeout.
8193 CACHE DIGEST OPTIONS
8194 -----------------------------------------------------------------------------
8197 NAME: digest_generation
8198 IFDEF: USE_CACHE_DIGESTS
8200 LOC: Config.onoff.digest_generation
8203 This controls whether the server will generate a Cache Digest
8204 of its contents. By default, Cache Digest generation is
8205 enabled if Squid is compiled with --enable-cache-digests defined.
8208 NAME: digest_bits_per_entry
8209 IFDEF: USE_CACHE_DIGESTS
8211 LOC: Config.digest.bits_per_entry
8214 This is the number of bits of the server's Cache Digest which
8215 will be associated with the Digest entry for a given HTTP
8216 Method and URL (public key) combination. The default is 5.
8219 NAME: digest_rebuild_period
8220 IFDEF: USE_CACHE_DIGESTS
8223 LOC: Config.digest.rebuild_period
8226 This is the wait time between Cache Digest rebuilds.
8229 NAME: digest_rewrite_period
8231 IFDEF: USE_CACHE_DIGESTS
8233 LOC: Config.digest.rewrite_period
8236 This is the wait time between Cache Digest writes to
8240 NAME: digest_swapout_chunk_size
8243 IFDEF: USE_CACHE_DIGESTS
8244 LOC: Config.digest.swapout_chunk_size
8247 This is the number of bytes of the Cache Digest to write to
8248 disk at a time. It defaults to 4096 bytes (4KB), the Squid
8252 NAME: digest_rebuild_chunk_percentage
8253 COMMENT: (percent, 0-100)
8254 IFDEF: USE_CACHE_DIGESTS
8256 LOC: Config.digest.rebuild_chunk_percentage
8259 This is the percentage of the Cache Digest to be scanned at a
8260 time. By default it is set to 10% of the Cache Digest.
8265 -----------------------------------------------------------------------------
8270 LOC: Config.Port.snmp
8272 DEFAULT_DOC: SNMP disabled.
8275 The port number where Squid listens for SNMP requests. To enable
8276 SNMP support set this to a suitable port number. Port number
8277 3401 is often used for the Squid SNMP agent. By default it's
8278 set to "0" (disabled)
8286 LOC: Config.accessList.snmp
8288 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
8291 Allowing or denying access to the SNMP port.
8293 All access to the agent is denied by default.
8296 snmp_access allow|deny [!]aclname ...
8298 This clause only supports fast acl types.
8299 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
8302 snmp_access allow snmppublic localhost
8303 snmp_access deny all
8306 NAME: snmp_incoming_address
8308 LOC: Config.Addrs.snmp_incoming
8310 DEFAULT_DOC: Accept SNMP packets from all machine interfaces.
8313 Just like 'udp_incoming_address', but for the SNMP port.
8315 snmp_incoming_address is used for the SNMP socket receiving
8316 messages from SNMP agents.
8318 The default snmp_incoming_address is to listen on all
8319 available network interfaces.
8322 NAME: snmp_outgoing_address
8324 LOC: Config.Addrs.snmp_outgoing
8326 DEFAULT_DOC: Use snmp_incoming_address or an address selected by the operating system.
8329 Just like 'udp_outgoing_address', but for the SNMP port.
8331 snmp_outgoing_address is used for SNMP packets returned to SNMP
8334 If snmp_outgoing_address is not set it will use the same socket
8335 as snmp_incoming_address. Only change this if you want to have
8336 SNMP replies sent using another address than where this Squid
8337 listens for SNMP queries.
8339 NOTE, snmp_incoming_address and snmp_outgoing_address can not have
8340 the same value since they both use the same port.
8345 -----------------------------------------------------------------------------
8348 NAME: icp_port udp_port
8351 DEFAULT_DOC: ICP disabled.
8352 LOC: Config.Port.icp
8354 The port number where Squid sends and receives ICP queries to
8355 and from neighbor caches. The standard UDP port for ICP is 3130.
8358 icp_port @DEFAULT_ICP_PORT@
8365 DEFAULT_DOC: HTCP disabled.
8366 LOC: Config.Port.htcp
8368 The port number where Squid sends and receives HTCP queries to
8369 and from neighbor caches. To turn it on you want to set it to
8376 NAME: log_icp_queries
8380 LOC: Config.onoff.log_udp
8382 If set, ICP queries are logged to access.log. You may wish
8383 do disable this if your ICP load is VERY high to speed things
8384 up or to simplify log analysis.
8387 NAME: udp_incoming_address
8389 LOC:Config.Addrs.udp_incoming
8391 DEFAULT_DOC: Accept packets from all machine interfaces.
8393 udp_incoming_address is used for UDP packets received from other
8396 The default behavior is to not bind to any specific address.
8398 Only change this if you want to have all UDP queries received on
8399 a specific interface/address.
8401 NOTE: udp_incoming_address is used by the ICP, HTCP, and DNS
8402 modules. Altering it will affect all of them in the same manner.
8404 see also; udp_outgoing_address
8406 NOTE, udp_incoming_address and udp_outgoing_address can not
8407 have the same value since they both use the same port.
8410 NAME: udp_outgoing_address
8412 LOC: Config.Addrs.udp_outgoing
8414 DEFAULT_DOC: Use udp_incoming_address or an address selected by the operating system.
8416 udp_outgoing_address is used for UDP packets sent out to other
8419 The default behavior is to not bind to any specific address.
8421 Instead it will use the same socket as udp_incoming_address.
8422 Only change this if you want to have UDP queries sent using another
8423 address than where this Squid listens for UDP queries from other
8426 NOTE: udp_outgoing_address is used by the ICP, HTCP, and DNS
8427 modules. Altering it will affect all of them in the same manner.
8429 see also; udp_incoming_address
8431 NOTE, udp_incoming_address and udp_outgoing_address can not
8432 have the same value since they both use the same port.
8439 LOC: Config.onoff.icp_hit_stale
8441 If you want to return ICP_HIT for stale cache objects, set this
8442 option to 'on'. If you have sibling relationships with caches
8443 in other administrative domains, this should be 'off'. If you only
8444 have sibling relationships with caches under your control,
8445 it is probably okay to set this to 'on'.
8446 If set to 'on', your siblings should use the option "allow-miss"
8447 on their cache_peer lines for connecting to you.
8450 NAME: minimum_direct_hops
8453 LOC: Config.minDirectHops
8455 If using the ICMP pinging stuff, do direct fetches for sites
8456 which are no more than this many hops away.
8459 NAME: minimum_direct_rtt
8463 LOC: Config.minDirectRtt
8465 If using the ICMP pinging stuff, do direct fetches for sites
8466 which are no more than this many rtt milliseconds away.
8472 LOC: Config.Netdb.low
8474 The low water mark for the ICMP measurement database.
8476 Note: high watermark controlled by netdb_high directive.
8478 These watermarks are counts, not percents. The defaults are
8479 (low) 900 and (high) 1000. When the high water mark is
8480 reached, database entries will be deleted until the low
8487 LOC: Config.Netdb.high
8489 The high water mark for the ICMP measurement database.
8491 Note: low watermark controlled by netdb_low directive.
8493 These watermarks are counts, not percents. The defaults are
8494 (low) 900 and (high) 1000. When the high water mark is
8495 reached, database entries will be deleted until the low
8499 NAME: netdb_ping_period
8501 LOC: Config.Netdb.period
8504 The minimum period for measuring a site. There will be at
8505 least this much delay between successive pings to the same
8506 network. The default is five minutes.
8513 LOC: Config.onoff.query_icmp
8515 If you want to ask your peers to include ICMP data in their ICP
8516 replies, enable this option.
8518 If your peer has configured Squid (during compilation) with
8519 '--enable-icmp' that peer will send ICMP pings to origin server
8520 sites of the URLs it receives. If you enable this option the
8521 ICP replies from that peer will include the ICMP data (if available).
8522 Then, when choosing a parent cache, Squid will choose the parent with
8523 the minimal RTT to the origin server. When this happens, the
8524 hierarchy field of the access.log will be
8525 "CLOSEST_PARENT_MISS". This option is off by default.
8528 NAME: test_reachability
8532 LOC: Config.onoff.test_reachability
8534 When this is 'on', ICP MISS replies will be ICP_MISS_NOFETCH
8535 instead of ICP_MISS if the target host is NOT in the ICMP
8536 database, or has a zero RTT.
8539 NAME: icp_query_timeout
8542 DEFAULT_DOC: Dynamic detection.
8544 LOC: Config.Timeout.icp_query
8546 Normally Squid will automatically determine an optimal ICP
8547 query timeout value based on the round-trip-time of recent ICP
8548 queries. If you want to override the value determined by
8549 Squid, set this 'icp_query_timeout' to a non-zero value. This
8550 value is specified in MILLISECONDS, so, to use a 2-second
8551 timeout (the old default), you would write:
8553 icp_query_timeout 2000
8556 NAME: maximum_icp_query_timeout
8560 LOC: Config.Timeout.icp_query_max
8562 Normally the ICP query timeout is determined dynamically. But
8563 sometimes it can lead to very large values (say 5 seconds).
8564 Use this option to put an upper limit on the dynamic timeout
8565 value. Do NOT use this option to always use a fixed (instead
8566 of a dynamic) timeout value. To set a fixed timeout see the
8567 'icp_query_timeout' directive.
8570 NAME: minimum_icp_query_timeout
8574 LOC: Config.Timeout.icp_query_min
8576 Normally the ICP query timeout is determined dynamically. But
8577 sometimes it can lead to very small timeouts, even lower than
8578 the normal latency variance on your link due to traffic.
8579 Use this option to put an lower limit on the dynamic timeout
8580 value. Do NOT use this option to always use a fixed (instead
8581 of a dynamic) timeout value. To set a fixed timeout see the
8582 'icp_query_timeout' directive.
