2 # SQUID Web Proxy Cache http://www.squid-cache.org/
3 # ----------------------------------------------------------
5 # Squid is the result of efforts by numerous individuals from
6 # the Internet community; see the CONTRIBUTORS file for full
7 # details. Many organizations have provided support for Squid's
8 # development; see the SPONSORS file for full details. Squid is
9 # Copyrighted (C) 2000 by the Regents of the University of
10 # California; see the COPYRIGHT file for full details. Squid
11 # incorporates software developed and/or copyrighted by other
12 # sources; see the CREDITS file for full details.
14 # This program is free software; you can redistribute it and/or modify
15 # it under the terms of the GNU General Public License as published by
16 # the Free Software Foundation; either version 2 of the License, or
17 # (at your option) any later version.
19 # This program is distributed in the hope that it will be useful,
20 # but WITHOUT ANY WARRANTY; without even the implied warranty of
21 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
22 # GNU General Public License for more details.
24 # You should have received a copy of the GNU General Public License
25 # along with this program; if not, write to the Free Software
26 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA.
31 ----------------------------
33 This is the documentation for the Squid configuration file.
34 This documentation can also be found online at:
35 http://www.squid-cache.org/Doc/config/
37 You may wish to look at the Squid home page and wiki for the
38 FAQ and other documentation:
39 http://www.squid-cache.org/
40 http://wiki.squid-cache.org/SquidFaq
41 http://wiki.squid-cache.org/ConfigExamples
43 This documentation shows what the defaults for various directives
44 happen to be. If you don't need to change the default, you should
45 leave the line out of your squid.conf in most cases.
47 In some cases "none" refers to no default setting at all,
48 while in other cases it refers to the value of the option
49 - the comments for that keyword indicate if this is the case.
54 Configuration options can be included using the "include" directive.
55 Include takes a list of files to include. Quoting and wildcards are
60 include /path/to/included/file/squid.acl.config
62 Includes can be nested up to a hard-coded depth of 16 levels.
63 This arbitrary restriction is to prevent recursive include references
64 from causing Squid entering an infinite loop whilst trying to load
68 Conditional configuration
70 If-statements can be used to make configuration directives
74 ... regular configuration directives ...
76 ... regular configuration directives ...]
79 The else part is optional. The keywords "if", "else", and "endif"
80 must be typed on their own lines, as if they were regular
81 configuration directives.
83 NOTE: An else-if condition is not supported.
85 These individual conditions types are supported:
88 Always evaluates to true.
90 Always evaluates to false.
92 Equality comparison of two integer numbers.
97 The following SMP-related preprocessor macros can be used.
99 ${process_name} expands to the current Squid process "name"
100 (e.g., squid1, squid2, or cache1).
102 ${process_number} expands to the current Squid process
103 identifier, which is an integer number (e.g., 1, 2, 3) unique
104 across all Squid processes.
107 # Options Removed in 3.2
108 NAME: ignore_expect_100
111 Remove this line. The HTTP/1.1 feature is now fully supported by default.
114 NAME: dns_v4_fallback
123 Remove this line. Configure FTP page display using the CSS controls in errorpages.css instead.
126 NAME: url_rewrite_concurrency
129 Remove this line. Set the 'concurrency=' option of url_rewrite_children instead.
132 # Options Removed in 3.1
136 Remove this line. DNS is no longer tested on startup.
139 NAME: extension_methods
142 Remove this line. All valid methods for HTTP are accepted by default.
145 # 2.7 Options Removed/Replaced in 3.1
153 Remove this line. HTTP/1.1 is supported by default.
156 NAME: upgrade_http0.9
159 Remove this line. ICY/1.0 streaming protocol is supported by default.
162 NAME: zph_local zph_mode zph_option zph_parent zph_sibling
165 Alter these entries. Use the qos_flows directive instead.
168 # Options Removed in 3.0
172 Since squid-3.0 replace with request_header_access or reply_header_access
173 depending on whether you wish to match client requests or server replies.
176 NAME: httpd_accel_no_pmtu_disc
179 Since squid-3.0 use the 'disable-pmtu-discovery' flag on http_port instead.
183 OPTIONS FOR AUTHENTICATION
184 -----------------------------------------------------------------------------
193 This is used to define parameters for the various authentication
194 schemes supported by Squid.
196 format: auth_param scheme parameter [setting]
198 The order in which authentication schemes are presented to the client is
199 dependent on the order the scheme first appears in config file. IE
200 has a bug (it's not RFC 2617 compliant) in that it will use the basic
201 scheme if basic is the first entry presented, even if more secure
202 schemes are presented. For now use the order in the recommended
203 settings section below. If other browsers have difficulties (don't
204 recognize the schemes offered even if you are using basic) either
205 put basic first, or disable the other schemes (by commenting out their
208 Once an authentication scheme is fully configured, it can only be
209 shutdown by shutting squid down and restarting. Changes can be made on
210 the fly and activated with a reconfigure. I.E. You can change to a
211 different helper, but not unconfigure the helper completely.
213 Please note that while this directive defines how Squid processes
214 authentication it does not automatically activate authentication.
215 To use authentication you must in addition make use of ACLs based
216 on login name in http_access (proxy_auth, proxy_auth_regex or
217 external with %LOGIN used in the format tag). The browser will be
218 challenged for authentication on the first such acl encountered
219 in http_access processing and will also be re-challenged for new
220 login credentials if the request is being denied by a proxy_auth
223 WARNING: authentication can't be used in a transparently intercepting
224 proxy as the client then thinks it is talking to an origin server and
225 not the proxy. This is a limitation of bending the TCP/IP protocol to
226 transparently intercepting port 80, not a limitation in Squid.
227 Ports flagged 'transparent', 'intercept', or 'tproxy' have
228 authentication disabled.
230 === Parameters for the basic scheme follow. ===
233 Specify the command for the external authenticator. Such a program
234 reads a line containing "username password" and replies "OK" or
235 "ERR" in an endless loop. "ERR" responses may optionally be followed
236 by a error description available as %m in the returned error page.
237 If you use an authenticator, make sure you have 1 acl of type
240 By default, the basic authentication scheme is not used unless a
241 program is specified.
243 If you want to use the traditional NCSA proxy authentication, set
244 this line to something like
246 auth_param basic program @DEFAULT_PREFIX@/libexec/ncsa_auth @DEFAULT_PREFIX@/etc/passwd
249 HTTP uses iso-latin-1 as characterset, while some authentication
250 backends such as LDAP expects UTF-8. If this is set to on Squid will
251 translate the HTTP iso-latin-1 charset to UTF-8 before sending the
252 username & password to the helper.
254 "children" numberofchildren [startup=N] [idle=N] [concurrency=N]
255 The maximum number of authenticator processes to spawn. If you start too few
256 Squid will have to wait for them to process a backlog of credential
257 verifications, slowing it down. When password verifications are
258 done via a (slow) network you are likely to need lots of
259 authenticator processes.
261 The startup= and idle= options permit some skew in the exact amount
262 run. A minimum of startup=N will begin during startup and reconfigure.
263 Squid will start more in groups of up to idle=N in an attempt to meet
264 traffic needs and to keep idle=N free above those traffic needs up to
267 The concurrency= option sets the number of concurrent requests the
268 helper can process. The default of 0 is used for helpers who only
269 supports one request at a time. Setting this to a number greater than
270 0 changes the protocol used to include a channel number first on the
271 request/response line, allowing multiple requests to be sent to the
272 same helper in parallell without wating for the response.
273 Must not be set unless it's known the helper supports this.
275 auth_param basic children 20 startup=0 idle=1
278 Specifies the realm name which is to be reported to the
279 client for the basic proxy authentication scheme (part of
280 the text the user will see when prompted their username and
281 password). There is no default.
282 auth_param basic realm Squid proxy-caching web server
284 "credentialsttl" timetolive
285 Specifies how long squid assumes an externally validated
286 username:password pair is valid for - in other words how
287 often the helper program is called for that user. Set this
288 low to force revalidation with short lived passwords. Note
289 setting this high does not impact your susceptibility
290 to replay attacks unless you are using an one-time password
291 system (such as SecureID). If you are using such a system,
292 you will be vulnerable to replay attacks unless you also
293 use the max_user_ip ACL in an http_access rule.
295 "casesensitive" on|off
296 Specifies if usernames are case sensitive. Most user databases are
297 case insensitive allowing the same username to be spelled using both
298 lower and upper case letters, but some are case sensitive. This
299 makes a big difference for user_max_ip ACL processing and similar.
300 auth_param basic casesensitive off
302 === Parameters for the digest scheme follow ===
305 Specify the command for the external authenticator. Such
306 a program reads a line containing "username":"realm" and
307 replies with the appropriate H(A1) value hex encoded or
308 ERR if the user (or his H(A1) hash) does not exists.
309 See rfc 2616 for the definition of H(A1).
310 "ERR" responses may optionally be followed by a error description
311 available as %m in the returned error page.
313 By default, the digest authentication scheme is not used unless a
314 program is specified.
316 If you want to use a digest authenticator, set this line to
319 auth_param digest program @DEFAULT_PREFIX@/bin/digest_pw_auth @DEFAULT_PREFIX@/etc/digpass
322 HTTP uses iso-latin-1 as characterset, while some authentication
323 backends such as LDAP expects UTF-8. If this is set to on Squid will
324 translate the HTTP iso-latin-1 charset to UTF-8 before sending the
325 username & password to the helper.
327 "children" numberofchildren [startup=N] [idle=N] [concurrency=N]
328 The maximum number of authenticator processes to spawn (default 5).
329 If you start too few Squid will have to wait for them to
330 process a backlog of H(A1) calculations, slowing it down.
331 When the H(A1) calculations are done via a (slow) network
332 you are likely to need lots of authenticator processes.
334 The startup= and idle= options permit some skew in the exact amount
335 run. A minimum of startup=N will begin during startup and reconfigure.
336 Squid will start more in groups of up to idle=N in an attempt to meet
337 traffic needs and to keep idle=N free above those traffic needs up to
340 The concurrency= option sets the number of concurrent requests the
341 helper can process. The default of 0 is used for helpers who only
342 supports one request at a time. Setting this to a number greater than
343 0 changes the protocol used to include a channel number first on the
344 request/response line, allowing multiple requests to be sent to the
345 same helper in parallell without wating for the response.
346 Must not be set unless it's known the helper supports this.
348 auth_param digest children 20 startup=0 idle=1
351 Specifies the realm name which is to be reported to the
352 client for the digest proxy authentication scheme (part of
353 the text the user will see when prompted their username and
354 password). There is no default.
355 auth_param digest realm Squid proxy-caching web server
357 "nonce_garbage_interval" timeinterval
358 Specifies the interval that nonces that have been issued
359 to client_agent's are checked for validity.
361 "nonce_max_duration" timeinterval
362 Specifies the maximum length of time a given nonce will be
365 "nonce_max_count" number
366 Specifies the maximum number of times a given nonce can be
369 "nonce_strictness" on|off
370 Determines if squid requires strict increment-by-1 behavior
371 for nonce counts, or just incrementing (off - for use when
372 useragents generate nonce counts that occasionally miss 1
373 (ie, 1,2,4,6)). Default off.
375 "check_nonce_count" on|off
376 This directive if set to off can disable the nonce count check
377 completely to work around buggy digest qop implementations in
378 certain mainstream browser versions. Default on to check the
379 nonce count to protect from authentication replay attacks.
381 "post_workaround" on|off
382 This is a workaround to certain buggy browsers who sends
383 an incorrect request digest in POST requests when reusing
384 the same nonce as acquired earlier on a GET request.
386 === NTLM scheme options follow ===
389 Specify the command for the external NTLM authenticator.
390 Such a program reads exchanged NTLMSSP packets with
391 the browser via Squid until authentication is completed.
392 If you use an NTLM authenticator, make sure you have 1 acl
393 of type proxy_auth. By default, the NTLM authenticator_program
396 auth_param ntlm program @DEFAULT_PREFIX@/bin/ntlm_auth
398 "children" numberofchildren [startup=N] [idle=N]
399 The maximum number of authenticator processes to spawn (default 5).
400 If you start too few Squid will have to wait for them to
401 process a backlog of credential verifications, slowing it
402 down. When credential verifications are done via a (slow)
403 network you are likely to need lots of authenticator
406 The startup= and idle= options permit some skew in the exact amount
407 run. A minimum of startup=N will begin during startup and reconfigure.