8585 NAME: background_ping_rate
8589 LOC: Config.backgroundPingRate
8591 Controls how often the ICP pings are sent to siblings that
8592 have background-ping set.
8596 MULTICAST ICP OPTIONS
8597 -----------------------------------------------------------------------------
8602 LOC: Config.mcast_group_list
8605 This tag specifies a list of multicast groups which your server
8606 should join to receive multicasted ICP queries.
8608 NOTE! Be very careful what you put here! Be sure you
8609 understand the difference between an ICP _query_ and an ICP
8610 _reply_. This option is to be set only if you want to RECEIVE
8611 multicast queries. Do NOT set this option to SEND multicast
8612 ICP (use cache_peer for that). ICP replies are always sent via
8613 unicast, so this option does not affect whether or not you will
8614 receive replies from multicast group members.
8616 You must be very careful to NOT use a multicast address which
8617 is already in use by another group of caches.
8619 If you are unsure about multicast, please read the Multicast
8620 chapter in the Squid FAQ (http://www.squid-cache.org/FAQ/).
8622 Usage: mcast_groups 239.128.16.128 224.0.1.20
8624 By default, Squid doesn't listen on any multicast groups.
8627 NAME: mcast_miss_addr
8628 IFDEF: MULTICAST_MISS_STREAM
8630 LOC: Config.mcast_miss.addr
8632 DEFAULT_DOC: disabled.
8634 If you enable this option, every "cache miss" URL will
8635 be sent out on the specified multicast address.
8637 Do not enable this option unless you are are absolutely
8638 certain you understand what you are doing.
8641 NAME: mcast_miss_ttl
8642 IFDEF: MULTICAST_MISS_STREAM
8644 LOC: Config.mcast_miss.ttl
8647 This is the time-to-live value for packets multicasted
8648 when multicasting off cache miss URLs is enabled. By
8649 default this is set to 'site scope', i.e. 16.
8652 NAME: mcast_miss_port
8653 IFDEF: MULTICAST_MISS_STREAM
8655 LOC: Config.mcast_miss.port
8658 This is the port number to be used in conjunction with
8662 NAME: mcast_miss_encode_key
8663 IFDEF: MULTICAST_MISS_STREAM
8665 LOC: Config.mcast_miss.encode_key
8666 DEFAULT: XXXXXXXXXXXXXXXX
8668 The URLs that are sent in the multicast miss stream are
8669 encrypted. This is the encryption key.
8672 NAME: mcast_icp_query_timeout
8676 LOC: Config.Timeout.mcast_icp_query
8678 For multicast peers, Squid regularly sends out ICP "probes" to
8679 count how many other peers are listening on the given multicast
8680 address. This value specifies how long Squid should wait to
8681 count all the replies. The default is 2000 msec, or 2
8686 INTERNAL ICON OPTIONS
8687 -----------------------------------------------------------------------------
8690 NAME: icon_directory
8692 LOC: Config.icons.directory
8693 DEFAULT: @DEFAULT_ICON_DIR@
8695 Where the icons are stored. These are normally kept in
8699 NAME: global_internal_static
8701 LOC: Config.onoff.global_internal_static
8704 This directive controls is Squid should intercept all requests for
8705 /squid-internal-static/ no matter which host the URL is requesting
8706 (default on setting), or if nothing special should be done for
8707 such URLs (off setting). The purpose of this directive is to make
8708 icons etc work better in complex cache hierarchies where it may
8709 not always be possible for all corners in the cache mesh to reach
8710 the server generating a directory listing.
8713 NAME: short_icon_urls
8715 LOC: Config.icons.use_short_names
8718 If this is enabled Squid will use short URLs for icons.
8719 If disabled it will revert to the old behavior of including
8720 it's own name and port in the URL.
8722 If you run a complex cache hierarchy with a mix of Squid and
8723 other proxies you may need to disable this directive.
8728 -----------------------------------------------------------------------------
8731 NAME: error_directory
8733 LOC: Config.errorDirectory
8735 DEFAULT_DOC: Send error pages in the clients preferred language
8737 If you wish to create your own versions of the default
8738 error files to customize them to suit your company copy
8739 the error/template files to another directory and point
8742 WARNING: This option will disable multi-language support
8743 on error pages if used.
8745 The squid developers are interested in making squid available in
8746 a wide variety of languages. If you are making translations for a
8747 language that Squid does not currently provide please consider
8748 contributing your translation back to the project.
8749 http://wiki.squid-cache.org/Translations
8751 The squid developers working on translations are happy to supply drop-in
8752 translated error files in exchange for any new language contributions.
8755 NAME: error_default_language
8756 IFDEF: USE_ERR_LOCALES
8758 LOC: Config.errorDefaultLanguage
8760 DEFAULT_DOC: Generate English language pages.
8762 Set the default language which squid will send error pages in
8763 if no existing translation matches the clients language
8766 If unset (default) generic English will be used.
8768 The squid developers are interested in making squid available in
8769 a wide variety of languages. If you are interested in making
8770 translations for any language see the squid wiki for details.
8771 http://wiki.squid-cache.org/Translations
8774 NAME: error_log_languages
8775 IFDEF: USE_ERR_LOCALES
8777 LOC: Config.errorLogMissingLanguages
8780 Log to cache.log what languages users are attempting to
8781 auto-negotiate for translations.
8783 Successful negotiations are not logged. Only failures
8784 have meaning to indicate that Squid may need an upgrade
8785 of its error page translations.
8788 NAME: err_page_stylesheet
8790 LOC: Config.errorStylesheet
8791 DEFAULT: @DEFAULT_CONFIG_DIR@/errorpage.css
8793 CSS Stylesheet to pattern the display of Squid default error pages.
8795 For information on CSS see http://www.w3.org/Style/CSS/
8800 LOC: Config.errHtmlText
8803 HTML text to include in error messages. Make this a "mailto"
8804 URL to your admin address, or maybe just a link to your
8805 organizations Web page.
8807 To include this in your error messages, you must rewrite
8808 the error template files (found in the "errors" directory).
8809 Wherever you want the 'err_html_text' line to appear,
8810 insert a %L tag in the error template file.
8813 NAME: email_err_data
8816 LOC: Config.onoff.emailErrData
8819 If enabled, information about the occurred error will be
8820 included in the mailto links of the ERR pages (if %W is set)
8821 so that the email body contains the data.
8822 Syntax is <A HREF="mailto:%w%W">%w</A>
8827 LOC: Config.denyInfoList
8830 Usage: deny_info err_page_name acl
8831 or deny_info http://... acl
8832 or deny_info TCP_RESET acl
8834 This can be used to return a ERR_ page for requests which
8835 do not pass the 'http_access' rules. Squid remembers the last
8836 acl it evaluated in http_access, and if a 'deny_info' line exists
8837 for that ACL Squid returns a corresponding error page.
8839 The acl is typically the last acl on the http_access deny line which
8840 denied access. The exceptions to this rule are:
8841 - When Squid needs to request authentication credentials. It's then
8842 the first authentication related acl encountered
8843 - When none of the http_access lines matches. It's then the last
8844 acl processed on the last http_access line.
8845 - When the decision to deny access was made by an adaptation service,
8846 the acl name is the corresponding eCAP or ICAP service_name.
8848 NP: If providing your own custom error pages with error_directory
8849 you may also specify them by your custom file name:
8850 Example: deny_info ERR_CUSTOM_ACCESS_DENIED bad_guys
8852 By default Squid will send "403 Forbidden". A different 4xx or 5xx
8853 may be specified by prefixing the file name with the code and a colon.
8854 e.g. 404:ERR_CUSTOM_ACCESS_DENIED
8856 Alternatively you can tell Squid to reset the TCP connection
8857 by specifying TCP_RESET.
8859 Or you can specify an error URL or URL pattern. The browsers will
8860 get redirected to the specified URL after formatting tags have
8861 been replaced. Redirect will be done with 302 or 307 according to
8862 HTTP/1.1 specs. A different 3xx code may be specified by prefixing
8863 the URL. e.g. 303:http://example.com/
8866 %a - username (if available. Password NOT included)
8867 %A - Local listening IP address the client connection was connected to
8870 %E - Error description
8872 %H - Request domain name
8873 %i - Client IP Address
8875 %O - Unescaped message result from external ACL helper
8876 %o - Message result from external ACL helper
8877 %p - Request Port number
8878 %P - Request Protocol name
8879 %R - Request URL path
8880 %T - Timestamp in RFC 1123 format
8881 %U - Full canonical URL from client
8882 (HTTPS URLs terminate with *)
8883 %u - Full canonical URL from client
8884 %w - Admin email from squid.conf
8886 %% - Literal percent (%) code
8891 OPTIONS INFLUENCING REQUEST FORWARDING
8892 -----------------------------------------------------------------------------
8895 NAME: nonhierarchical_direct
8897 LOC: Config.onoff.nonhierarchical_direct
8900 By default, Squid will send any non-hierarchical requests
8901 (not cacheable request type) direct to origin servers.
8903 When this is set to "off", Squid will prefer to send these
8904 requests to parents.
8906 Note that in most configurations, by turning this off you will only
8907 add latency to these request without any improvement in global hit
8910 This option only sets a preference. If the parent is unavailable a
8911 direct connection to the origin server may still be attempted. To
8912 completely prevent direct connections use never_direct.
8917 LOC: Config.onoff.prefer_direct
8920 Normally Squid tries to use parents for most requests. If you for some
8921 reason like it to first try going direct and only use a parent if
8922 going direct fails set this to on.
8924 By combining nonhierarchical_direct off and prefer_direct on you
8925 can set up Squid to use a parent as a backup path if going direct
8928 Note: If you want Squid to use parents for all requests see
8929 the never_direct directive. prefer_direct only modifies how Squid
8930 acts on cacheable requests.
8933 NAME: cache_miss_revalidate
8937 LOC: Config.onoff.cache_miss_revalidate
8939 RFC 7232 defines a conditional request mechanism to prevent
8940 response objects being unnecessarily transferred over the network.
8941 If that mechanism is used by the client and a cache MISS occurs
8942 it can prevent new cache entries being created.
8944 This option determines whether Squid on cache MISS will pass the
8945 client revalidation request to the server or tries to fetch new
8946 content for caching. It can be useful while the cache is mostly
8947 empty to more quickly have the cache populated by generating
8948 non-conditional GETs.
8950 When set to 'on' (default), Squid will pass all client If-* headers
8951 to the server. This permits server responses without a cacheable
8952 payload to be delivered and on MISS no new cache entry is created.
8954 When set to 'off' and if the request is cacheable, Squid will
8955 remove the clients If-Modified-Since and If-None-Match headers from
8956 the request sent to the server. This requests a 200 status response
8957 from the server to create a new cache entry with.
8962 LOC: Config.accessList.AlwaysDirect
8964 DEFAULT_DOC: Prevent any cache_peer being used for this request.
8966 Usage: always_direct allow|deny [!]aclname ...
8968 Here you can use ACL elements to specify requests which should
8969 ALWAYS be forwarded by Squid to the origin servers without using
8970 any peers. For example, to always directly forward requests for
8971 local servers ignoring any parents or siblings you may have use
8974 acl local-servers dstdomain my.domain.net
8975 always_direct allow local-servers
8977 To always forward FTP requests directly, use
8980 always_direct allow FTP
8982 NOTE: There is a similar, but opposite option named
8983 'never_direct'. You need to be aware that "always_direct deny
8984 foo" is NOT the same thing as "never_direct allow foo". You
8985 may need to use a deny rule to exclude a more-specific case of
8986 some other rule. Example:
8988 acl local-external dstdomain external.foo.net
8989 acl local-servers dstdomain .foo.net
8990 always_direct deny local-external
8991 always_direct allow local-servers
8993 NOTE: If your goal is to make the client forward the request
8994 directly to the origin server bypassing Squid then this needs
8995 to be done in the client configuration. Squid configuration
8996 can only tell Squid how Squid should fetch the object.
8998 NOTE: This directive is not related to caching. The replies
8999 is cached as usual even if you use always_direct. To not cache
9000 the replies see the 'cache' directive.
9002 This clause supports both fast and slow acl types.
9003 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
9008 LOC: Config.accessList.NeverDirect
9010 DEFAULT_DOC: Allow DNS results to be used for this request.
9012 Usage: never_direct allow|deny [!]aclname ...
9014 never_direct is the opposite of always_direct. Please read
9015 the description for always_direct if you have not already.
9017 With 'never_direct' you can use ACL elements to specify
9018 requests which should NEVER be forwarded directly to origin
9019 servers. For example, to force the use of a proxy for all
9020 requests, except those in your local domain use something like:
9022 acl local-servers dstdomain .foo.net
9023 never_direct deny local-servers
9024 never_direct allow all
9026 or if Squid is inside a firewall and there are local intranet
9027 servers inside the firewall use something like:
9029 acl local-intranet dstdomain .foo.net
9030 acl local-external dstdomain external.foo.net
9031 always_direct deny local-external
9032 always_direct allow local-intranet
9033 never_direct allow all
9035 This clause supports both fast and slow acl types.
9036 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
9040 ADVANCED NETWORKING OPTIONS
9041 -----------------------------------------------------------------------------
9044 NAME: incoming_udp_average incoming_icp_average
9047 LOC: Config.comm_incoming.udp.average
9049 Heavy voodoo here. I can't even believe you are reading this.
9050 Are you crazy? Don't even think about adjusting these unless
9051 you understand the algorithms in comm_select.c first!
9054 NAME: incoming_tcp_average incoming_http_average
9057 LOC: Config.comm_incoming.tcp.average
9059 Heavy voodoo here. I can't even believe you are reading this.
9060 Are you crazy? Don't even think about adjusting these unless
9061 you understand the algorithms in comm_select.c first!
9064 NAME: incoming_dns_average
9067 LOC: Config.comm_incoming.dns.average
9069 Heavy voodoo here. I can't even believe you are reading this.
9070 Are you crazy? Don't even think about adjusting these unless
9071 you understand the algorithms in comm_select.c first!
9074 NAME: min_udp_poll_cnt min_icp_poll_cnt
9077 LOC: Config.comm_incoming.udp.min_poll
9079 Heavy voodoo here. I can't even believe you are reading this.
9080 Are you crazy? Don't even think about adjusting these unless
9081 you understand the algorithms in comm_select.c first!
9084 NAME: min_dns_poll_cnt
9087 LOC: Config.comm_incoming.dns.min_poll
9089 Heavy voodoo here. I can't even believe you are reading this.
9090 Are you crazy? Don't even think about adjusting these unless
9091 you understand the algorithms in comm_select.c first!
9094 NAME: min_tcp_poll_cnt min_http_poll_cnt
9097 LOC: Config.comm_incoming.tcp.min_poll
9099 Heavy voodoo here. I can't even believe you are reading this.
9100 Are you crazy? Don't even think about adjusting these unless
9101 you understand the algorithms in comm_select.c first!
9107 LOC: Config.accept_filter
9111 The name of an accept(2) filter to install on Squid's
9112 listen socket(s). This feature is perhaps specific to
9113 FreeBSD and requires support in the kernel.
9115 The 'httpready' filter delays delivering new connections
9116 to Squid until a full HTTP request has been received.
9117 See the accf_http(9) man page for details.
9119 The 'dataready' filter delays delivering new connections
9120 to Squid until there is some data to process.
9121 See the accf_dataready(9) man page for details.
9125 The 'data' filter delays delivering of new connections
9126 to Squid until there is some data to process by TCP_ACCEPT_DEFER.
9127 You may optionally specify a number of seconds to wait by
9128 'data=N' where N is the number of seconds. Defaults to 30
9129 if not specified. See the tcp(7) man page for details.
9132 accept_filter httpready
9137 NAME: client_ip_max_connections
9139 LOC: Config.client_ip_max_connections
9141 DEFAULT_DOC: No limit.
9143 Set an absolute limit on the number of connections a single
9144 client IP can use. Any more than this and Squid will begin to drop
9145 new connections from the client until it closes some links.
9147 Note that this is a global limit. It affects all HTTP, HTCP, Gopher and FTP
9148 connections from the client. For finer control use the ACL access controls.
9150 Requires client_db to be enabled (the default).
9152 WARNING: This may noticeably slow down traffic received via external proxies
9153 or NAT devices and cause them to rebound error messages back to their clients.
9156 NAME: tcp_recv_bufsize
9160 DEFAULT_DOC: Use operating system TCP defaults.
9161 LOC: Config.tcpRcvBufsz
9163 Size of receive buffer to set for TCP sockets. Probably just
9164 as easy to change your kernel's default.
9165 Omit from squid.conf to use the default buffer size.
9170 -----------------------------------------------------------------------------
9177 LOC: Adaptation::Icap::TheConfig.onoff
9180 If you want to enable the ICAP module support, set this to on.
9183 NAME: icap_connect_timeout
9186 LOC: Adaptation::Icap::TheConfig.connect_timeout_raw
9189 This parameter specifies how long to wait for the TCP connect to
9190 the requested ICAP server to complete before giving up and either
9191 terminating the HTTP transaction or bypassing the failure.
9193 The default for optional services is peer_connect_timeout.
9194 The default for essential services is connect_timeout.
9195 If this option is explicitly set, its value applies to all services.
9198 NAME: icap_io_timeout
9202 DEFAULT_DOC: Use read_timeout.
9203 LOC: Adaptation::Icap::TheConfig.io_timeout_raw
9206 This parameter specifies how long to wait for an I/O activity on
9207 an established, active ICAP connection before giving up and
9208 either terminating the HTTP transaction or bypassing the
9212 NAME: icap_service_failure_limit
9213 COMMENT: limit [in memory-depth time-units]
9214 TYPE: icap_service_failure_limit
9216 LOC: Adaptation::Icap::TheConfig
9219 The limit specifies the number of failures that Squid tolerates
9220 when establishing a new TCP connection with an ICAP service. If
9221 the number of failures exceeds the limit, the ICAP service is
9222 not used for new ICAP requests until it is time to refresh its
9225 A negative value disables the limit. Without the limit, an ICAP
9226 service will not be considered down due to connectivity failures
9227 between ICAP OPTIONS requests.
9229 Squid forgets ICAP service failures older than the specified
9230 value of memory-depth. The memory fading algorithm
9231 is approximate because Squid does not remember individual
9232 errors but groups them instead, splitting the option
9233 value into ten time slots of equal length.
9235 When memory-depth is 0 and by default this option has no
9236 effect on service failure expiration.
9238 Squid always forgets failures when updating service settings
9239 using an ICAP OPTIONS transaction, regardless of this option
9243 # suspend service usage after 10 failures in 5 seconds:
9244 icap_service_failure_limit 10 in 5 seconds
9247 NAME: icap_service_revival_delay
9250 LOC: Adaptation::Icap::TheConfig.service_revival_delay
9253 The delay specifies the number of seconds to wait after an ICAP
9254 OPTIONS request failure before requesting the options again. The
9255 failed ICAP service is considered "down" until fresh OPTIONS are
9258 The actual delay cannot be smaller than the hardcoded minimum
9259 delay of 30 seconds.
9262 NAME: icap_preview_enable
9266 LOC: Adaptation::Icap::TheConfig.preview_enable
9269 The ICAP Preview feature allows the ICAP server to handle the
9270 HTTP message by looking only at the beginning of the message body
9271 or even without receiving the body at all. In some environments,
9272 previews greatly speedup ICAP processing.