408 Squid will start more in groups of up to idle=N in an attempt to meet
409 traffic needs and to keep idle=N free above those traffic needs up to
412 auth_param ntlm children 20 startup=0 idle=1
415 If you experience problems with PUT/POST requests when using the
416 Negotiate authentication scheme then you can try setting this to
417 off. This will cause Squid to forcibly close the connection on
418 the initial requests where the browser asks which schemes are
419 supported by the proxy.
421 auth_param ntlm keep_alive on
423 === Options for configuring the NEGOTIATE auth-scheme follow ===
426 Specify the command for the external Negotiate authenticator.
427 This protocol is used in Microsoft Active-Directory enabled setups with
428 the Microsoft Internet Explorer or Mozilla Firefox browsers.
429 Its main purpose is to exchange credentials with the Squid proxy
430 using the Kerberos mechanisms.
431 If you use a Negotiate authenticator, make sure you have at least
432 one acl of type proxy_auth active. By default, the negotiate
433 authenticator_program is not used.
434 The only supported program for this role is the ntlm_auth
435 program distributed as part of Samba, version 4 or later.
437 auth_param negotiate program @DEFAULT_PREFIX@/bin/ntlm_auth --helper-protocol=gss-spnego
439 "children" numberofchildren [startup=N] [idle=N]
440 The maximum number of authenticator processes to spawn (default 5).
441 If you start too few Squid will have to wait for them to
442 process a backlog of credential verifications, slowing it
443 down. When crendential verifications are done via a (slow)
444 network you are likely to need lots of authenticator
447 The startup= and idle= options permit some skew in the exact amount
448 run. A minimum of startup=N will begin during startup and reconfigure.
449 Squid will start more in groups of up to idle=N in an attempt to meet
450 traffic needs and to keep idle=N free above those traffic needs up to
453 auth_param negotiate children 20 startup=0 idle=1
456 If you experience problems with PUT/POST requests when using the
457 Negotiate authentication scheme then you can try setting this to
458 off. This will cause Squid to forcibly close the connection on
459 the initial requests where the browser asks which schemes are
460 supported by the proxy.
462 auth_param negotiate keep_alive on
467 #Recommended minimum configuration per scheme:
468 #auth_param negotiate program <uncomment and complete this line to activate>
469 #auth_param negotiate children 20 startup=0 idle=1
470 #auth_param negotiate keep_alive on
472 #auth_param ntlm program <uncomment and complete this line to activate>
473 #auth_param ntlm children 20 startup=0 idle=1
474 #auth_param ntlm keep_alive on
476 #auth_param digest program <uncomment and complete this line>
477 #auth_param digest children 20 startup=0 idle=1
478 #auth_param digest realm Squid proxy-caching web server
479 #auth_param digest nonce_garbage_interval 5 minutes
480 #auth_param digest nonce_max_duration 30 minutes
481 #auth_param digest nonce_max_count 50
483 #auth_param basic program <uncomment and complete this line>
484 #auth_param basic children 5 startup=5 idle=1
485 #auth_param basic realm Squid proxy-caching web server
486 #auth_param basic credentialsttl 2 hours
489 NAME: authenticate_cache_garbage_interval
492 LOC: Config.authenticateGCInterval
494 The time period between garbage collection across the username cache.
495 This is a tradeoff between memory utilization (long intervals - say
496 2 days) and CPU (short intervals - say 1 minute). Only change if you
500 NAME: authenticate_ttl
503 LOC: Config.authenticateTTL
505 The time a user & their credentials stay in the logged in
506 user cache since their last request. When the garbage
507 interval passes, all user credentials that have passed their
508 TTL are removed from memory.
511 NAME: authenticate_ip_ttl
513 LOC: Config.authenticateIpTTL
516 If you use proxy authentication and the 'max_user_ip' ACL,
517 this directive controls how long Squid remembers the IP
518 addresses associated with each user. Use a small value
519 (e.g., 60 seconds) if your users might change addresses
520 quickly, as is the case with dialups. You might be safe
521 using a larger value (e.g., 2 hours) in a corporate LAN
522 environment with relatively static address assignments.
527 -----------------------------------------------------------------------------
530 NAME: external_acl_type
531 TYPE: externalAclHelper
532 LOC: Config.externalAclHelperList
535 This option defines external acl classes using a helper program
536 to look up the status
538 external_acl_type name [options] FORMAT.. /path/to/helper [helper arguments..]
542 ttl=n TTL in seconds for cached results (defaults to 3600
545 TTL for cached negative lookups (default same
548 Maximum number of acl helper processes spawned to service
549 external acl lookups of this type. (default 20)
551 Minimum number of acl helper processes to spawn during
552 startup and reconfigure to service external acl lookups
553 of this type. (default 0)
555 Number of acl helper processes to keep ahead of traffic
556 loads. Squid will spawn this many at once whenever load
557 rises above the capabilities of existing processes.
558 Up to the value of children-max. (default 1)
559 concurrency=n concurrency level per process. Only used with helpers
560 capable of processing more than one query at a time.
561 cache=n limit the result cache size, default is unbounded.
562 grace=n Percentage remaining of TTL where a refresh of a
563 cached entry should be initiated without needing to
564 wait for a new reply. (default is for no grace period)
565 protocol=2.5 Compatibility mode for Squid-2.5 external acl helpers
566 ipv4 / ipv6 IP protocol used to communicate with this helper.
567 The default is to auto-detect IPv6 and use it when available.
569 FORMAT specifications
571 %LOGIN Authenticated user login name
572 %EXT_USER Username from previous external acl
573 %EXT_LOG Log details from previous external acl
574 %EXT_TAG Tag from previous external acl
575 %IDENT Ident user name
577 %SRCPORT Client source port
580 %PROTO Requested protocol
582 %PATH Requested URL path
583 %METHOD Request method
584 %MYADDR Squid interface address
585 %MYPORT Squid http_port number
586 %PATH Requested URL-path (including query-string if any)
587 %USER_CERT SSL User certificate in PEM format
588 %USER_CERTCHAIN SSL User certificate chain in PEM format
589 %USER_CERT_xx SSL User certificate subject attribute xx
590 %USER_CA_xx SSL User certificate issuer attribute xx
592 %>{Header} HTTP request header "Header"
594 HTTP request header "Hdr" list member "member"
596 HTTP request header list member using ; as
597 list separator. ; can be any non-alphanumeric
600 %<{Header} HTTP reply header "Header"
602 HTTP reply header "Hdr" list member "member"
604 HTTP reply header list member using ; as
605 list separator. ; can be any non-alphanumeric
608 %% The percent sign. Useful for helpers which need
609 an unchanging input format.
611 In addition to the above, any string specified in the referencing
612 acl will also be included in the helper request line, after the
613 specified formats (see the "acl external" directive)
615 The helper receives lines per the above format specification,
616 and returns lines starting with OK or ERR indicating the validity
617 of the request and optionally followed by additional keywords with
620 General result syntax:
622 OK/ERR keyword=value ...
626 user= The users name (login)
627 password= The users password (for login= cache_peer option)
628 message= Message describing the reason. Available as %o
630 tag= Apply a tag to a request (for both ERR and OK results)
631 Only sets a tag, does not alter existing tags.
632 log= String to be logged in access.log. Available as
633 %ea in logformat specifications
635 If protocol=3.0 (the default) then URL escaping is used to protect
636 each value in both requests and responses.
638 If using protocol=2.5 then all values need to be enclosed in quotes
639 if they may contain whitespace, or the whitespace escaped using \.
640 And quotes or \ characters within the keyword value must be \ escaped.
642 When using the concurrency= option the protocol is changed by
643 introducing a query channel tag infront of the request/response.
644 The query channel tag is a number between 0 and concurrency-1.
651 DEFAULT: ssl::certHasExpired ssl_error X509_V_ERR_CERT_HAS_EXPIRED
652 DEFAULT: ssl::certNotYetValid ssl_error X509_V_ERR_CERT_NOT_YET_VALID
653 DEFAULT: ssl::certDomainMismatch ssl_error SQUID_X509_V_ERR_DOMAIN_MISMATCH
654 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
655 DEFAULT: ssl::certSelfSigned ssl_error X509_V_ERR_DEPTH_ZERO_SELF_SIGNED_CERT
658 DEFAULT: manager url_regex -i ^cache_object:// +i ^https?://[^/]+/squid-internal-mgr/
659 DEFAULT: localhost src 127.0.0.1/32 ::1
660 DEFAULT: to_localhost dst 127.0.0.0/8 0.0.0.0/32 ::1
661 DEFAULT_DOC: ACLs all, manager, localhost, and to_localhost are predefined.
663 Defining an Access List
665 Every access list definition must begin with an aclname and acltype,
666 followed by either type-specific arguments or a quoted filename that
669 acl aclname acltype argument ...
670 acl aclname acltype "file" ...
672 When using "file", the file should contain one item per line.
674 By default, regular expressions are CASE-SENSITIVE.
675 To make them case-insensitive, use the -i option. To return case-sensitive
676 use the +i option between patterns, or make a new ACL line without -i.
678 Some acl types require suspending the current request in order
679 to access some external data source.
680 Those which do are marked with the tag [slow], those which
681 don't are marked as [fast].
682 See http://wiki.squid-cache.org/SquidFaq/SquidAcl
683 for further information
685 ***** ACL TYPES AVAILABLE *****
687 acl aclname src ip-address/mask ... # clients IP address [fast]
688 acl aclname src addr1-addr2/mask ... # range of addresses [fast]
689 acl aclname dst ip-address/mask ... # URL host's IP address [slow]
690 acl aclname localip ip-address/mask ... # IP address the client connected to [fast]
692 acl aclname arp mac-address ... (xx:xx:xx:xx:xx:xx notation)
693 # The arp ACL requires the special configure option --enable-arp-acl.
694 # Furthermore, the ARP ACL code is not portable to all operating systems.
695 # It works on Linux, Solaris, Windows, FreeBSD, and some
696 # other *BSD variants.
699 # NOTE: Squid can only determine the MAC address for clients that are on
700 # the same subnet. If the client is on a different subnet,
701 # then Squid cannot find out its MAC address.
703 acl aclname srcdomain .foo.com ...
704 # reverse lookup, from client IP [slow]
705 acl aclname dstdomain .foo.com ...
706 # Destination server from URL [fast]
707 acl aclname srcdom_regex [-i] \.foo\.com ...
708 # regex matching client name [slow]
709 acl aclname dstdom_regex [-i] \.foo\.com ...
710 # regex matching server [fast]
712 # For dstdomain and dstdom_regex a reverse lookup is tried if a IP
713 # based URL is used and no match is found. The name "none" is used
714 # if the reverse lookup fails.
716 acl aclname src_as number ...
717 acl aclname dst_as number ...
719 # Except for access control, AS numbers can be used for
720 # routing of requests to specific caches. Here's an
721 # example for routing all requests for AS#1241 and only
722 # those to mycache.mydomain.net:
723 # acl asexample dst_as 1241
724 # cache_peer_access mycache.mydomain.net allow asexample
725 # cache_peer_access mycache_mydomain.net deny all
727 acl aclname peername myPeer ...
729 # match against a named cache_peer entry
730 # set unique name= on cache_peer lines for reliable use.
732 acl aclname time [day-abbrevs] [h1:m1-h2:m2]
742 # h1:m1 must be less than h2:m2
744 acl aclname url_regex [-i] ^http:// ...
745 # regex matching on whole URL [fast]
746 acl aclname urlpath_regex [-i] \.gif$ ...
747 # regex matching on URL path [fast]
749 acl aclname port 80 70 21 0-1024... # destination TCP port [fast]
751 acl aclname localport 3128 ... # TCP port the client connected to [fast]
752 # NP: for interception mode this is usually '80'
754 acl aclname myportname 3128 ... # http(s)_port name [fast]
756 acl aclname proto HTTP FTP ... # request protocol [fast]
758 acl aclname method GET POST ... # HTTP request method [fast]
760 acl aclname http_status 200 301 500- 400-403 ...
761 # status code in reply [fast]
763 acl aclname browser [-i] regexp ...
764 # pattern match on User-Agent header (see also req_header below) [fast]
766 acl aclname referer_regex [-i] regexp ...
767 # pattern match on Referer header [fast]
768 # Referer is highly unreliable, so use with care
770 acl aclname ident username ...
771 acl aclname ident_regex [-i] pattern ...