9274 During an ICAP OPTIONS transaction, the server may tell Squid what
9275 HTTP messages should be previewed and how big the preview should be.
9276 Squid will not use Preview if the server did not request one.
9278 To disable ICAP Preview for all ICAP services, regardless of
9279 individual ICAP server OPTIONS responses, set this option to "off".
9281 icap_preview_enable off
9284 NAME: icap_preview_size
9287 LOC: Adaptation::Icap::TheConfig.preview_size
9289 DEFAULT_DOC: No preview sent.
9291 The default size of preview data to be sent to the ICAP server.
9292 This value might be overwritten on a per server basis by OPTIONS requests.
9295 NAME: icap_206_enable
9299 LOC: Adaptation::Icap::TheConfig.allow206_enable
9302 206 (Partial Content) responses is an ICAP extension that allows the
9303 ICAP agents to optionally combine adapted and original HTTP message
9304 content. The decision to combine is postponed until the end of the
9305 ICAP response. Squid supports Partial Content extension by default.
9307 Activation of the Partial Content extension is negotiated with each
9308 ICAP service during OPTIONS exchange. Most ICAP servers should handle
9309 negotiation correctly even if they do not support the extension, but
9310 some might fail. To disable Partial Content support for all ICAP
9311 services and to avoid any negotiation, set this option to "off".
9317 NAME: icap_default_options_ttl
9320 LOC: Adaptation::Icap::TheConfig.default_options_ttl
9323 The default TTL value for ICAP OPTIONS responses that don't have
9324 an Options-TTL header.
9327 NAME: icap_persistent_connections
9331 LOC: Adaptation::Icap::TheConfig.reuse_connections
9334 Whether or not Squid should use persistent connections to
9338 NAME: adaptation_send_client_ip icap_send_client_ip
9340 IFDEF: USE_ADAPTATION
9342 LOC: Adaptation::Config::send_client_ip
9345 If enabled, Squid shares HTTP client IP information with adaptation
9346 services. For ICAP, Squid adds the X-Client-IP header to ICAP requests.
9347 For eCAP, Squid sets the libecap::metaClientIp transaction option.
9349 See also: adaptation_uses_indirect_client
9352 NAME: adaptation_send_username icap_send_client_username
9354 IFDEF: USE_ADAPTATION
9356 LOC: Adaptation::Config::send_username
9359 This sends authenticated HTTP client username (if available) to
9360 the adaptation service.
9362 For ICAP, the username value is encoded based on the
9363 icap_client_username_encode option and is sent using the header
9364 specified by the icap_client_username_header option.
9367 NAME: icap_client_username_header
9370 LOC: Adaptation::Icap::TheConfig.client_username_header
9371 DEFAULT: X-Client-Username
9373 ICAP request header name to use for adaptation_send_username.
9376 NAME: icap_client_username_encode
9380 LOC: Adaptation::Icap::TheConfig.client_username_encode
9383 Whether to base64 encode the authenticated client username.
9387 TYPE: icap_service_type
9389 LOC: Adaptation::Icap::TheConfig
9392 Defines a single ICAP service using the following format:
9394 icap_service id vectoring_point uri [option ...]
9397 an opaque identifier or name which is used to direct traffic to
9398 this specific service. Must be unique among all adaptation
9399 services in squid.conf.
9401 vectoring_point: reqmod_precache|reqmod_postcache|respmod_precache|respmod_postcache
9402 This specifies at which point of transaction processing the
9403 ICAP service should be activated. *_postcache vectoring points
9404 are not yet supported.
9406 uri: icap://servername:port/servicepath
9407 ICAP server and service location.
9408 icaps://servername:port/servicepath
9409 The "icap:" URI scheme is used for traditional ICAP server and
9410 service location (default port is 1344, connections are not
9411 encrypted). The "icaps:" URI scheme is for Secure ICAP
9412 services that use SSL/TLS-encrypted ICAP connections (by
9413 default, on port 11344).
9415 ICAP does not allow a single service to handle both REQMOD and RESPMOD
9416 transactions. Squid does not enforce that requirement. You can specify
9417 services with the same service_url and different vectoring_points. You
9418 can even specify multiple identical services as long as their
9419 service_names differ.
9421 To activate a service, use the adaptation_access directive. To group
9422 services, use adaptation_service_chain and adaptation_service_set.
9424 Service options are separated by white space. ICAP services support
9425 the following name=value options:
9428 If set to 'on' or '1', the ICAP service is treated as
9429 optional. If the service cannot be reached or malfunctions,
9430 Squid will try to ignore any errors and process the message as
9431 if the service was not enabled. No all ICAP errors can be
9432 bypassed. If set to 0, the ICAP service is treated as
9433 essential and all ICAP errors will result in an error page
9434 returned to the HTTP client.
9436 Bypass is off by default: services are treated as essential.
9439 If set to 'on' or '1', the ICAP service is allowed to
9440 dynamically change the current message adaptation plan by
9441 returning a chain of services to be used next. The services
9442 are specified using the X-Next-Services ICAP response header
9443 value, formatted as a comma-separated list of service names.
9444 Each named service should be configured in squid.conf. Other
9445 services are ignored. An empty X-Next-Services value results
9446 in an empty plan which ends the current adaptation.
9448 Dynamic adaptation plan may cross or cover multiple supported
9449 vectoring points in their natural processing order.
9451 Routing is not allowed by default: the ICAP X-Next-Services
9452 response header is ignored.
9455 Only has effect on split-stack systems. The default on those systems
9456 is to use IPv4-only connections. When set to 'on' this option will
9457 make Squid use IPv6-only connections to contact this ICAP service.
9459 on-overload=block|bypass|wait|force
9460 If the service Max-Connections limit has been reached, do
9461 one of the following for each new ICAP transaction:
9462 * block: send an HTTP error response to the client
9463 * bypass: ignore the "over-connected" ICAP service
9464 * wait: wait (in a FIFO queue) for an ICAP connection slot
9465 * force: proceed, ignoring the Max-Connections limit
9467 In SMP mode with N workers, each worker assumes the service
9468 connection limit is Max-Connections/N, even though not all
9469 workers may use a given service.
9471 The default value is "bypass" if service is bypassable,
9472 otherwise it is set to "wait".
9476 Use the given number as the Max-Connections limit, regardless
9477 of the Max-Connections value given by the service, if any.
9479 connection-encryption=on|off
9480 Determines the ICAP service effect on the connections_encrypted
9483 The default is "on" for Secure ICAP services (i.e., those
9484 with the icaps:// service URIs scheme) and "off" for plain ICAP
9487 Does not affect ICAP connections (e.g., does not turn Secure
9490 ==== ICAPS / TLS OPTIONS ====
9492 These options are used for Secure ICAP (icaps://....) services only.
9494 tls-cert=/path/to/ssl/certificate
9495 A client X.509 certificate to use when connecting to
9498 tls-key=/path/to/ssl/key
9499 The private key corresponding to the previous
9502 If tls-key= is not specified tls-cert= is assumed to
9503 reference a PEM file containing both the certificate
9506 tls-cipher=... The list of valid TLS/SSL ciphers to use when connecting
9507 to this icap server.
9510 The minimum TLS protocol version to permit. To control
9511 SSLv3 use the tls-options= parameter.
9512 Supported Values: 1.0 (default), 1.1, 1.2
9514 tls-options=... Specify various OpenSSL library options:
9516 NO_SSLv3 Disallow the use of SSLv3
9519 Always create a new key when using
9520 temporary/ephemeral DH key exchanges
9522 ALL Enable various bug workarounds
9523 suggested as "harmless" by OpenSSL
9524 Be warned that this reduces SSL/TLS
9525 strength to some attacks.
9527 See the OpenSSL SSL_CTX_set_options documentation for a
9528 more complete list. Options relevant only to SSLv2 are
9531 tls-cafile= PEM file containing CA certificates to use when verifying
9532 the icap server certificate.
9533 Use to specify intermediate CA certificate(s) if not sent
9534 by the server. Or the full CA chain for the server when
9535 using the tls-default-ca=off flag.
9536 May be repeated to load multiple files.
9538 tls-capath=... A directory containing additional CA certificates to
9539 use when verifying the icap server certificate.
9540 Requires OpenSSL or LibreSSL.
9542 tls-crlfile=... A certificate revocation list file to use when
9543 verifying the icap server certificate.
9545 tls-flags=... Specify various flags modifying the Squid TLS implementation:
9548 Accept certificates even if they fail to
9551 Don't verify the icap server certificate
9552 matches the server name
9554 tls-default-ca[=off]
9555 Whether to use the system Trusted CAs. Default is ON.
9557 tls-domain= The icap server name as advertised in it's certificate.
9558 Used for verifying the correctness of the received icap
9559 server certificate. If not specified the icap server
9560 hostname extracted from ICAP URI will be used.
9562 Older icap_service format without optional named parameters is
9563 deprecated but supported for backward compatibility.
9566 icap_service svcBlocker reqmod_precache icap://icap1.mydomain.net:1344/reqmod bypass=0
9567 icap_service svcLogger reqmod_precache icaps://icap2.mydomain.net:11344/reqmod routing=on
9571 TYPE: icap_class_type
9576 This deprecated option was documented to define an ICAP service
9577 chain, even though it actually defined a set of similar, redundant
9578 services, and the chains were not supported.
9580 To define a set of redundant services, please use the
9581 adaptation_service_set directive. For service chains, use
9582 adaptation_service_chain.
9586 TYPE: icap_access_type
9591 This option is deprecated. Please use adaptation_access, which
9592 has the same ICAP functionality, but comes with better
9593 documentation, and eCAP support.
9598 -----------------------------------------------------------------------------
9605 LOC: Adaptation::Ecap::TheConfig.onoff
9608 Controls whether eCAP support is enabled.
9612 TYPE: ecap_service_type
9614 LOC: Adaptation::Ecap::TheConfig
9617 Defines a single eCAP service
9619 ecap_service id vectoring_point uri [option ...]
9622 an opaque identifier or name which is used to direct traffic to
9623 this specific service. Must be unique among all adaptation
9624 services in squid.conf.
9626 vectoring_point: reqmod_precache|reqmod_postcache|respmod_precache|respmod_postcache
9627 This specifies at which point of transaction processing the
9628 eCAP service should be activated. *_postcache vectoring points
9629 are not yet supported.
9631 uri: ecap://vendor/service_name?custom&cgi=style¶meters=optional
9632 Squid uses the eCAP service URI to match this configuration
9633 line with one of the dynamically loaded services. Each loaded
9634 eCAP service must have a unique URI. Obtain the right URI from
9635 the service provider.
9637 To activate a service, use the adaptation_access directive. To group
9638 services, use adaptation_service_chain and adaptation_service_set.
9640 Service options are separated by white space. eCAP services support
9641 the following name=value options:
9644 If set to 'on' or '1', the eCAP service is treated as optional.
9645 If the service cannot be reached or malfunctions, Squid will try
9646 to ignore any errors and process the message as if the service
9647 was not enabled. No all eCAP errors can be bypassed.
9648 If set to 'off' or '0', the eCAP service is treated as essential
9649 and all eCAP errors will result in an error page returned to the
9652 Bypass is off by default: services are treated as essential.
9655 If set to 'on' or '1', the eCAP service is allowed to
9656 dynamically change the current message adaptation plan by
9657 returning a chain of services to be used next.
9659 Dynamic adaptation plan may cross or cover multiple supported
9660 vectoring points in their natural processing order.
9662 Routing is not allowed by default.
9664 connection-encryption=on|off
9665 Determines the eCAP service effect on the connections_encrypted
9668 Defaults to "on", which does not taint the master transaction
9671 Does not affect eCAP API calls.
9673 Older ecap_service format without optional named parameters is
9674 deprecated but supported for backward compatibility.
9678 ecap_service s1 reqmod_precache ecap://filters.R.us/leakDetector?on_error=block bypass=off
9679 ecap_service s2 respmod_precache ecap://filters.R.us/virusFilter config=/etc/vf.cfg bypass=on
9682 NAME: loadable_modules
9684 IFDEF: USE_LOADABLE_MODULES
9685 LOC: Config.loadable_module_names
9688 Instructs Squid to load the specified dynamic module(s) or activate
9689 preloaded module(s).
9691 loadable_modules @DEFAULT_PREFIX@/lib/MinimalAdapter.so
9695 MESSAGE ADAPTATION OPTIONS
9696 -----------------------------------------------------------------------------
9699 NAME: adaptation_service_set
9700 TYPE: adaptation_service_set_type
9701 IFDEF: USE_ADAPTATION
9706 Configures an ordered set of similar, redundant services. This is
9707 useful when hot standby or backup adaptation servers are available.
9709 adaptation_service_set set_name service_name1 service_name2 ...
9711 The named services are used in the set declaration order. The first
9712 applicable adaptation service from the set is used first. The next
9713 applicable service is tried if and only if the transaction with the
9714 previous service fails and the message waiting to be adapted is still
9717 When adaptation starts, broken services are ignored as if they were
9718 not a part of the set. A broken service is a down optional service.
9720 The services in a set must be attached to the same vectoring point
9721 (e.g., pre-cache) and use the same adaptation method (e.g., REQMOD).
9723 If all services in a set are optional then adaptation failures are
9724 bypassable. If all services in the set are essential, then a
9725 transaction failure with one service may still be retried using
9726 another service from the set, but when all services fail, the master
9727 transaction fails as well.
9729 A set may contain a mix of optional and essential services, but that
9730 is likely to lead to surprising results because broken services become
9731 ignored (see above), making previously bypassable failures fatal.
9732 Technically, it is the bypassability of the last failed service that
9735 See also: adaptation_access adaptation_service_chain
9738 adaptation_service_set svcBlocker urlFilterPrimary urlFilterBackup
9739 adaptation service_set svcLogger loggerLocal loggerRemote
9742 NAME: adaptation_service_chain
9743 TYPE: adaptation_service_chain_type
9744 IFDEF: USE_ADAPTATION
9749 Configures a list of complementary services that will be applied
9750 one-by-one, forming an adaptation chain or pipeline. This is useful
9751 when Squid must perform different adaptations on the same message.
9753 adaptation_service_chain chain_name service_name1 svc_name2 ...
9755 The named services are used in the chain declaration order. The first
9756 applicable adaptation service from the chain is used first. The next
9757 applicable service is applied to the successful adaptation results of
9758 the previous service in the chain.
9760 When adaptation starts, broken services are ignored as if they were
9761 not a part of the chain. A broken service is a down optional service.
9763 Request satisfaction terminates the adaptation chain because Squid
9764 does not currently allow declaration of RESPMOD services at the
9765 "reqmod_precache" vectoring point (see icap_service or ecap_service).
9767 The services in a chain must be attached to the same vectoring point
9768 (e.g., pre-cache) and use the same adaptation method (e.g., REQMOD).
9770 A chain may contain a mix of optional and essential services. If an
9771 essential adaptation fails (or the failure cannot be bypassed for
9772 other reasons), the master transaction fails. Otherwise, the failure
9773 is bypassed as if the failed adaptation service was not in the chain.
9775 See also: adaptation_access adaptation_service_set
9778 adaptation_service_chain svcRequest requestLogger urlFilter leakDetector
9781 NAME: adaptation_access
9782 TYPE: adaptation_access_type
9783 IFDEF: USE_ADAPTATION
9786 DEFAULT_DOC: Allow, unless rules exist in squid.conf.
9788 Sends an HTTP transaction to an ICAP or eCAP adaptation service.
9790 adaptation_access service_name allow|deny [!]aclname...
9791 adaptation_access set_name allow|deny [!]aclname...
9793 At each supported vectoring point, the adaptation_access
9794 statements are processed in the order they appear in this
9795 configuration file. Statements pointing to the following services
9796 are ignored (i.e., skipped without checking their ACL):
9798 - services serving different vectoring points
9799 - "broken-but-bypassable" services
9800 - "up" services configured to ignore such transactions
9801 (e.g., based on the ICAP Transfer-Ignore header).
9803 When a set_name is used, all services in the set are checked
9804 using the same rules, to find the first applicable one. See
9805 adaptation_service_set for details.
9807 If an access list is checked and there is a match, the
9808 processing stops: For an "allow" rule, the corresponding
9809 adaptation service is used for the transaction. For a "deny"
9810 rule, no adaptation service is activated.
9812 It is currently not possible to apply more than one adaptation
9813 service at the same vectoring point to the same HTTP transaction.
9815 See also: icap_service and ecap_service
9818 adaptation_access service_1 allow all
9821 NAME: adaptation_service_iteration_limit
9823 IFDEF: USE_ADAPTATION
9824 LOC: Adaptation::Config::service_iteration_limit
9827 Limits the number of iterations allowed when applying adaptation
9828 services to a message. If your longest adaptation set or chain
9829 may have more than 16 services, increase the limit beyond its
9830 default value of 16. If detecting infinite iteration loops sooner
9831 is critical, make the iteration limit match the actual number
9832 of services in your longest adaptation set or chain.
9834 Infinite adaptation loops are most likely with routing services.
9836 See also: icap_service routing=1
9839 NAME: adaptation_masterx_shared_names
9841 IFDEF: USE_ADAPTATION
9842 LOC: Adaptation::Config::masterx_shared_name
9845 For each master transaction (i.e., the HTTP request and response
9846 sequence, including all related ICAP and eCAP exchanges), Squid
9847 maintains a table of metadata. The table entries are (name, value)
9848 pairs shared among eCAP and ICAP exchanges. The table is destroyed
9849 with the master transaction.
9851 This option specifies the table entry names that Squid must accept
9852 from and forward to the adaptation transactions.
9854 An ICAP REQMOD or RESPMOD transaction may set an entry in the
9855 shared table by returning an ICAP header field with a name
9856 specified in adaptation_masterx_shared_names.
9858 An eCAP REQMOD or RESPMOD transaction may set an entry in the
9859 shared table by implementing the libecap::visitEachOption() API
9860 to provide an option with a name specified in
9861 adaptation_masterx_shared_names.
9863 Squid will store and forward the set entry to subsequent adaptation
9864 transactions within the same master transaction scope.
9866 Only one shared entry name is supported at this time.
9869 # share authentication information among ICAP services
9870 adaptation_masterx_shared_names X-Subscriber-ID
9873 NAME: adaptation_meta
9875 IFDEF: USE_ADAPTATION
9876 LOC: Adaptation::Config::metaHeaders
9879 This option allows Squid administrator to add custom ICAP request
9880 headers or eCAP options to Squid ICAP requests or eCAP transactions.
9881 Use it to pass custom authentication tokens and other
9882 transaction-state related meta information to an ICAP/eCAP service.
9884 The addition of a meta header is ACL-driven:
9885 adaptation_meta name value [!]aclname ...