772 # string match on ident output [slow]
773 # use REQUIRED to accept any non-null ident.
775 acl aclname proxy_auth [-i] username ...
776 acl aclname proxy_auth_regex [-i] pattern ...
777 # perform http authentication challenge to the client and match against
778 # supplied credentials [slow]
780 # takes a list of allowed usernames.
781 # use REQUIRED to accept any valid username.
783 # Will use proxy authentication in forward-proxy scenarios, and plain
784 # http authenticaiton in reverse-proxy scenarios
786 # NOTE: when a Proxy-Authentication header is sent but it is not
787 # needed during ACL checking the username is NOT logged
790 # NOTE: proxy_auth requires a EXTERNAL authentication program
791 # to check username/password combinations (see
792 # auth_param directive).
794 # NOTE: proxy_auth can't be used in a transparent/intercepting proxy
795 # as the browser needs to be configured for using a proxy in order
796 # to respond to proxy authentication.
798 acl aclname snmp_community string ...
799 # A community string to limit access to your SNMP Agent [fast]
802 # acl snmppublic snmp_community public
804 acl aclname maxconn number
805 # This will be matched when the client's IP address has
806 # more than <number> TCP connections established. [fast]
807 # NOTE: This only measures direct TCP links so X-Forwarded-For
808 # indirect clients are not counted.
810 acl aclname max_user_ip [-s] number
811 # This will be matched when the user attempts to log in from more
812 # than <number> different ip addresses. The authenticate_ip_ttl
813 # parameter controls the timeout on the ip entries. [fast]
814 # If -s is specified the limit is strict, denying browsing
815 # from any further IP addresses until the ttl has expired. Without
816 # -s Squid will just annoy the user by "randomly" denying requests.
817 # (the counter is reset each time the limit is reached and a
819 # NOTE: in acceleration mode or where there is mesh of child proxies,
820 # clients may appear to come from multiple addresses if they are
821 # going through proxy farms, so a limit of 1 may cause user problems.
823 acl aclname random probability
824 # Pseudo-randomly match requests. Based on the probability given.
825 # Probability may be written as a decimal (0.333), fraction (1/3)
826 # or ratio of matches:non-matches (3:5).
828 acl aclname req_mime_type [-i] mime-type ...
829 # regex match against the mime type of the request generated
830 # by the client. Can be used to detect file upload or some
831 # types HTTP tunneling requests [fast]
832 # NOTE: This does NOT match the reply. You cannot use this
833 # to match the returned file type.
835 acl aclname req_header header-name [-i] any\.regex\.here
836 # regex match against any of the known request headers. May be
837 # thought of as a superset of "browser", "referer" and "mime-type"
840 acl aclname rep_mime_type [-i] mime-type ...
841 # regex match against the mime type of the reply received by
842 # squid. Can be used to detect file download or some
843 # types HTTP tunneling requests. [fast]
844 # NOTE: This has no effect in http_access rules. It only has
845 # effect in rules that affect the reply data stream such as
848 acl aclname rep_header header-name [-i] any\.regex\.here
849 # regex match against any of the known reply headers. May be
850 # thought of as a superset of "browser", "referer" and "mime-type"
853 acl aclname external class_name [arguments...]
854 # external ACL lookup via a helper class defined by the
855 # external_acl_type directive [slow]
857 acl aclname user_cert attribute values...
858 # match against attributes in a user SSL certificate
859 # attribute is one of DN/C/O/CN/L/ST [fast]
861 acl aclname ca_cert attribute values...
862 # match against attributes a users issuing CA SSL certificate
863 # attribute is one of DN/C/O/CN/L/ST [fast]
865 acl aclname ext_user username ...
866 acl aclname ext_user_regex [-i] pattern ...
867 # string match on username returned by external acl helper [slow]
868 # use REQUIRED to accept any non-null user name.
870 acl aclname tag tagvalue ...
871 # string match on tag returned by external acl helper [slow]
873 acl aclname hier_code codename ...
874 # string match against squid hierarchy code(s); [fast]
875 # e.g., DIRECT, PARENT_HIT, NONE, etc.
877 # NOTE: This has no effect in http_access rules. It only has
878 # effect in rules that affect the reply data stream such as
882 acl aclname ssl_error errorname
883 # match against SSL certificate validation error [fast]
884 # For valid error names see in @DEFAULT_ERROR_DIR@/templates/error-details.txt template file
885 # The user aditionaly can use as error name the following error name
887 # [ssl::]certHasExpired: certificate "not after" field is in the past
888 # [ssl::]certNotYetValid: certificate "not before" field is in the
890 # [ssl::]certDomainMismatch: The certificate CN domain does not match
891 # connecting host name
892 # [ssl::]certUntrusted: The certificate is untrusted because of an
893 # error says that the certificate issuer is not trusted.
894 # [ssl::]certSelfSigned: The certificate is self signed
896 # The ssl::certHasExpired, ssl::certNotYetValid ssl::certDomainMismatch,
897 # ssl::certUntrusted and ssl::certSelfSigned also exists as predefined
900 # NOTE: The ssl_error acl has effect only when used with
901 # sslproxy_cert_error, sslproxy_cert_sign and sslproxy_cert_adapt
906 acl macaddress arp 09:00:2b:23:45:67
907 acl myexample dst_as 1241
908 acl password proxy_auth REQUIRED
909 acl fileupload req_mime_type -i ^multipart/form-data$
910 acl javascript rep_mime_type -i ^application/x-javascript$
914 # Recommended minimum configuration:
917 # Example rule allowing access from your local networks.
918 # Adapt to list your (internal) IP networks from where browsing
920 acl localnet src 10.0.0.0/8 # RFC1918 possible internal network
921 acl localnet src 172.16.0.0/12 # RFC1918 possible internal network
922 acl localnet src 192.168.0.0/16 # RFC1918 possible internal network
923 acl localnet src fc00::/7 # RFC 4193 local private network range
924 acl localnet src fe80::/10 # RFC 4291 link-local (directly plugged) machines
926 acl SSL_ports port 443
927 acl Safe_ports port 80 # http
928 acl Safe_ports port 21 # ftp
929 acl Safe_ports port 443 # https
930 acl Safe_ports port 70 # gopher
931 acl Safe_ports port 210 # wais
932 acl Safe_ports port 1025-65535 # unregistered ports
933 acl Safe_ports port 280 # http-mgmt
934 acl Safe_ports port 488 # gss-http
935 acl Safe_ports port 591 # filemaker
936 acl Safe_ports port 777 # multiling http
937 acl CONNECT method CONNECT
941 NAME: follow_x_forwarded_for
943 IFDEF: FOLLOW_X_FORWARDED_FOR
944 LOC: Config.accessList.followXFF
945 DEFAULT_IF_NONE: deny all
947 Allowing or Denying the X-Forwarded-For header to be followed to
948 find the original source of a request.
950 Requests may pass through a chain of several other proxies
951 before reaching us. The X-Forwarded-For header will contain a
952 comma-separated list of the IP addresses in the chain, with the
953 rightmost address being the most recent.
955 If a request reaches us from a source that is allowed by this
956 configuration item, then we consult the X-Forwarded-For header
957 to see where that host received the request from. If the
958 X-Forwarded-For header contains multiple addresses, we continue
959 backtracking until we reach an address for which we are not allowed
960 to follow the X-Forwarded-For header, or until we reach the first
961 address in the list. For the purpose of ACL used in the
962 follow_x_forwarded_for directive the src ACL type always matches
963 the address we are testing and srcdomain matches its rDNS.
965 The end result of this process is an IP address that we will
966 refer to as the indirect client address. This address may
967 be treated as the client address for access control, ICAP, delay
968 pools and logging, depending on the acl_uses_indirect_client,
969 icap_uses_indirect_client, delay_pool_uses_indirect_client,
970 log_uses_indirect_client and tproxy_uses_indirect_client options.
972 This clause only supports fast acl types.
973 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
975 SECURITY CONSIDERATIONS:
977 Any host for which we follow the X-Forwarded-For header
978 can place incorrect information in the header, and Squid
979 will use the incorrect information as if it were the
980 source address of the request. This may enable remote
981 hosts to bypass any access control restrictions that are
982 based on the client's source addresses.
986 acl localhost src 127.0.0.1
987 acl my_other_proxy srcdomain .proxy.example.com
988 follow_x_forwarded_for allow localhost
989 follow_x_forwarded_for allow my_other_proxy
992 NAME: acl_uses_indirect_client
995 IFDEF: FOLLOW_X_FORWARDED_FOR
997 LOC: Config.onoff.acl_uses_indirect_client
999 Controls whether the indirect client address
1000 (see follow_x_forwarded_for) is used instead of the
1001 direct client address in acl matching.
1003 NOTE: maxconn ACL considers direct TCP links and indirect
1004 clients will always have zero. So no match.
1007 NAME: delay_pool_uses_indirect_client
1010 IFDEF: FOLLOW_X_FORWARDED_FOR&&USE_DELAY_POOLS
1012 LOC: Config.onoff.delay_pool_uses_indirect_client
1014 Controls whether the indirect client address
1015 (see follow_x_forwarded_for) is used instead of the
1016 direct client address in delay pools.
1019 NAME: log_uses_indirect_client
1022 IFDEF: FOLLOW_X_FORWARDED_FOR
1024 LOC: Config.onoff.log_uses_indirect_client
1026 Controls whether the indirect client address
1027 (see follow_x_forwarded_for) is used instead of the
1028 direct client address in the access log.
1031 NAME: tproxy_uses_indirect_client
1034 IFDEF: FOLLOW_X_FORWARDED_FOR&&LINUX_NETFILTER
1036 LOC: Config.onoff.tproxy_uses_indirect_client
1038 Controls whether the indirect client address
1039 (see follow_x_forwarded_for) is used instead of the
1040 direct client address when spoofing the outgoing client.
1042 This has no effect on requests arriving in non-tproxy
1045 SECURITY WARNING: Usage of this option is dangerous
1046 and should not be used trivially. Correct configuration
1047 of follow_x_forewarded_for with a limited set of trusted
1048 sources is required to prevent abuse of your proxy.
1053 LOC: Config.accessList.http
1054 DEFAULT_IF_NONE: deny all
1056 Allowing or Denying access based on defined access lists
1058 Access to the HTTP port:
1059 http_access allow|deny [!]aclname ...
1061 NOTE on default values:
1063 If there are no "access" lines present, the default is to deny
1066 If none of the "access" lines cause a match, the default is the
1067 opposite of the last line in the list. If the last line was
1068 deny, the default is allow. Conversely, if the last line
1069 is allow, the default will be deny. For these reasons, it is a
1070 good idea to have an "deny all" entry at the end of your access
1071 lists to avoid potential confusion.
1073 This clause supports both fast and slow acl types.
1074 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1079 # Recommended minimum Access Permission configuration:
1081 # Only allow cachemgr access from localhost
1082 http_access allow localhost manager
1083 http_access deny manager
1085 # Deny requests to certain unsafe ports
1086 http_access deny !Safe_ports
1088 # Deny CONNECT to other than secure SSL ports
1089 http_access deny CONNECT !SSL_ports
1091 # We strongly recommend the following be uncommented to protect innocent
1092 # web applications running on the proxy server who think the only
1093 # one who can access services on "localhost" is a local user
1094 #http_access deny to_localhost
1097 # INSERT YOUR OWN RULE(S) HERE TO ALLOW ACCESS FROM YOUR CLIENTS
1100 # Example rule allowing access from your local networks.
1101 # Adapt localnet in the ACL section to list your (internal) IP networks
1102 # from where browsing should be allowed
1103 http_access allow localnet
1104 http_access allow localhost
1106 # And finally deny all other access to this proxy
1107 http_access deny all
1111 NAME: adapted_http_access http_access2
1113 LOC: Config.accessList.adapted_http
1116 Allowing or Denying access based on defined access lists
1118 Essentially identical to http_access, but runs after redirectors
1119 and ICAP/eCAP adaptation. Allowing access control based on their
1122 If not set then only http_access is used.
1125 NAME: http_reply_access
1127 LOC: Config.accessList.reply
1130 Allow replies to client requests. This is complementary to http_access.
1132 http_reply_access allow|deny [!] aclname ...
1134 NOTE: if there are no access lines present, the default is to allow
1137 If none of the access lines cause a match the opposite of the
1138 last line will apply. Thus it is good practice to end the rules
1139 with an "allow all" or "deny all" entry.
1141 This clause supports both fast and slow acl types.
1142 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1147 LOC: Config.accessList.icp
1148 DEFAULT_IF_NONE: deny all
1150 Allowing or Denying access to the ICP port based on defined
1153 icp_access allow|deny [!]aclname ...
1155 See http_access for details
1157 This clause only supports fast acl types.
1158 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1160 # Allow ICP queries from local networks only
1161 #icp_access allow localnet
1162 #icp_access deny all
1168 LOC: Config.accessList.htcp
1169 DEFAULT_IF_NONE: deny all
1171 Allowing or Denying access to the HTCP port based on defined
1174 htcp_access allow|deny [!]aclname ...
1176 See http_access for details
1178 NOTE: The default if no htcp_access lines are present is to
1179 deny all traffic. This default may cause problems with peers
1180 using the htcp option.
1182 This clause only supports fast acl types.
1183 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1185 # Allow HTCP queries from local networks only
1186 #htcp_access allow localnet
1187 #htcp_access deny all
1190 NAME: htcp_clr_access
1193 LOC: Config.accessList.htcp_clr
1194 DEFAULT_IF_NONE: deny all
1196 Allowing or Denying access to purge content using HTCP based
1197 on defined access lists
1199 htcp_clr_access allow|deny [!]aclname ...
1201 See http_access for details
1203 This clause only supports fast acl types.
1204 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1206 # Allow HTCP CLR requests from trusted peers
1207 acl htcp_clr_peer src 172.16.1.2
1208 htcp_clr_access allow htcp_clr_peer
1213 LOC: Config.accessList.miss
1216 Determins whether network access is permitted when satisfying a request.
1219 to force your neighbors to use you as a sibling instead of
1222 acl localclients src 172.16.0.0/16
1223 miss_access allow localclients
1224 miss_access deny !localclients
1226 This means only your local clients are allowed to fetch relayed/MISS
1227 replies from the network and all other clients can only fetch cached
1231 The default for this setting allows all clients who passed the
1232 http_access rules to relay via this proxy.
1234 This clause only supports fast acl types.
1235 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1238 NAME: ident_lookup_access
1241 DEFAULT_IF_NONE: deny all
1242 LOC: Ident::TheConfig.identLookup
1244 A list of ACL elements which, if matched, cause an ident
1245 (RFC 931) lookup to be performed for this request. For
1246 example, you might choose to always perform ident lookups
1247 for your main multi-user Unix boxes, but not for your Macs
1248 and PCs. By default, ident lookups are not performed for
1251 To enable ident lookups for specific client addresses, you
1252 can follow this example:
1254 acl ident_aware_hosts src 198.168.1.0/24
1255 ident_lookup_access allow ident_aware_hosts
1256 ident_lookup_access deny all
1258 Only src type ACL checks are fully supported. A srcdomain
1259 ACL might work at times, but it will not always provide
1262 This clause only supports fast acl types.
1263 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
1266 NAME: reply_body_max_size
1267 COMMENT: size [acl acl...]
1270 LOC: Config.ReplyBodySize
1272 This option specifies the maximum size of a reply body. It can be
1273 used to prevent users from downloading very large files, such as
1274 MP3's and movies. When the reply headers are received, the
1275 reply_body_max_size lines are processed, and the first line where
1276 all (if any) listed ACLs are true is used as the maximum body size
1279 This size is checked twice. First when we get the reply headers,
1280 we check the content-length value. If the content length value exists
1281 and is larger than the allowed size, the request is denied and the
1282 user receives an error message that says "the request or reply
1283 is too large." If there is no content-length, and the reply
1284 size exceeds this limit, the client's connection is just closed
1285 and they will receive a partial reply.
1287 WARNING: downstream caches probably can not detect a partial reply
1288 if there is no content-length header, so they will cache
1289 partial responses and give them out as hits. You should NOT
1290 use this option if you have downstream caches.
1292 WARNING: A maximum size smaller than the size of squid's error messages
1293 will cause an infinite loop and crash squid. Ensure that the smallest
1294 non-zero value you use is greater that the maximum header size plus
1295 the size of your largest error page.
1297 If you set this parameter none (the default), there will be
1300 Configuration Format is:
1301 reply_body_max_size SIZE UNITS [acl ...]
1303 reply_body_max_size 10 MB
1309 -----------------------------------------------------------------------------
1312 NAME: http_port ascii_port
1313 TYPE: http_port_list
1315 LOC: Config.Sockaddr.http
1317 Usage: port [mode] [options]
1318 hostname:port [mode] [options]
1319 1.2.3.4:port [mode] [options]
1321 The socket addresses where Squid will listen for HTTP client
1322 requests. You may specify multiple socket addresses.
1323 There are three forms: port alone, hostname with port, and
1324 IP address with port. If you specify a hostname or IP
1325 address, Squid binds the socket to that specific
1326 address. Most likely, you do not need to bind to a specific
1327 address, so you can use the port number alone.
1329 If you are running Squid in accelerator mode, you
1330 probably want to listen on port 80 also, or instead.
1332 The -a command line option may be used to specify additional
1333 port(s) where Squid listens for proxy request. Such ports will
1334 be plain proxy ports with no options.
1336 You may specify multiple socket addresses on multiple lines.
1340 intercept Support for IP-Layer interception of
1341 outgoing requests without browser settings.
1342 NP: disables authentication and IPv6 on the port.
1344 tproxy Support Linux TPROXY for spoofing outgoing
1345 connections using the client IP address.
1346 NP: disables authentication and maybe IPv6 on the port.
1348 accel Accelerator / reverse proxy mode
1350 ssl-bump For each CONNECT request allowed by ssl_bump ACLs,
1351 establish secure connection with the client and with
1352 the server, decrypt HTTPS messages as they pass through
1353 Squid, and treat them as unencrypted HTTP messages,
1354 becoming the man-in-the-middle.
1356 The "ssl_bump" option is required to fully enable
1357 bumping of CONNECT requests.
1359 Omitting the mode flag causes default forward proxy mode to be used.
1362 Accelerator Mode Options:
1364 defaultsite=domainname
1365 What to use for the Host: header if it is not present
1366 in a request. Determines what site (not origin server)
1367 accelerators should consider the default.
1369 no-vhost Disable using HTTP/1.1 Host header for virtual domain support.
1371 protocol= Protocol to reconstruct accelerated requests with.
1372 Defaults to http for http_port and https for
1375 vport Virtual host port support. Using the http_port number
1376 instead of the port passed on Host: headers.
1378 vport=NN Virtual host port support. Using the specified port
1379 number instead of the port passed on Host: headers.
1382 Act as if this Squid is the origin server.
1383 This currently means generate new Date: and Expires:
1384 headers on HIT instead of adding Age:.
1386 ignore-cc Ignore request Cache-Control headers.
1388 WARNING: This option violates HTTP specifications if
1389 used in non-accelerator setups.
1391 allow-direct Allow direct forwarding in accelerator mode. Normally
1392 accelerated requests are denied direct forwarding as if
1393 never_direct was used.
1395 WARNING: this option opens accelerator mode to security
1396 vulnerabilities usually only affecting in interception
1397 mode. Make sure to protect forwarding with suitable
1398 http_access rules when using this.
1401 SSL Bump Mode Options:
1402 In addition to these options ssl-bump requires TLS/SSL options.
1404 generate-host-certificates[=<on|off>]
1405 Dynamically create SSL server certificates for the
1406 destination hosts of bumped CONNECT requests.When
1407 enabled, the cert and key options are used to sign
1408 generated certificates. Otherwise generated
1409 certificate will be selfsigned.
1410 If there is a CA certificate lifetime of the generated
1411 certificate equals lifetime of the CA certificate. If
1412 generated certificate is selfsigned lifetime is three
1414 This option is enabled by default when ssl-bump is used.
1415 See the ssl-bump option above for more information.
1417 dynamic_cert_mem_cache_size=SIZE
1418 Approximate total RAM size spent on cached generated
1419 certificates. If set to zero, caching is disabled. The
1420 default value is 4MB. An average XXX-bit certificate
1421 consumes about XXX bytes of RAM.
1425 cert= Path to SSL certificate (PEM format).
1427 key= Path to SSL private key file (PEM format)
1428 if not specified, the certificate file is
1429 assumed to be a combined certificate and
1432 version= The version of SSL/TLS supported
1433 1 automatic (default)
1438 cipher= Colon separated list of supported ciphers.
1439 NOTE: some ciphers such as EDH ciphers depend on
1440 additional settings. If those settings are
1441 omitted the ciphers may be silently ignored
1442 by the OpenSSL library.
1444 options= Various SSL implementation options. The most important
1446 NO_SSLv2 Disallow the use of SSLv2
1447 NO_SSLv3 Disallow the use of SSLv3
1448 NO_TLSv1 Disallow the use of TLSv1
1449 SINGLE_DH_USE Always create a new key when using
1450 temporary/ephemeral DH key exchanges
1451 ALL Enable various bug workarounds
1452 suggested as "harmless" by OpenSSL
1453 Be warned that this reduces SSL/TLS
1454 strength to some attacks.
1455 See OpenSSL SSL_CTX_set_options documentation for a
1456 complete list of options.
1458 clientca= File containing the list of CAs to use when
1459 requesting a client certificate.
1461 cafile= File containing additional CA certificates to
1462 use when verifying client certificates. If unset
1463 clientca will be used.
1465 capath= Directory containing additional CA certificates
1466 and CRL lists to use when verifying client certificates.
1468 crlfile= File of additional CRL lists to use when verifying
1469 the client certificate, in addition to CRLs stored in
1470 the capath. Implies VERIFY_CRL flag below.
1472 dhparams= File containing DH parameters for temporary/ephemeral
1473 DH key exchanges. See OpenSSL documentation for details
1474 on how to create this file.
1475 WARNING: EDH ciphers will be silently disabled if this
1478 sslflags= Various flags modifying the use of SSL:
1480 Don't request client certificates
1481 immediately, but wait until acl processing
1482 requires a certificate (not yet implemented).
1484 Don't use the default CA lists built in
1487 Don't allow for session reuse. Each connection
1488 will result in a new SSL session.
1490 Verify CRL lists when accepting client
1493 Verify CRL lists for all certificates in the
1494 client certificate chain.
1496 sslcontext= SSL session ID context identifier.
1500 connection-auth[=on|off]
1501 use connection-auth=off to tell Squid to prevent
1502 forwarding Microsoft connection oriented authentication
1503 (NTLM, Negotiate and Kerberos)
1505 disable-pmtu-discovery=
1506 Control Path-MTU discovery usage:
1507 off lets OS decide on what to do (default).
1508 transparent disable PMTU discovery when transparent
1510 always disable always PMTU discovery.
1512 In many setups of transparently intercepting proxies
1513 Path-MTU discovery can not work on traffic towards the
1514 clients. This is the case when the intercepting device
1515 does not fully track connections and fails to forward
1516 ICMP must fragment messages to the cache server. If you
1517 have such setup and experience that certain clients
1518 sporadically hang or never complete requests set
1519 disable-pmtu-discovery option to 'transparent'.
1521 name= Specifies a internal name for the port. Defaults to
1522 the port specification (port or addr:port)
1524 tcpkeepalive[=idle,interval,timeout]
1525 Enable TCP keepalive probes of idle connections.
1526 In seconds; idle is the initial time before TCP starts
1527 probing the connection, interval how often to probe, and
1528 timeout the time before giving up.
1530 If you run Squid on a dual-homed machine with an internal
1531 and an external interface we recommend you to specify the
1532 internal address:port in http_port. This way Squid will only be
1533 visible on the internal address.
1537 # Squid normally listens to port 3128
1538 http_port @DEFAULT_HTTP_PORT@
1544 TYPE: https_port_list
1546 LOC: Config.Sockaddr.https
1548 Usage: [ip:]port cert=certificate.pem [key=key.pem] [mode] [options...]
1550 The socket address where Squid will listen for client requests made
1551 over TLS or SSL connections. Commonly referred to as HTTPS.
1553 This is most useful for situations where you are running squid in
1554 accelerator mode and you want to do the SSL work at the accelerator level.
1556 You may specify multiple socket addresses on multiple lines,
1557 each with their own SSL certificate and/or options.
1561 accel Accelerator / reverse proxy mode
1563 tproxy Support Linux TPROXY for spoofing outgoing
1564 connections using the client IP address.