9887 Processing for a given header name stops after the first ACL list match.
9888 Thus, it is impossible to add two headers with the same name. If no ACL
9889 lists match for a given header name, no such header is added. For
9892 # do not debug transactions except for those that need debugging
9893 adaptation_meta X-Debug 1 needs_debugging
9895 # log all transactions except for those that must remain secret
9896 adaptation_meta X-Log 1 !keep_secret
9898 # mark transactions from users in the "G 1" group
9899 adaptation_meta X-Authenticated-Groups "G 1" authed_as_G1
9901 The "value" parameter may be a regular squid.conf token or a "double
9902 quoted string". Within the quoted string, use backslash (\) to escape
9903 any character, which is currently only useful for escaping backslashes
9904 and double quotes. For example,
9905 "this string has one backslash (\\) and two \"quotes\""
9907 Used adaptation_meta header values may be logged via %note
9908 logformat code. If multiple adaptation_meta headers with the same name
9909 are used during master transaction lifetime, the header values are
9910 logged in the order they were used and duplicate values are ignored
9911 (only the first repeated value will be logged).
9917 LOC: Adaptation::Icap::TheConfig.repeat
9918 DEFAULT_IF_NONE: deny all
9920 This ACL determines which retriable ICAP transactions are
9921 retried. Transactions that received a complete ICAP response
9922 and did not have to consume or produce HTTP bodies to receive
9923 that response are usually retriable.
9925 icap_retry allow|deny [!]aclname ...
9927 Squid automatically retries some ICAP I/O timeouts and errors
9928 due to persistent connection race conditions.
9930 See also: icap_retry_limit
9933 NAME: icap_retry_limit
9936 LOC: Adaptation::Icap::TheConfig.repeat_limit
9938 DEFAULT_DOC: No retries are allowed.
9940 Limits the number of retries allowed.
9942 Communication errors due to persistent connection race
9943 conditions are unavoidable, automatically retried, and do not
9944 count against this limit.
9946 See also: icap_retry
9952 -----------------------------------------------------------------------------
9955 NAME: check_hostnames
9958 LOC: Config.onoff.check_hostnames
9960 For security and stability reasons Squid can check
9961 hostnames for Internet standard RFC compliance. If you want
9962 Squid to perform these checks turn this directive on.
9965 NAME: allow_underscore
9968 LOC: Config.onoff.allow_underscore
9970 Underscore characters is not strictly allowed in Internet hostnames
9971 but nevertheless used by many sites. Set this to off if you want
9972 Squid to be strict about the standard.
9973 This check is performed only when check_hostnames is set to on.
9976 NAME: dns_retransmit_interval
9979 LOC: Config.Timeout.idns_retransmit
9981 Initial retransmit interval for DNS queries. The interval is
9982 doubled each time all configured DNS servers have been tried.
9988 LOC: Config.Timeout.idns_query
9990 DNS Query timeout. If no response is received to a DNS query
9991 within this time all DNS servers for the queried domain
9992 are assumed to be unavailable.
9995 NAME: dns_packet_max
9997 DEFAULT_DOC: EDNS disabled
9999 LOC: Config.dns.packet_max
10001 Maximum number of bytes packet size to advertise via EDNS.
10002 Set to "none" to disable EDNS large packet support.
10004 For legacy reasons DNS UDP replies will default to 512 bytes which
10005 is too small for many responses. EDNS provides a means for Squid to
10006 negotiate receiving larger responses back immediately without having
10007 to failover with repeat requests. Responses larger than this limit
10008 will retain the old behaviour of failover to TCP DNS.
10010 Squid has no real fixed limit internally, but allowing packet sizes
10011 over 1500 bytes requires network jumbogram support and is usually not
10014 WARNING: The RFC also indicates that some older resolvers will reply
10015 with failure of the whole request if the extension is added. Some
10016 resolvers have already been identified which will reply with mangled
10017 EDNS response on occasion. Usually in response to many-KB jumbogram
10018 sizes being advertised by Squid.
10019 Squid will currently treat these both as an unable-to-resolve domain
10020 even if it would be resolvable without EDNS.
10027 DEFAULT_DOC: Search for single-label domain names is disabled.
10028 LOC: Config.onoff.res_defnames
10030 Normally the RES_DEFNAMES resolver option is disabled
10031 (see res_init(3)). This prevents caches in a hierarchy
10032 from interpreting single-component hostnames locally. To allow
10033 Squid to handle single-component names, enable this option.
10036 NAME: dns_multicast_local
10040 DEFAULT_DOC: Search for .local and .arpa names is disabled.
10041 LOC: Config.onoff.dns_mdns
10043 When set to on, Squid sends multicast DNS lookups on the local
10044 network for domains ending in .local and .arpa.
10045 This enables local servers and devices to be contacted in an
10046 ad-hoc or zero-configuration network environment.
10049 NAME: dns_nameservers
10052 DEFAULT_DOC: Use operating system definitions
10053 LOC: Config.dns.nameservers
10055 Use this if you want to specify a list of DNS name servers
10056 (IP addresses) to use instead of those given in your
10057 /etc/resolv.conf file.
10059 On Windows platforms, if no value is specified here or in
10060 the /etc/resolv.conf file, the list of DNS name servers are
10061 taken from the Windows registry, both static and dynamic DHCP
10062 configurations are supported.
10064 Example: dns_nameservers 10.0.0.1 192.172.0.4
10069 DEFAULT: @DEFAULT_HOSTS@
10070 LOC: Config.etcHostsPath
10072 Location of the host-local IP name-address associations
10073 database. Most Operating Systems have such a file on different
10075 - Un*X & Linux: /etc/hosts
10076 - Windows NT/2000: %SystemRoot%\system32\drivers\etc\hosts
10077 (%SystemRoot% value install default is c:\winnt)
10078 - Windows XP/2003: %SystemRoot%\system32\drivers\etc\hosts
10079 (%SystemRoot% value install default is c:\windows)
10080 - Windows 9x/Me: %windir%\hosts
10081 (%windir% value is usually c:\windows)
10082 - Cygwin: /etc/hosts
10084 The file contains newline-separated definitions, in the
10085 form ip_address_in_dotted_form name [name ...] names are
10086 whitespace-separated. Lines beginning with an hash (#)
10087 character are comments.
10089 The file is checked at startup and upon configuration.
10090 If set to 'none', it won't be checked.
10091 If append_domain is used, that domain will be added to
10092 domain-local (i.e. not containing any dot character) host
10096 NAME: append_domain
10098 LOC: Config.appendDomain
10100 DEFAULT_DOC: Use operating system definitions
10102 Appends local domain name to hostnames without any dots in
10103 them. append_domain must begin with a period.
10105 Be warned there are now Internet names with no dots in
10106 them using only top-domain names, so setting this may
10107 cause some Internet sites to become unavailable.
10110 append_domain .yourdomain.com
10113 NAME: ignore_unknown_nameservers
10115 LOC: Config.onoff.ignore_unknown_nameservers
10118 By default Squid checks that DNS responses are received
10119 from the same IP addresses they are sent to. If they
10120 don't match, Squid ignores the response and writes a warning
10121 message to cache.log. You can allow responses from unknown
10122 nameservers by setting this option to 'off'.
10126 COMMENT: (number of entries)
10129 LOC: Config.ipcache.size
10131 Maximum number of DNS IP cache entries.
10138 LOC: Config.ipcache.low
10145 LOC: Config.ipcache.high
10147 The size, low-, and high-water marks for the IP cache.
10150 NAME: fqdncache_size
10151 COMMENT: (number of entries)
10154 LOC: Config.fqdncache.size
10156 Maximum number of FQDN cache entries.
10161 -----------------------------------------------------------------------------
10164 NAME: configuration_includes_quoted_values
10166 TYPE: configuration_includes_quoted_values
10168 LOC: ConfigParser::RecognizeQuotedValues
10170 If set, Squid will recognize each "quoted string" after a configuration
10171 directive as a single parameter. The quotes are stripped before the
10172 parameter value is interpreted or used.
10173 See "Values with spaces, quotes, and other special characters"
10174 section for more details.
10181 LOC: Config.onoff.mem_pools
10183 If set, Squid will keep pools of allocated (but unused) memory
10184 available for future use. If memory is a premium on your
10185 system and you believe your malloc library outperforms Squid
10186 routines, disable this.
10189 NAME: memory_pools_limit
10193 LOC: Config.MemPools.limit
10195 Used only with memory_pools on:
10196 memory_pools_limit 50 MB
10198 If set to a non-zero value, Squid will keep at most the specified
10199 limit of allocated (but unused) memory in memory pools. All free()
10200 requests that exceed this limit will be handled by your malloc
10201 library. Squid does not pre-allocate any memory, just safe-keeps
10202 objects that otherwise would be free()d. Thus, it is safe to set
10203 memory_pools_limit to a reasonably high value even if your
10204 configuration will use less memory.
10206 If set to none, Squid will keep all memory it can. That is, there
10207 will be no limit on the total amount of memory used for safe-keeping.
10209 To disable memory allocation optimization, do not set
10210 memory_pools_limit to 0 or none. Set memory_pools to "off" instead.
10212 An overhead for maintaining memory pools is not taken into account
10213 when the limit is checked. This overhead is close to four bytes per
10214 object kept. However, pools may actually _save_ memory because of
10215 reduced memory thrashing in your malloc library.
10218 NAME: forwarded_for
10219 COMMENT: on|off|transparent|truncate|delete
10222 LOC: opt_forwarded_for
10224 If set to "on", Squid will append your client's IP address
10225 in the HTTP requests it forwards. By default it looks like:
10227 X-Forwarded-For: 192.1.2.3
10229 If set to "off", it will appear as
10231 X-Forwarded-For: unknown
10233 If set to "transparent", Squid will not alter the
10234 X-Forwarded-For header in any way.
10236 If set to "delete", Squid will delete the entire
10237 X-Forwarded-For header.
10239 If set to "truncate", Squid will remove all existing
10240 X-Forwarded-For entries, and place the client IP as the sole entry.
10243 NAME: cachemgr_passwd
10244 TYPE: cachemgrpasswd
10246 DEFAULT_DOC: No password. Actions which require password are denied.
10247 LOC: Config.passwd_list
10249 Specify passwords for cachemgr operations.
10251 Usage: cachemgr_passwd password action action ...
10253 Some valid actions are (see cache manager menu for a full list):
10293 * Indicates actions which will not be performed without a
10294 valid password, others can be performed if not listed here.