1565 NP: disables authentication and maybe IPv6 on the port.
1567 ssl-bump For each intercepted connection allowed by ssl_bump
1568 ACLs, establish a secure connection with the client and with
1569 the server, decrypt HTTPS messages as they pass through
1570 Squid, and treat them as unencrypted HTTP messages,
1571 becoming the man-in-the-middle.
1573 An "ssl_bump server-first" match is required to
1574 fully enable bumping of intercepted SSL connections.
1578 Omitting the mode flag causes default forward proxy mode to be used.
1581 See http_port for a list of generic options
1586 cert= Path to SSL certificate (PEM format).
1588 key= Path to SSL private key file (PEM format)
1589 if not specified, the certificate file is
1590 assumed to be a combined certificate and
1593 version= The version of SSL/TLS supported
1594 1 automatic (default)
1599 cipher= Colon separated list of supported ciphers.
1601 options= Various SSL engine options. The most important
1603 NO_SSLv2 Disallow the use of SSLv2
1604 NO_SSLv3 Disallow the use of SSLv3
1605 NO_TLSv1 Disallow the use of TLSv1
1606 SINGLE_DH_USE Always create a new key when using
1607 temporary/ephemeral DH key exchanges
1608 See src/ssl_support.c or OpenSSL SSL_CTX_set_options
1609 documentation for a complete list of options.
1611 clientca= File containing the list of CAs to use when
1612 requesting a client certificate.
1614 cafile= File containing additional CA certificates to
1615 use when verifying client certificates. If unset
1616 clientca will be used.
1618 capath= Directory containing additional CA certificates
1619 and CRL lists to use when verifying client certificates.
1621 crlfile= File of additional CRL lists to use when verifying
1622 the client certificate, in addition to CRLs stored in
1623 the capath. Implies VERIFY_CRL flag below.
1625 dhparams= File containing DH parameters for temporary/ephemeral
1628 sslflags= Various flags modifying the use of SSL:
1630 Don't request client certificates
1631 immediately, but wait until acl processing
1632 requires a certificate (not yet implemented).
1634 Don't use the default CA lists built in
1637 Don't allow for session reuse. Each connection
1638 will result in a new SSL session.
1640 Verify CRL lists when accepting client
1643 Verify CRL lists for all certificates in the
1644 client certificate chain.
1646 sslcontext= SSL session ID context identifier.
1648 generate-host-certificates[=<on|off>]
1649 Dynamically create SSL server certificates for the
1650 destination hosts of bumped SSL requests.When
1651 enabled, the cert and key options are used to sign
1652 generated certificates. Otherwise generated
1653 certificate will be selfsigned.
1654 If there is CA certificate life time of generated
1655 certificate equals lifetime of CA certificate. If
1656 generated certificate is selfsigned lifetime is three
1658 This option is enabled by default when SslBump is used.
1659 See the sslBump option above for more information.
1661 dynamic_cert_mem_cache_size=SIZE
1662 Approximate total RAM size spent on cached generated
1663 certificates. If set to zero, caching is disabled. The
1664 default value is 4MB. An average XXX-bit certificate
1665 consumes about XXX bytes of RAM.
1667 See http_port for a list of available options.
1670 NAME: tcp_outgoing_tos tcp_outgoing_ds tcp_outgoing_dscp
1673 LOC: Ip::Qos::TheConfig.tosToServer
1675 Allows you to select a TOS/Diffserv value for packets outgoing
1676 on the server side, based on an ACL.
1678 tcp_outgoing_tos ds-field [!]aclname ...
1680 Example where normal_service_net uses the TOS value 0x00
1681 and good_service_net uses 0x20
1683 acl normal_service_net src 10.0.0.0/24
1684 acl good_service_net src 10.0.1.0/24
1685 tcp_outgoing_tos 0x00 normal_service_net
1686 tcp_outgoing_tos 0x20 good_service_net
1688 TOS/DSCP values really only have local significance - so you should
1689 know what you're specifying. For more information, see RFC2474,
1690 RFC2475, and RFC3260.
1692 The TOS/DSCP byte must be exactly that - a octet value 0 - 255, or
1693 "default" to use whatever default your host has. Note that in
1694 practice often only multiples of 4 is usable as the two rightmost bits
1695 have been redefined for use by ECN (RFC 3168 section 23.1).
1697 Processing proceeds in the order specified, and stops at first fully
1701 NAME: clientside_tos
1704 LOC: Ip::Qos::TheConfig.tosToClient
1706 Allows you to select a TOS/Diffserv value for packets being transmitted
1707 on the client-side, based on an ACL.
1709 clientside_tos ds-field [!]aclname ...
1711 Example where normal_service_net uses the TOS value 0x00
1712 and good_service_net uses 0x20
1714 acl normal_service_net src 10.0.0.0/24
1715 acl good_service_net src 10.0.1.0/24
1716 clientside_tos 0x00 normal_service_net
1717 clientside_tos 0x20 good_service_net
1719 Note: This feature is incompatible with qos_flows. Any TOS values set here
1720 will be overwritten by TOS values in qos_flows.
1723 NAME: tcp_outgoing_mark
1725 IFDEF: SO_MARK&&USE_LIBCAP
1727 LOC: Ip::Qos::TheConfig.nfmarkToServer
1729 Allows you to apply a Netfilter mark value to outgoing packets
1730 on the server side, based on an ACL.
1732 tcp_outgoing_mark mark-value [!]aclname ...
1734 Example where normal_service_net uses the mark value 0x00
1735 and good_service_net uses 0x20
1737 acl normal_service_net src 10.0.0.0/24
1738 acl good_service_net src 10.0.1.0/24
1739 tcp_outgoing_mark 0x00 normal_service_net
1740 tcp_outgoing_mark 0x20 good_service_net
1743 NAME: clientside_mark
1745 IFDEF: SO_MARK&&USE_LIBCAP
1747 LOC: Ip::Qos::TheConfig.nfmarkToClient
1749 Allows you to apply a Netfilter mark value to packets being transmitted
1750 on the client-side, based on an ACL.
1752 clientside_mark mark-value [!]aclname ...
1754 Example where normal_service_net uses the mark value 0x00
1755 and good_service_net uses 0x20
1757 acl normal_service_net src 10.0.0.0/24
1758 acl good_service_net src 10.0.1.0/24
1759 clientside_mark 0x00 normal_service_net
1760 clientside_mark 0x20 good_service_net
1762 Note: This feature is incompatible with qos_flows. Any mark values set here
1763 will be overwritten by mark values in qos_flows.
1770 LOC: Ip::Qos::TheConfig
1772 Allows you to select a TOS/DSCP value to mark outgoing
1773 connections with, based on where the reply was sourced. For
1774 platforms using netfilter, allows you to set a netfilter mark
1775 value instead of, or in addition to, a TOS value.
1777 TOS values really only have local significance - so you should
1778 know what you're specifying. For more information, see RFC2474,
1779 RFC2475, and RFC3260.
1781 The TOS/DSCP byte must be exactly that - a octet value 0 - 255. Note that
1782 in practice often only multiples of 4 is usable as the two rightmost bits
1783 have been redefined for use by ECN (RFC 3168 section 23.1).
1785 Mark values can be any unsigned 32-bit integer value.
1787 This setting is configured by setting the following values:
1789 tos|mark Whether to set TOS or netfilter mark values
1791 local-hit=0xFF Value to mark local cache hits.
1793 sibling-hit=0xFF Value to mark hits from sibling peers.
1795 parent-hit=0xFF Value to mark hits from parent peers.
1797 miss=0xFF[/mask] Value to mark cache misses. Takes precedence
1798 over the preserve-miss feature (see below), unless
1799 mask is specified, in which case only the bits
1800 specified in the mask are written.
1802 The TOS variant of the following features are only possible on Linux
1803 and require your kernel to be patched with the TOS preserving ZPH
1804 patch, available from http://zph.bratcheda.org
1805 No patch is needed to preserve the netfilter mark, which will work
1806 with all variants of netfilter.
1808 disable-preserve-miss
1809 This option disables the preservation of the TOS or netfilter
1810 mark. By default, the existing TOS or netfilter mark value of
1811 the response coming from the remote server will be retained
1812 and masked with miss-mark.
1813 NOTE: in the case of a netfilter mark, the mark must be set on
1814 the connection (using the CONNMARK target) not on the packet
1818 Allows you to mask certain bits in the TOS or mark value
1819 received from the remote server, before copying the value to
1820 the TOS sent towards clients.
1821 Default for tos: 0xFF (TOS from server is not changed).
1822 Default for mark: 0xFFFFFFFF (mark from server is not changed).
1824 All of these features require the --enable-zph-qos compilation flag
1825 (enabled by default). Netfilter marking also requires the
1826 libnetfilter_conntrack libraries (--with-netfilter-conntrack) and
1827 libcap 2.09+ (--with-libcap).
1831 NAME: tcp_outgoing_address
1834 LOC: Config.accessList.outgoing_address
1836 Allows you to map requests to different outgoing IP addresses
1837 based on the username or source address of the user making
1840 tcp_outgoing_address ipaddr [[!]aclname] ...
1843 Forwarding clients with dedicated IPs for certain subnets.
1845 acl normal_service_net src 10.0.0.0/24
1846 acl good_service_net src 10.0.2.0/24
1848 tcp_outgoing_address 2001:db8::c001 good_service_net
1849 tcp_outgoing_address 10.1.0.2 good_service_net
1851 tcp_outgoing_address 2001:db8::beef normal_service_net
1852 tcp_outgoing_address 10.1.0.1 normal_service_net
1854 tcp_outgoing_address 2001:db8::1
1855 tcp_outgoing_address 10.1.0.3
1857 Processing proceeds in the order specified, and stops at first fully
1860 Squid will add an implicit IP version test to each line.
1861 Requests going to IPv4 websites will use the outgoing 10.1.0.* addresses.
1862 Requests going to IPv6 websites will use the outgoing 2001:db8:* addresses.
1865 NOTE: The use of this directive using client dependent ACLs is
1866 incompatible with the use of server side persistent connections. To
1867 ensure correct results it is best to set server_persistent_connections
1868 to off when using this directive in such configurations.
1870 NOTE: The use of this directive to set a local IP on outgoing TCP links
1871 is incompatible with using TPROXY to set client IP out outbound TCP links.
1872 When needing to contact peers use the no-tproxy cache_peer option and the
1873 client_dst_passthru directive re-enable normal forwarding such as this.
1877 NAME: host_verify_strict
1880 LOC: Config.onoff.hostStrictVerify
1882 Regardless of this option setting, when dealing with intercepted
1883 traffic, Squid always verifies that the destination IP address matches
1884 the Host header domain or IP (called 'authority form URL').
1886 This enforcement is performed to satisfy a MUST-level requirement in
1887 RFC 2616 section 14.23: "The Host field value MUST represent the naming
1888 authority of the origin server or gateway given by the original URL".
1891 Squid always responds with an HTTP 409 (Conflict) error
1892 page and logs a security warning if there is no match.
1894 Squid verifies that the destination IP address matches
1895 the Host header for forward-proxy and reverse-proxy traffic
1896 as well. For those traffic types, Squid also enables the
1897 following checks, comparing the corresponding Host header
1898 and Request-URI components:
1900 * The host names (domain or IP) must be identical,
1901 but valueless or missing Host header disables all checks.
1902 For the two host names to match, both must be either IP
1905 * Port numbers must be identical, but if a port is missing
1906 the scheme-default port is assumed.
1909 When set to OFF (the default):
1910 Squid allows suspicious requests to continue but logs a
1911 security warning and blocks caching of the response.
1913 * Forward-proxy traffic is not checked at all.
1915 * Reverse-proxy traffic is not checked at all.
1917 * Intercepted traffic which passes verification is handled
1920 For now it also forces suspicious requests to go DIRECT to the
1921 original destination, overriding client_dst_passthru for
1922 intercepted requests which fail Host: verification.
1924 For now suspicious intercepted CONNECT requests are always
1925 responded to with an HTTP 409 (Conflict) error page.
1928 NAME: client_dst_passthru
1931 LOC: Config.onoff.client_dst_passthru
1933 With NAT or TPROXY intercepted traffic Squid may pass the request
1934 directly to the original client destination IP or seek a faster
1937 This option (on by default) prevents cache_peer and alternative DNS
1938 entries being used on intercepted traffic. Both of which lead to
1939 the security vulnerability outlined below.