10296 To disable an action, set the password to "disable".
10297 To allow performing an action without a password, set the
10298 password to "none".
10300 Use the keyword "all" to set the same password for all actions.
10303 cachemgr_passwd secret shutdown
10304 cachemgr_passwd lesssssssecret info stats/objects
10305 cachemgr_passwd disable all
10312 LOC: Config.onoff.client_db
10314 If you want to disable collecting per-client statistics,
10315 turn off client_db here.
10318 NAME: refresh_all_ims
10322 LOC: Config.onoff.refresh_all_ims
10324 When you enable this option, squid will always check
10325 the origin server for an update when a client sends an
10326 If-Modified-Since request. Many browsers use IMS
10327 requests when the user requests a reload, and this
10328 ensures those clients receive the latest version.
10330 By default (off), squid may return a Not Modified response
10331 based on the age of the cached version.
10334 NAME: reload_into_ims
10335 IFDEF: USE_HTTP_VIOLATIONS
10339 LOC: Config.onoff.reload_into_ims
10341 When you enable this option, client no-cache or ``reload''
10342 requests will be changed to If-Modified-Since requests.
10343 Doing this VIOLATES the HTTP standard. Enabling this
10344 feature could make you liable for problems which it
10347 see also refresh_pattern for a more selective approach.
10350 NAME: connect_retries
10352 LOC: Config.connect_retries
10354 DEFAULT_DOC: Do not retry failed connections.
10356 Limits the number of reopening attempts when establishing a single
10357 TCP connection. All these attempts must still complete before the
10358 applicable connection opening timeout expires.
10360 By default and when connect_retries is set to zero, Squid does not
10361 retry failed connection opening attempts.
10363 The (not recommended) maximum is 10 tries. An attempt to configure a
10364 higher value results in the value of 10 being used (with a warning).
10366 Squid may open connections to retry various high-level forwarding
10367 failures. For an outside observer, that activity may look like a
10368 low-level connection reopening attempt, but those high-level retries
10369 are governed by forward_max_tries instead.
10371 See also: connect_timeout, forward_timeout, icap_connect_timeout,
10372 ident_timeout, and forward_max_tries.
10375 NAME: retry_on_error
10377 LOC: Config.retry.onerror
10380 If set to ON Squid will automatically retry requests when
10381 receiving an error response with status 403 (Forbidden),
10382 500 (Internal Error), 501 or 503 (Service not available).
10383 Status 502 and 504 (Gateway errors) are always retried.
10385 This is mainly useful if you are in a complex cache hierarchy to
10386 work around access control errors.
10388 NOTE: This retry will attempt to find another working destination.
10389 Which is different from the server which just failed.
10392 NAME: as_whois_server
10394 LOC: Config.as_whois_server
10395 DEFAULT: whois.ra.net
10397 WHOIS server to query for AS numbers. NOTE: AS numbers are
10398 queried only when Squid starts up, not for every request.
10403 LOC: Config.onoff.offline
10406 Enable this option and Squid will never try to validate cached
10410 NAME: uri_whitespace
10411 TYPE: uri_whitespace
10412 LOC: Config.uri_whitespace
10415 What to do with requests that have whitespace characters in the
10418 strip: The whitespace characters are stripped out of the URL.
10419 This is the behavior recommended by RFC2396 and RFC3986
10420 for tolerant handling of generic URI.
10421 NOTE: This is one difference between generic URI and HTTP URLs.
10423 deny: The request is denied. The user receives an "Invalid
10425 This is the behaviour recommended by RFC2616 for safe
10426 handling of HTTP request URL.
10428 allow: The request is allowed and the URI is not changed. The
10429 whitespace characters remain in the URI. Note the
10430 whitespace is passed to redirector processes if they
10432 Note this may be considered a violation of RFC2616
10433 request parsing where whitespace is prohibited in the
10436 encode: The request is allowed and the whitespace characters are
10437 encoded according to RFC1738.
10439 chop: The request is allowed and the URI is chopped at the
10443 NOTE the current Squid implementation of encode and chop violates
10444 RFC2616 by not using a 301 redirect after altering the URL.
10449 LOC: Config.chroot_dir
10452 Specifies a directory where Squid should do a chroot() while
10453 initializing. This also causes Squid to fully drop root
10454 privileges after initializing. This means, for example, if you
10455 use a HTTP port less than 1024 and try to reconfigure, you may
10456 get an error saying that Squid can not open the port.
10459 NAME: pipeline_prefetch
10460 TYPE: pipelinePrefetch
10461 LOC: Config.pipeline_max_prefetch
10463 DEFAULT_DOC: Do not pre-parse pipelined requests.
10465 HTTP clients may send a pipeline of 1+N requests to Squid using a
10466 single connection, without waiting for Squid to respond to the first
10467 of those requests. This option limits the number of concurrent
10468 requests Squid will try to handle in parallel. If set to N, Squid
10469 will try to receive and process up to 1+N requests on the same
10470 connection concurrently.
10472 Defaults to 0 (off) for bandwidth management and access logging
10475 NOTE: pipelining requires persistent connections to clients.
10477 WARNING: pipelining breaks NTLM and Negotiate/Kerberos authentication.
10480 NAME: high_response_time_warning
10483 LOC: Config.warnings.high_rptm
10485 DEFAULT_DOC: disabled.
10487 If the one-minute median response time exceeds this value,
10488 Squid prints a WARNING with debug level 0 to get the
10489 administrators attention. The value is in milliseconds.
10492 NAME: high_page_fault_warning
10494 LOC: Config.warnings.high_pf
10496 DEFAULT_DOC: disabled.
10498 If the one-minute average page fault rate exceeds this
10499 value, Squid prints a WARNING with debug level 0 to get
10500 the administrators attention. The value is in page faults
10504 NAME: high_memory_warning
10506 LOC: Config.warnings.high_memory
10507 IFDEF: HAVE_MSTATS&&HAVE_GNUMALLOC_H
10509 DEFAULT_DOC: disabled.
10511 If the memory usage (as determined by gnumalloc, if available and used)
10512 exceeds this amount, Squid prints a WARNING with debug level 0 to get
10513 the administrators attention.
10515 # TODO: link high_memory_warning to mempools?
10517 NAME: sleep_after_fork
10518 COMMENT: (microseconds)
10520 LOC: Config.sleep_after_fork
10523 When this is set to a non-zero value, the main Squid process
10524 sleeps the specified number of microseconds after a fork()
10525 system call. This sleep may help the situation where your
10526 system reports fork() failures due to lack of (virtual)
10527 memory. Note, however, if you have a lot of child
10528 processes, these sleep delays will add up and your
10529 Squid will not service requests for some amount of time
10530 until all the child processes have been started.
10531 On Windows value less then 1000 (1 milliseconds) are
10535 NAME: windows_ipaddrchangemonitor
10536 IFDEF: _SQUID_WINDOWS_
10540 LOC: Config.onoff.WIN32_IpAddrChangeMonitor
10542 On Windows Squid by default will monitor IP address changes and will
10543 reconfigure itself after any detected event. This is very useful for
10544 proxies connected to internet with dial-up interfaces.
10545 In some cases (a Proxy server acting as VPN gateway is one) it could be
10546 desiderable to disable this behaviour setting this to 'off'.
10547 Note: after changing this, Squid service must be restarted.
10552 IFDEF: USE_SQUID_EUI
10554 LOC: Eui::TheConfig.euiLookup
10556 Whether to lookup the EUI or MAC address of a connected client.
10559 NAME: max_filedescriptors max_filedesc
10562 DEFAULT_DOC: Use operating system soft limit set by ulimit.
10563 LOC: Config.max_filedescriptors
10565 Set the maximum number of filedescriptors, either below the
10566 operating system default or up to the hard limit.
10568 Remove from squid.conf to inherit the current ulimit soft
10571 Note: Changing this requires a restart of Squid. Also
10572 not all I/O types supports large values (eg on Windows).
10575 NAME: force_request_body_continuation
10577 LOC: Config.accessList.forceRequestBodyContinuation
10579 DEFAULT_DOC: Deny, unless rules exist in squid.conf.
10581 This option controls how Squid handles data upload requests from HTTP
10582 and FTP agents that require a "Please Continue" control message response
10583 to actually send the request body to Squid. It is mostly useful in
10584 adaptation environments.
10586 When Squid receives an HTTP request with an "Expect: 100-continue"
10587 header or an FTP upload command (e.g., STOR), Squid normally sends the
10588 request headers or FTP command information to an adaptation service (or
10589 peer) and waits for a response. Most adaptation services (and some
10590 broken peers) may not respond to Squid at that stage because they may
10591 decide to wait for the HTTP request body or FTP data transfer. However,
10592 that request body or data transfer may never come because Squid has not
10593 responded with the HTTP 100 or FTP 150 (Please Continue) control message
10594 to the request sender yet!
10596 An allow match tells Squid to respond with the HTTP 100 or FTP 150
10597 (Please Continue) control message on its own, before forwarding the
10598 request to an adaptation service or peer. Such a response usually forces
10599 the request sender to proceed with sending the body. A deny match tells
10600 Squid to delay that control response until the origin server confirms
10601 that the request body is needed. Delaying is the default behavior.
10604 NAME: http_upgrade_request_protocols
10605 TYPE: http_upgrade_request_protocols
10606 LOC: Config.http_upgrade_request_protocols
10608 DEFAULT_DOC: Upgrade header dropped, effectively blocking an upgrade attempt.
10610 Controls client-initiated and server-confirmed switching from HTTP to
10611 another protocol (or to several protocols) using HTTP Upgrade mechanism
10612 defined in RFC 7230 Section 6.7. Squid itself does not understand the
10613 protocols being upgraded to and participates in the upgraded
10614 communication only as a dumb TCP proxy. Admins should not allow
10615 upgrading to protocols that require a more meaningful proxy
10618 Usage: http_upgrade_request_protocols <protocol> allow|deny [!]acl ...
10620 The required "protocol" parameter is either an all-caps word OTHER or an
10621 explicit protocol name (e.g. "WebSocket") optionally followed by a slash
10622 and a version token (e.g. "HTTP/3"). Explicit protocol names and
10623 versions are case sensitive.
10625 When an HTTP client sends an Upgrade request header, Squid iterates over
10626 the client-offered protocols and, for each protocol P (with an optional
10627 version V), evaluates the first non-empty set of
10628 http_upgrade_request_protocols rules (if any) from the following list:
10630 * All rules with an explicit protocol name equal to P.
10631 * All rules that use OTHER instead of a protocol name.
10633 In other words, rules using OTHER are considered for protocol P if and
10634 only if there are no rules mentioning P by name.
10636 If both of the above sets are empty, then Squid removes protocol P from
10639 If the client sent a versioned protocol offer P/X, then explicit rules
10640 referring to the same-name but different-version protocol P/Y are
10641 declared inapplicable. Inapplicable rules are not evaluated (i.e. are
10642 ignored). However, inapplicable rules still belong to the first set of
10645 Within the applicable rule subset, individual rules are evaluated in
10646 their configuration order. If all ACLs of an applicable "allow" rule
10647 match, then the protocol offered by the client is forwarded to the next
10648 hop as is. If all ACLs of an applicable "deny" rule match, then the
10649 offer is dropped. If no applicable rules have matching ACLs, then the
10650 offer is also dropped. The first matching rule also ends rules
10651 evaluation for the offered protocol.
10653 If all client-offered protocols are removed, then Squid forwards the
10654 client request without the Upgrade header. Squid never sends an empty
10655 Upgrade request header.
10657 An Upgrade request header with a value violating HTTP syntax is dropped
10658 and ignored without an attempt to use extractable individual protocol
10661 Upon receiving an HTTP 101 (Switching Protocols) control message, Squid
10662 checks that the server listed at least one protocol name and sent a
10663 Connection:upgrade response header. Squid does not understand individual
10664 protocol naming and versioning concepts enough to implement stricter
10665 checks, but an admin can restrict HTTP 101 (Switching Protocols)
10666 responses further using http_reply_access. Responses denied by
10667 http_reply_access rules and responses flagged by the internal Upgrade
10668 checks result in HTTP 502 (Bad Gateway) ERR_INVALID_RESP errors and
10669 Squid-to-server connection closures.
10671 If Squid sends an Upgrade request header, and the next hop (e.g., the
10672 origin server) responds with an acceptable HTTP 101 (Switching
10673 Protocols), then Squid forwards that message to the client and becomes
10676 The presence of an Upgrade request header alone does not preclude cache
10677 lookups. In other words, an Upgrade request might be satisfied from the
10678 cache, using regular HTTP caching rules.
10680 This clause only supports fast acl types.
10681 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
10683 Each of the following groups of configuration lines represents a
10684 separate configuration example:
10686 # never upgrade to protocol Foo; all others are OK
10687 http_upgrade_request_protocols Foo deny all
10688 http_upgrade_request_protocols OTHER allow all
10690 # only allow upgrades to protocol Bar (except for its first version)
10691 http_upgrade_request_protocols Bar/1 deny all
10692 http_upgrade_request_protocols Bar allow all
10693 http_upgrade_request_protocols OTHER deny all # this rule is optional
10695 # only allow upgrades to protocol Baz, and only if Baz is the only offer
10696 acl UpgradeHeaderHasMultipleOffers ...
10697 http_upgrade_request_protocols Baz deny UpgradeHeaderHasMultipleOffers
10698 http_upgrade_request_protocols Baz allow all
10701 NAME: server_pconn_for_nonretriable
10704 DEFAULT_DOC: Open new connections for forwarding requests Squid cannot retry safely.
10705 LOC: Config.accessList.serverPconnForNonretriable
10707 This option provides fine-grained control over persistent connection
10708 reuse when forwarding HTTP requests that Squid cannot retry. It is useful
10709 in environments where opening new connections is very expensive
10710 (e.g., all connections are secured with TLS with complex client and server
10711 certificate validation) and race conditions associated with persistent
10712 connections are very rare and/or only cause minor problems.
10714 HTTP prohibits retrying unsafe and non-idempotent requests (e.g., POST).
10715 Squid limitations also prohibit retrying all requests with bodies (e.g., PUT).
10716 By default, when forwarding such "risky" requests, Squid opens a new
10717 connection to the server or cache_peer, even if there is an idle persistent
10718 connection available. When Squid is configured to risk sending a non-retriable
10719 request on a previously used persistent connection, and the server closes
10720 the connection before seeing that risky request, the user gets an error response
10721 from Squid. In most cases, that error response will be HTTP 502 (Bad Gateway)
10722 with ERR_ZERO_SIZE_OBJECT or ERR_WRITE_ERROR (peer connection reset) error detail.
10724 If an allow rule matches, Squid reuses an available idle persistent connection
10725 (if any) for the request that Squid cannot retry. If a deny rule matches, then
10726 Squid opens a new connection for the request that Squid cannot retry.
10728 This option does not affect requests that Squid can retry. They will reuse idle
10729 persistent connections (if any).
10731 This clause only supports fast acl types.
10732 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
10735 acl SpeedIsWorthTheRisk method POST
10736 server_pconn_for_nonretriable allow SpeedIsWorthTheRisk
10739 NAME: happy_eyeballs_connect_timeout
10743 LOC: Config.happyEyeballs.connect_timeout
10745 This Happy Eyeballs (RFC 8305) tuning directive specifies the minimum
10746 delay between opening a primary to-server connection and opening a
10747 spare to-server connection for the same master transaction. This delay
10748 is similar to the Connection Attempt Delay in RFC 8305, but it is only
10749 applied to the first spare connection attempt. Subsequent spare
10750 connection attempts use happy_eyeballs_connect_gap, and primary
10751 connection attempts are not artificially delayed at all.
10753 Terminology: The "primary" and "spare" designations are determined by
10754 the order of DNS answers received by Squid: If Squid DNS AAAA query
10755 was answered first, then primary connections are connections to IPv6
10756 peer addresses (while spare connections use IPv4 addresses).
10757 Similarly, if Squid DNS A query was answered first, then primary
10758 connections are connections to IPv4 peer addresses (while spare
10759 connections use IPv6 addresses).
10761 Shorter happy_eyeballs_connect_timeout values reduce master
10762 transaction response time, potentially improving user-perceived
10763 response times (i.e., making user eyeballs happier). Longer delays
10764 reduce both concurrent connection level and server bombardment with
10765 connection requests, potentially improving overall Squid performance
10766 and reducing the chance of being blocked by servers for opening too
10767 many unused connections.
10769 RFC 8305 prohibits happy_eyeballs_connect_timeout values smaller than
10770 10 (milliseconds) to "avoid congestion collapse in the presence of
10771 high packet-loss rates".
10773 The following Happy Eyeballs directives place additional connection
10774 opening restrictions: happy_eyeballs_connect_gap and
10775 happy_eyeballs_connect_limit.
10778 NAME: happy_eyeballs_connect_gap
10782 DEFAULT_DOC: no artificial delays between spare attempts
10783 LOC: Config.happyEyeballs.connect_gap
10785 This Happy Eyeballs (RFC 8305) tuning directive specifies the
10786 minimum delay between opening spare to-server connections (to any
10787 server; i.e. across all concurrent master transactions in a Squid
10788 instance). Each SMP worker currently multiplies the configured gap
10789 by the total number of workers so that the combined spare connection
10790 opening rate of a Squid instance obeys the configured limit. The
10791 workers do not coordinate connection openings yet; a micro burst
10792 of spare connection openings may violate the configured gap.
10794 This directive has similar trade-offs as
10795 happy_eyeballs_connect_timeout, but its focus is on limiting traffic
10796 amplification effects for Squid as a whole, while
10797 happy_eyeballs_connect_timeout works on an individual master
10800 The following Happy Eyeballs directives place additional connection
10801 opening restrictions: happy_eyeballs_connect_timeout and
10802 happy_eyeballs_connect_limit. See the former for related terminology.
10805 NAME: happy_eyeballs_connect_limit
10808 DEFAULT_DOC: no artificial limit on the number of concurrent spare attempts
10809 LOC: Config.happyEyeballs.connect_limit
10811 This Happy Eyeballs (RFC 8305) tuning directive specifies the
10812 maximum number of spare to-server connections (to any server; i.e.
10813 across all concurrent master transactions in a Squid instance).
10814 Each SMP worker gets an equal share of the total limit. However,
10815 the workers do not share the actual connection counts yet, so one
10816 (busier) worker cannot "borrow" spare connection slots from another
10817 (less loaded) worker.
10819 Setting this limit to zero disables concurrent use of primary and
10820 spare TCP connections: Spare connection attempts are made only after
10821 all primary attempts fail. However, Squid would still use the
10822 DNS-related optimizations of the Happy Eyeballs approach.
10824 This directive has similar trade-offs as happy_eyeballs_connect_gap,
10825 but its focus is on limiting Squid overheads, while
10826 happy_eyeballs_connect_gap focuses on the origin server and peer
10829 The following Happy Eyeballs directives place additional connection
10830 opening restrictions: happy_eyeballs_connect_timeout and
10831 happy_eyeballs_connect_gap. See the former for related terminology.