1943 This directive should only be disabled if cache_peer are required.
1945 As described in CVE-2009-0801 when the Host: header alone is used
1946 to determine the destination of a request it becomes trivial for
1947 malicious scripts on remote websites to bypass browser same-origin
1948 security policy and sandboxing protections.
1950 The cause of this is that such applets are allowed to perform their
1951 own HTTP stack, in which case the same-origin policy of the browser
1952 sandbox only verifies that the applet tries to contact the same IP
1953 as from where it was loaded at the IP level. The Host: header may
1954 be different from the connected IP and approved origin.
1960 -----------------------------------------------------------------------------
1963 NAME: ssl_unclean_shutdown
1967 LOC: Config.SSL.unclean_shutdown
1969 Some browsers (especially MSIE) bugs out on SSL shutdown
1976 LOC: Config.SSL.ssl_engine
1979 The OpenSSL engine to use. You will need to set this if you
1980 would like to use hardware SSL acceleration for example.
1983 NAME: sslproxy_client_certificate
1986 LOC: Config.ssl_client.cert
1989 Client SSL Certificate to use when proxying https:// URLs
1992 NAME: sslproxy_client_key
1995 LOC: Config.ssl_client.key
1998 Client SSL Key to use when proxying https:// URLs
2001 NAME: sslproxy_version
2004 LOC: Config.ssl_client.version
2007 SSL version level to use when proxying https:// URLs
2010 NAME: sslproxy_options
2013 LOC: Config.ssl_client.options
2016 SSL implementation options to use when proxying https:// URLs
2018 The most important being:
2020 NO_SSLv2 Disallow the use of SSLv2
2021 NO_SSLv3 Disallow the use of SSLv3
2022 NO_TLSv1 Disallow the use of TLSv1
2024 Always create a new key when using temporary/ephemeral
2027 Disable use of RFC5077 session tickets. Some servers
2028 may have problems understanding the TLS extension due
2029 to ambiguous specification in RFC4507.
2030 ALL Enable various bug workarounds suggested as "harmless"
2031 by OpenSSL. Be warned that this may reduce SSL/TLS
2032 strength to some attacks.
2034 See the OpenSSL SSL_CTX_set_options documentation for a
2035 complete list of possible options.
2038 NAME: sslproxy_cipher
2041 LOC: Config.ssl_client.cipher
2044 SSL cipher list to use when proxying https:// URLs
2046 Colon separated list of supported ciphers.
2049 NAME: sslproxy_cafile
2052 LOC: Config.ssl_client.cafile
2055 file containing CA certificates to use when verifying server
2056 certificates while proxying https:// URLs
2059 NAME: sslproxy_capath
2062 LOC: Config.ssl_client.capath
2065 directory containing CA certificates to use when verifying
2066 server certificates while proxying https:// URLs
2071 TYPE: sslproxy_ssl_bump
2072 LOC: Config.accessList.ssl_bump
2075 This option is consulted when a CONNECT request is received on
2076 an http_port (or a new connection is intercepted at an
2077 https_port), provided that port was configured with an ssl-bump
2078 flag. The subsequent data on the connection is either treated as
2079 HTTPS and decrypted OR tunneled at TCP level without decryption,
2080 depending on the first bumping "mode" which ACLs match.
2082 ssl_bump <mode> [!]acl ...
2084 The following bumping modes are supported:
2087 Allow bumping of the connection. Establish a secure connection
2088 with the client first, then connect to the server. This old mode
2089 does not allow Squid to mimic server SSL certificate and does
2090 not work with intercepted SSL connections.
2093 Allow bumping of the connection. Establish a secure connection
2094 with the server first, then establish a secure connection with
2095 the client, using a mimicked server certificate. Works with both
2096 CONNECT requests and intercepted SSL connections.
2099 Become a TCP tunnel without decoding the connection.
2100 Works with both CONNECT requests and intercepted SSL
2101 connections. This is the default behavior when no
2102 ssl_bump option is given or no ssl_bump ACLs match.
2104 By default, no connections are bumped.
2106 The first matching ssl_bump option wins. If no ACLs match, the
2107 connection is not bumped. Unlike most allow/deny ACL lists, ssl_bump
2108 does not have an implicit "negate the last given option" rule. You
2109 must make that rule explicit if you convert old ssl_bump allow/deny
2110 rules that rely on such an implicit rule.
2112 This clause supports both fast and slow acl types.
2113 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2115 See also: http_port ssl-bump, https_port ssl-bump
2118 # Example: Bump all requests except those originating from
2119 # localhost and those going to example.com.
2121 acl broken_sites dstdomain .example.com
2122 ssl_bump none localhost
2123 ssl_bump none broken_sites
2124 ssl_bump server-first all
2127 NAME: sslproxy_flags
2130 LOC: Config.ssl_client.flags
2133 Various flags modifying the use of SSL while proxying https:// URLs:
2134 DONT_VERIFY_PEER Accept certificates that fail verification.
2135 For refined control, see sslproxy_cert_error.
2136 NO_DEFAULT_CA Don't use the default CA list built in
2140 NAME: sslproxy_cert_error
2143 LOC: Config.ssl_client.cert_error
2146 Use this ACL to bypass server certificate validation errors.
2148 For example, the following lines will bypass all validation errors
2149 when talking to servers located at 172.16.0.0/16. All other
2150 validation errors will result in ERR_SECURE_CONNECT_FAIL error.
2152 acl BrokenServersAtTrustedIP dst 172.16.0.0/16
2153 sslproxy_cert_error allow BrokenServersAtTrustedIP
2154 sslproxy_cert_error deny all
2156 This clause only supports fast acl types.
2157 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
2158 Using slow acl types may result in server crashes
2160 Without this option, all server certificate validation errors
2161 terminate the transaction. Bypassing validation errors is dangerous
2162 because an error usually implies that the server cannot be trusted and
2163 the connection may be insecure.
2165 See also: sslproxy_flags and DONT_VERIFY_PEER.
2167 Default setting: sslproxy_cert_error deny all
2170 NAME: sslproxy_cert_sign
2173 POSTSCRIPTUM: signUntrusted ssl::certUntrusted
2174 POSTSCRIPTUM: signSelf ssl::certSelfSigned
2175 POSTSCRIPTUM: signTrusted all
2176 TYPE: sslproxy_cert_sign
2177 LOC: Config.ssl_client.cert_sign
2180 sslproxy_cert_sign <signing algorithm> acl ...
2182 The following certificate signing algorithms are supported:
2184 Sign using the configured CA certificate which is usually
2185 placed in and trusted by end-user browsers. This is the
2186 default for trusted origin server certificates.
2188 Sign to guarantee an X509_V_ERR_CERT_UNTRUSTED browser error.
2189 This is the default for untrusted origin server certificates
2190 that are not self-signed (see ssl::certUntrusted).
2192 Sign using a self-signed certificate with the right CN to
2193 generate a X509_V_ERR_DEPTH_ZERO_SELF_SIGNED_CERT error in the
2194 browser. This is the default for self-signed origin server
2195 certificates (see ssl::certSelfSigned).
2197 This clause only supports fast acl types.
2199 When sslproxy_cert_sign acl(s) match, Squid uses the corresponding
2200 signing algorithm to generate the certificate and ignores all
2201 subsequent sslproxy_cert_sign options (the first match wins). If no
2202 acl(s) match, the default signing algorithm is determined by errors
2203 detected when obtaining and validating the origin server certificate.
2205 WARNING: SQUID_X509_V_ERR_DOMAIN_MISMATCH and ssl:certDomainMismatch can
2206 be used with sslproxy_cert_adapt, but if and only if Squid is bumping a
2207 CONNECT request that carries a domain name. In all other cases (CONNECT
2208 to an IP address or an intercepted SSL connection), Squid cannot detect
2209 the domain mismatch at certificate generation time when
2210 bump-server-first is used.
2213 NAME: sslproxy_cert_adapt
2216 TYPE: sslproxy_cert_adapt
2217 LOC: Config.ssl_client.cert_adapt
2220 sslproxy_cert_adapt <adaptation algorithm> acl ...
2222 The following certificate adaptation algorithms are supported:
2224 Sets the "Not After" property to the "Not After" property of
2225 the CA certificate used to sign generated certificates.
2227 Sets the "Not Before" property to the "Not Before" property of
2228 the CA certificate used to sign generated certificates.
2229 setCommonName or setCommonName{CN}
2230 Sets Subject.CN property to the host name specified as a
2231 CN parameter or, if no explicit CN parameter was specified,
2232 extracted from the CONNECT request. It is a misconfiguration
2233 to use setCommonName without an explicit parameter for
2234 intercepted or tproxied SSL connections.
2236 This clause only supports fast acl types.
2238 Squid first groups sslproxy_cert_adapt options by adaptation algorithm.
2239 Within a group, when sslproxy_cert_adapt acl(s) match, Squid uses the
2240 corresponding adaptation algorithm to generate the certificate and
2241 ignores all subsequent sslproxy_cert_adapt options in that algorithm's
2242 group (i.e., the first match wins within each algorithm group). If no
2243 acl(s) match, the default mimicking action takes place.
2245 WARNING: SQUID_X509_V_ERR_DOMAIN_MISMATCH and ssl:certDomainMismatch can
2246 be used with sslproxy_cert_adapt, but if and only if Squid is bumping a
2247 CONNECT request that carries a domain name. In all other cases (CONNECT
2248 to an IP address or an intercepted SSL connection), Squid cannot detect
2249 the domain mismatch at certificate generation time when
2250 bump-server-first is used.
2253 NAME: sslpassword_program
2256 LOC: Config.Program.ssl_password
2259 Specify a program used for entering SSL key passphrases
2260 when using encrypted SSL certificate keys. If not specified
2261 keys must either be unencrypted, or Squid started with the -N
2262 option to allow it to query interactively for the passphrase.
2264 The key file name is given as argument to the program allowing
2265 selection of the right password if you have multiple encrypted
2270 OPTIONS RELATING TO EXTERNAL SSL_CRTD
2271 -----------------------------------------------------------------------------
2274 NAME: sslcrtd_program
2277 DEFAULT: @DEFAULT_SSL_CRTD@ -s @DEFAULT_SSL_DB_DIR@ -M 4MB
2278 LOC: Ssl::TheConfig.ssl_crtd
2280 Specify the location and options of the executable for ssl_crtd process.
2281 @DEFAULT_SSL_CRTD@ program requires -s and -M parameters
2282 For more information use:
2283 @DEFAULT_SSL_CRTD@ -h
2286 NAME: sslcrtd_children
2287 TYPE: HelperChildConfig
2289 DEFAULT: 32 startup=5 idle=1
2290 LOC: Ssl::TheConfig.ssl_crtdChildren
2292 The maximum number of processes spawn to service ssl server.
2293 The maximum this may be safely set to is 32.
2295 The startup= and idle= options allow some measure of skew in your
2300 Sets the minimum number of processes to spawn when Squid
2301 starts or reconfigures. When set to zero the first request will
2302 cause spawning of the first child process to handle it.
2304 Starting too few children temporary slows Squid under load while it
2305 tries to spawn enough additional processes to cope with traffic.
2309 Sets a minimum of how many processes Squid is to try and keep available
2310 at all times. When traffic begins to rise above what the existing
2311 processes can handle this many more will be spawned up to the maximum
2312 configured. A minimum setting of 1 is required.
2314 You must have at least one ssl_crtd process.
2318 OPTIONS WHICH AFFECT THE NEIGHBOR SELECTION ALGORITHM
2319 -----------------------------------------------------------------------------
2327 To specify other caches in a hierarchy, use the format:
2329 cache_peer hostname type http-port icp-port [options]
2334 # hostname type port port options
2335 # -------------------- -------- ----- ----- -----------
2336 cache_peer parent.foo.net parent 3128 3130 default
2337 cache_peer sib1.foo.net sibling 3128 3130 proxy-only
2338 cache_peer sib2.foo.net sibling 3128 3130 proxy-only
2339 cache_peer example.com parent 80 0 default
2340 cache_peer cdn.example.com sibling 3128 0
2342 type: either 'parent', 'sibling', or 'multicast'.
2344 proxy-port: The port number where the peer accept HTTP requests.
2345 For other Squid proxies this is usually 3128
2346 For web servers this is usually 80
2348 icp-port: Used for querying neighbor caches about objects.
2349 Set to 0 if the peer does not support ICP or HTCP.
2350 See ICP and HTCP options below for additional details.
2353 ==== ICP OPTIONS ====
2355 You MUST also set icp_port and icp_access explicitly when using these options.
2356 The defaults will prevent peer traffic using ICP.
2359 no-query Disable ICP queries to this neighbor.
2362 Indicates the named peer is a member of a multicast group.
2363 ICP queries will not be sent directly to the peer, but ICP
2364 replies will be accepted from it.
2366 closest-only Indicates that, for ICP_OP_MISS replies, we'll only forward
2367 CLOSEST_PARENT_MISSes and never FIRST_PARENT_MISSes.
2370 To only send ICP queries to this neighbor infrequently.
2371 This is used to keep the neighbor round trip time updated
2372 and is usually used in conjunction with weighted-round-robin.
2375 ==== HTCP OPTIONS ====
2377 You MUST also set htcp_port and htcp_access explicitly when using these options.
2378 The defaults will prevent peer traffic using HTCP.
2381 htcp Send HTCP, instead of ICP, queries to the neighbor.
2382 You probably also want to set the "icp-port" to 4827
2383 instead of 3130. This directive accepts a comma separated
2384 list of options described below.
2386 htcp=oldsquid Send HTCP to old Squid versions (2.5 or earlier).
2388 htcp=no-clr Send HTCP to the neighbor but without
2389 sending any CLR requests. This cannot be used with
2392 htcp=only-clr Send HTCP to the neighbor but ONLY CLR requests.
2393 This cannot be used with no-clr.
2396 Send HTCP to the neighbor including CLRs but only when
2397 they do not result from PURGE requests.
2400 Forward any HTCP CLR requests this proxy receives to the peer.
2403 ==== PEER SELECTION METHODS ====
2405 The default peer selection method is ICP, with the first responding peer
2406 being used as source. These options can be used for better load balancing.
2409 default This is a parent cache which can be used as a "last-resort"
2410 if a peer cannot be located by any of the peer-selection methods.
2411 If specified more than once, only the first is used.
2413 round-robin Load-Balance parents which should be used in a round-robin
2414 fashion in the absence of any ICP queries.
2415 weight=N can be used to add bias.
2417 weighted-round-robin
2418 Load-Balance parents which should be used in a round-robin
2419 fashion with the frequency of each parent being based on the
2420 round trip time. Closer parents are used more often.
2421 Usually used for background-ping parents.
2422 weight=N can be used to add bias.
2424 carp Load-Balance parents which should be used as a CARP array.
2425 The requests will be distributed among the parents based on the
2426 CARP load balancing hash function based on their weight.
2428 userhash Load-balance parents based on the client proxy_auth or ident username.
2430 sourcehash Load-balance parents based on the client source IP.
2433 To be used only for cache peers of type "multicast".
2434 ALL members of this multicast group have "sibling"
2435 relationship with it, not "parent". This is to a multicast
2436 group when the requested object would be fetched only from
2437 a "parent" cache, anyway. It's useful, e.g., when
2438 configuring a pool of redundant Squid proxies, being
2439 members of the same multicast group.
2442 ==== PEER SELECTION OPTIONS ====
2444 weight=N use to affect the selection of a peer during any weighted
2445 peer-selection mechanisms.
2446 The weight must be an integer; default is 1,
2447 larger weights are favored more.
2448 This option does not affect parent selection if a peering
2449 protocol is not in use.
2451 basetime=N Specify a base amount to be subtracted from round trip
2453 It is subtracted before division by weight in calculating
2454 which parent to fectch from. If the rtt is less than the
2455 base time the rtt is set to a minimal value.
2457 ttl=N Specify a TTL to use when sending multicast ICP queries
2459 Only useful when sending to a multicast group.
2460 Because we don't accept ICP replies from random
2461 hosts, you must configure other group members as
2462 peers with the 'multicast-responder' option.
2464 no-delay To prevent access to this neighbor from influencing the
2467 digest-url=URL Tell Squid to fetch the cache digest (if digests are
2468 enabled) for this host from the specified URL rather
2469 than the Squid default location.
2472 ==== CARP OPTIONS ====
2474 carp-key=key-specification
2475 use a different key than the full URL to hash against the peer.
2476 the key-specification is a comma-separated list of the keywords
2477 scheme, host, port, path, params
2478 Order is not important.
2480 ==== ACCELERATOR / REVERSE-PROXY OPTIONS ====
2482 originserver Causes this parent to be contacted as an origin server.
2483 Meant to be used in accelerator setups when the peer
2487 Set the Host header of requests forwarded to this peer.
2488 Useful in accelerator setups where the server (peer)
2489 expects a certain domain name but clients may request
2490 others. ie example.com or www.example.com
2492 no-digest Disable request of cache digests.
2495 Disables requesting ICMP RTT database (NetDB).
2498 ==== AUTHENTICATION OPTIONS ====
2501 If this is a personal/workgroup proxy and your parent
2502 requires proxy authentication.
2504 Note: The string can include URL escapes (i.e. %20 for
2505 spaces). This also means % must be written as %%.
2508 Send login details received from client to this peer.
2509 Both Proxy- and WWW-Authorization headers are passed
2510 without alteration to the peer.
2511 Authentication is not required by Squid for this to work.
2513 Note: This will pass any form of authentication but
2514 only Basic auth will work through a proxy unless the
2515 connection-auth options are also used.
2517 login=PASS Send login details received from client to this peer.
2518 Authentication is not required by this option.
2520 If there are no client-provided authentication headers
2521 to pass on, but username and password are available
2522 from an external ACL user= and password= result tags
2523 they may be sent instead.
2525 Note: To combine this with proxy_auth both proxies must
2526 share the same user database as HTTP only allows for
2527 a single login (one for proxy, one for origin server).
2528 Also be warned this will expose your users proxy
2529 password to the peer. USE WITH CAUTION
2532 Send the username to the upstream cache, but with a
2533 fixed password. This is meant to be used when the peer
2534 is in another administrative domain, but it is still
2535 needed to identify each user.
2536 The star can optionally be followed by some extra
2537 information which is added to the username. This can
2538 be used to identify this proxy to the peer, similar to
2539 the login=username:password option above.
2542 If this is a personal/workgroup proxy and your parent
2543 requires a secure proxy authentication.
2544 The first principal from the default keytab or defined by
2545 the environment variable KRB5_KTNAME will be used.
2547 WARNING: The connection may transmit requests from multiple
2548 clients. Negotiate often assumes end-to-end authentication
2549 and a single-client. Which is not strictly true here.
2551 login=NEGOTIATE:principal_name
2552 If this is a personal/workgroup proxy and your parent
2553 requires a secure proxy authentication.
2554 The principal principal_name from the default keytab or
2555 defined by the environment variable KRB5_KTNAME will be
2558 WARNING: The connection may transmit requests from multiple
2559 clients. Negotiate often assumes end-to-end authentication
2560 and a single-client. Which is not strictly true here.
2562 connection-auth=on|off
2563 Tell Squid that this peer does or not support Microsoft
2564 connection oriented authentication, and any such
2565 challenges received from there should be ignored.
2566 Default is auto to automatically determine the status
2570 ==== SSL / HTTPS / TLS OPTIONS ====
2572 ssl Encrypt connections to this peer with SSL/TLS.
2574 sslcert=/path/to/ssl/certificate
2575 A client SSL certificate to use when connecting to
2578 sslkey=/path/to/ssl/key
2579 The private SSL key corresponding to sslcert above.
2580 If 'sslkey' is not specified 'sslcert' is assumed to
2581 reference a combined file containing both the
2582 certificate and the key.
2585 The SSL version to use when connecting to this peer
2586 1 = automatic (default)
2591 sslcipher=... The list of valid SSL ciphers to use when connecting
2594 ssloptions=... Specify various SSL implementation options:
2596 NO_SSLv2 Disallow the use of SSLv2
2597 NO_SSLv3 Disallow the use of SSLv3
2598 NO_TLSv1 Disallow the use of TLSv1
2600 Always create a new key when using
2601 temporary/ephemeral DH key exchanges
2602 ALL Enable various bug workarounds
2603 suggested as "harmless" by OpenSSL
2604 Be warned that this reduces SSL/TLS
2605 strength to some attacks.
2607 See the OpenSSL SSL_CTX_set_options documentation for a
2610 sslcafile=... A file containing additional CA certificates to use
2611 when verifying the peer certificate.
2613 sslcapath=... A directory containing additional CA certificates to
2614 use when verifying the peer certificate.
2616 sslcrlfile=... A certificate revocation list file to use when
2617 verifying the peer certificate.
2619 sslflags=... Specify various flags modifying the SSL implementation:
2622 Accept certificates even if they fail to
2625 Don't use the default CA list built in
2628 Don't verify the peer certificate
2629 matches the server name
2631 ssldomain= The peer name as advertised in it's certificate.
2632 Used for verifying the correctness of the received peer
2633 certificate. If not specified the peer hostname will be
2637 Enable the "Front-End-Https: On" header needed when
2638 using Squid as a SSL frontend in front of Microsoft OWA.
2639 See MS KB document Q307347 for details on this header.
2640 If set to auto the header will only be added if the
2641 request is forwarded as a https:// URL.
2644 ==== GENERAL OPTIONS ====
2647 A peer-specific connect timeout.
2648 Also see the peer_connect_timeout directive.
2650 connect-fail-limit=N
2651 How many times connecting to a peer must fail before
2652 it is marked as down. Default is 10.
2654 allow-miss Disable Squid's use of only-if-cached when forwarding
2655 requests to siblings. This is primarily useful when
2656 icp_hit_stale is used by the sibling. To extensive use
2657 of this option may result in forwarding loops, and you
2658 should avoid having two-way peerings with this option.
2659 For example to deny peer usage on requests from peer
2660 by denying cache_peer_access if the source is a peer.
2662 max-conn=N Limit the amount of connections Squid may open to this
2665 name=xxx Unique name for the peer.
2666 Required if you have multiple peers on the same host
2667 but different ports.
2668 This name can be used in cache_peer_access and similar
2669 directives to dentify the peer.
2670 Can be used by outgoing access controls through the
2673 no-tproxy Do not use the client-spoof TPROXY support when forwarding
2674 requests to this peer. Use normal address selection instead.
2676 proxy-only objects fetched from the peer will not be stored locally.
2680 NAME: cache_peer_domain cache_host_domain
2685 Use to limit the domains for which a neighbor cache will be
2688 cache_peer_domain cache-host domain [domain ...]
2689 cache_peer_domain cache-host !domain
2691 For example, specifying
2693 cache_peer_domain parent.foo.net .edu
2695 has the effect such that UDP query packets are sent to
2696 'bigserver' only when the requested object exists on a
2697 server in the .edu domain. Prefixing the domainname
2698 with '!' means the cache will be queried for objects
2701 NOTE: * Any number of domains may be given for a cache-host,
2702 either on the same or separate lines.
2703 * When multiple domains are given for a particular
2704 cache-host, the first matched domain is applied.
2705 * Cache hosts with no domain restrictions are queried
2707 * There are no defaults.
2708 * There is also a 'cache_peer_access' tag in the ACL
2712 NAME: cache_peer_access
2717 Similar to 'cache_peer_domain' but provides more flexibility by
2720 cache_peer_access cache-host allow|deny [!]aclname ...
2722 The syntax is identical to 'http_access' and the other lists of
2723 ACL elements. See the comments for 'http_access' below, or
2724 the Squid FAQ (http://wiki.squid-cache.org/SquidFaq/SquidAcl).
2727 NAME: neighbor_type_domain
2728 TYPE: hostdomaintype
2732 usage: neighbor_type_domain neighbor parent|sibling domain domain ...
2734 Modifying the neighbor type for specific domains is now
2735 possible. You can treat some domains differently than the
2736 default neighbor type specified on the 'cache_peer' line.
2737 Normally it should only be necessary to list domains which
2738 should be treated differently because the default neighbor type
2739 applies for hostnames which do not match domains listed here.
2742 cache_peer cache.foo.org parent 3128 3130
2743 neighbor_type_domain cache.foo.org sibling .com .net
2744 neighbor_type_domain cache.foo.org sibling .au .de
2747 NAME: dead_peer_timeout
2751 LOC: Config.Timeout.deadPeer
2753 This controls how long Squid waits to declare a peer cache
2754 as "dead." If there are no ICP replies received in this
2755 amount of time, Squid will declare the peer dead and not
2756 expect to receive any further ICP replies. However, it
2757 continues to send ICP queries, and will mark the peer as
2758 alive upon receipt of the first subsequent ICP reply.
2760 This timeout also affects when Squid expects to receive ICP
2761 replies from peers. If more than 'dead_peer' seconds have
2762 passed since the last ICP reply was received, Squid will not
2763 expect to receive an ICP reply on the next query. Thus, if
2764 your time between requests is greater than this timeout, you
2765 will see a lot of requests sent DIRECT to origin servers
2766 instead of to your parents.
2769 NAME: forward_max_tries
2772 LOC: Config.forward_max_tries
2774 Controls how many different forward paths Squid will try
2775 before giving up. See also forward_timeout.
2777 NOTE: connect_retries (default: none) can make each of these
2778 possible forwarding paths be tried multiple times.
2781 NAME: hierarchy_stoplist
2784 LOC: Config.hierarchy_stoplist
2786 A list of words which, if found in a URL, cause the object to
2787 be handled directly by this cache. In other words, use this
2788 to not query neighbor caches for certain objects. You may
2789 list this option multiple times.
2792 hierarchy_stoplist cgi-bin ?
2794 Note: never_direct overrides this option.
2798 MEMORY CACHE OPTIONS
2799 -----------------------------------------------------------------------------
2806 LOC: Config.memMaxSize
2808 NOTE: THIS PARAMETER DOES NOT SPECIFY THE MAXIMUM PROCESS SIZE.
2809 IT ONLY PLACES A LIMIT ON HOW MUCH ADDITIONAL MEMORY SQUID WILL
2810 USE AS A MEMORY CACHE OF OBJECTS. SQUID USES MEMORY FOR OTHER
2811 THINGS AS WELL. SEE THE SQUID FAQ SECTION 8 FOR DETAILS.
2813 'cache_mem' specifies the ideal amount of memory to be used
2815 * In-Transit objects
2817 * Negative-Cached objects
2819 Data for these objects are stored in 4 KB blocks. This
2820 parameter specifies the ideal upper limit on the total size of
2821 4 KB blocks allocated. In-Transit objects take the highest
2824 In-transit objects have priority over the others. When
2825 additional space is needed for incoming data, negative-cached
2826 and hot objects will be released. In other words, the
2827 negative-cached and hot objects will fill up any unused space
2828 not needed for in-transit objects.
2830 If circumstances require, this limit will be exceeded.
2831 Specifically, if your incoming request rate requires more than
2832 'cache_mem' of memory to hold in-transit objects, Squid will
2833 exceed this limit to satisfy the new requests. When the load
2834 decreases, blocks will be freed until the high-water mark is
2835 reached. Thereafter, blocks will be used to store hot
2838 If shared memory caching is enabled, Squid does not use the shared
2839 cache space for in-transit objects, but they still consume as much
2840 local memory as they need. For more details about the shared memory
2841 cache, see memory_cache_shared.
2844 NAME: maximum_object_size_in_memory
2848 LOC: Config.Store.maxInMemObjSize
2850 Objects greater than this size will not be attempted to kept in
2851 the memory cache. This should be set high enough to keep objects
2852 accessed frequently in memory to improve performance whilst low
2853 enough to keep larger objects from hoarding cache_mem.
2856 NAME: memory_cache_shared
2859 LOC: Config.memShared
2861 DEFAULT_DOC: "on" where supported if doing memory caching with multiple SMP workers.
2863 Controls whether the memory cache is shared among SMP workers.
2865 The shared memory cache is meant to occupy cache_mem bytes and replace
2866 the non-shared memory cache, although some entities may still be
2867 cached locally by workers for now (e.g., internal and in-transit
2868 objects may be served from a local memory cache even if shared memory
2869 caching is enabled).
2871 By default, the memory cache is shared if and only if all of the
2872 following conditions are satisfied: Squid runs in SMP mode with
2873 multiple workers, cache_mem is positive, and Squid environment
2874 supports required IPC primitives (e.g., POSIX shared memory segments
2875 and GCC-style atomic operations).
2877 To avoid blocking locks, shared memory uses opportunistic algorithms
2878 that do not guarantee that every cachable entity that could have been
2879 shared among SMP workers will actually be shared.
2881 Currently, entities exceeding 32KB in size cannot be shared.
2884 NAME: memory_cache_mode
2889 Controls which objects to keep in the memory cache (cache_mem)
2891 always Keep most recently fetched objects in memory (default)
2893 disk Only disk cache hits are kept in memory, which means
2894 an object must first be cached on disk and then hit
2895 a second time before cached in memory.
2897 network Only objects fetched from network is kept in memory
2900 NAME: memory_replacement_policy
2902 LOC: Config.memPolicy
2905 The memory replacement policy parameter determines which
2906 objects are purged from memory when memory space is needed.
2908 See cache_replacement_policy for details.
2913 -----------------------------------------------------------------------------
2916 NAME: cache_replacement_policy
2918 LOC: Config.replPolicy
2921 The cache replacement policy parameter determines which
2922 objects are evicted (replaced) when disk space is needed.
2924 lru : Squid's original list based LRU policy
2925 heap GDSF : Greedy-Dual Size Frequency
2926 heap LFUDA: Least Frequently Used with Dynamic Aging
2927 heap LRU : LRU policy implemented using a heap
2929 Applies to any cache_dir lines listed below this.
2931 The LRU policies keeps recently referenced objects.
2933 The heap GDSF policy optimizes object hit rate by keeping smaller
2934 popular objects in cache so it has a better chance of getting a
2935 hit. It achieves a lower byte hit rate than LFUDA though since
2936 it evicts larger (possibly popular) objects.
2938 The heap LFUDA policy keeps popular objects in cache regardless of
2939 their size and thus optimizes byte hit rate at the expense of
2940 hit rate since one large, popular object will prevent many
2941 smaller, slightly less popular objects from being cached.
2943 Both policies utilize a dynamic aging mechanism that prevents
2944 cache pollution that can otherwise occur with frequency-based
2945 replacement policies.
2947 NOTE: if using the LFUDA replacement policy you should increase
2948 the value of maximum_object_size above its default of 4096 KB to
2949 to maximize the potential byte hit rate improvement of LFUDA.
2951 For more information about the GDSF and LFUDA cache replacement
2952 policies see http://www.hpl.hp.com/techreports/1999/HPL-1999-69.html
2953 and http://fog.hpl.external.hp.com/techreports/98/HPL-98-173.html.
2959 LOC: Config.cacheSwap
2963 cache_dir Type Directory-Name Fs-specific-data [options]
2965 You can specify multiple cache_dir lines to spread the
2966 cache among different disk partitions.
2968 Type specifies the kind of storage system to use. Only "ufs"
2969 is built by default. To enable any of the other storage systems
2970 see the --enable-storeio configure option.
2972 'Directory' is a top-level directory where cache swap
2973 files will be stored. If you want to use an entire disk
2974 for caching, this can be the mount-point directory.
2975 The directory must exist and be writable by the Squid
2976 process. Squid will NOT create this directory for you.
2978 In SMP configurations, cache_dir must not precede the workers option
2979 and should use configuration macros or conditionals to give each
2980 worker interested in disk caching a dedicated cache directory.
2984 "ufs" is the old well-known Squid storage format that has always
2987 cache_dir ufs Directory-Name Mbytes L1 L2 [options]
2989 'Mbytes' is the amount of disk space (MB) to use under this
2990 directory. The default is 100 MB. Change this to suit your
2991 configuration. Do NOT put the size of your disk drive here.
2992 Instead, if you want Squid to use the entire disk drive,
2993 subtract 20% and use that value.
2995 'L1' is the number of first-level subdirectories which
2996 will be created under the 'Directory'. The default is 16.
2998 'L2' is the number of second-level subdirectories which
2999 will be created under each first-level directory. The default
3002 The aufs store type:
3004 "aufs" uses the same storage format as "ufs", utilizing
3005 POSIX-threads to avoid blocking the main Squid process on
3006 disk-I/O. This was formerly known in Squid as async-io.
3008 cache_dir aufs Directory-Name Mbytes L1 L2 [options]
3010 see argument descriptions under ufs above
3012 The diskd store type:
3014 "diskd" uses the same storage format as "ufs", utilizing a
3015 separate process to avoid blocking the main Squid process on
3018 cache_dir diskd Directory-Name Mbytes L1 L2 [options] [Q1=n] [Q2=n]
3020 see argument descriptions under ufs above
3022 Q1 specifies the number of unacknowledged I/O requests when Squid
3023 stops opening new files. If this many messages are in the queues,
3024 Squid won't open new files. Default is 64
3026 Q2 specifies the number of unacknowledged messages when Squid
3027 starts blocking. If this many messages are in the queues,
3028 Squid blocks until it receives some replies. Default is 72
3030 When Q1 < Q2 (the default), the cache directory is optimized
3031 for lower response time at the expense of a decrease in hit
3032 ratio. If Q1 > Q2, the cache directory is optimized for
3033 higher hit ratio at the expense of an increase in response
3036 The rock store type:
3038 cache_dir rock Directory-Name Mbytes <max-size=bytes> [options]
3040 The Rock Store type is a database-style storage. All cached
3041 entries are stored in a "database" file, using fixed-size slots,
3042 one entry per slot. The database size is specified in MB. The
3043 slot size is specified in bytes using the max-size option. See
3044 below for more info on the max-size option.
3046 swap-timeout=msec: Squid will not start writing a miss to or
3047 reading a hit from disk if it estimates that the swap operation
3048 will take more than the specified number of milliseconds. By
3049 default and when set to zero, disables the disk I/O time limit
3050 enforcement. Ignored when using blocking I/O module because
3051 blocking synchronous I/O does not allow Squid to estimate the
3052 expected swap wait time.
3054 max-swap-rate=swaps/sec: Artificially limits disk access using
3055 the specified I/O rate limit. Swap out requests that
3056 would cause the average I/O rate to exceed the limit are
3057 delayed. Individual swap in requests (i.e., hits or reads) are
3058 not delayed, but they do contribute to measured swap rate and
3059 since they are placed in the same FIFO queue as swap out
3060 requests, they may wait longer if max-swap-rate is smaller.
3061 This is necessary on file systems that buffer "too
3062 many" writes and then start blocking Squid and other processes
3063 while committing those writes to disk. Usually used together
3064 with swap-timeout to avoid excessive delays and queue overflows
3065 when disk demand exceeds available disk "bandwidth". By default
3066 and when set to zero, disables the disk I/O rate limit
3067 enforcement. Currently supported by IpcIo module only.
3070 The coss store type:
3072 NP: COSS filesystem in Squid-3 has been deemed too unstable for
3073 production use and has thus been removed from this release.
3074 We hope that it can be made usable again soon.
3076 block-size=n defines the "block size" for COSS cache_dir's.
3077 Squid uses file numbers as block numbers. Since file numbers
3078 are limited to 24 bits, the block size determines the maximum
3079 size of the COSS partition. The default is 512 bytes, which
3080 leads to a maximum cache_dir size of 512<<24, or 8 GB. Note
3081 you should not change the coss block size after Squid
3082 has written some objects to the cache_dir.
3084 The coss file store has changed from 2.5. Now it uses a file
3085 called 'stripe' in the directory names in the config - and
3086 this will be created by squid -z.
3090 no-store, no new objects should be stored to this cache_dir
3092 min-size=n, refers to the min object size in bytes this cache_dir
3093 will accept. It's used to restrict a cache_dir to only store
3094 large objects (e.g. aufs) while other storedirs are optimized
3095 for smaller objects (e.g. COSS). Defaults to 0.
3097 max-size=n, refers to the max object size in bytes this cache_dir
3098 supports. It is used to select the cache_dir to store the object.
3099 Note: To make optimal use of the max-size limits you should order
3100 the cache_dir lines with the smallest max-size value first and the
3101 ones with no max-size specification last.
3103 Note for coss, max-size must be less than COSS_MEMBUF_SZ,
3104 which can be changed with the --with-coss-membuf-size=N configure
3108 # Uncomment and adjust the following to add a disk cache directory.
3109 #cache_dir ufs @DEFAULT_SWAP_DIR@ 100 16 256
3113 NAME: store_dir_select_algorithm
3115 LOC: Config.store_dir_select_algorithm
3118 Set this to 'round-robin' as an alternative.
3121 NAME: max_open_disk_fds
3123 LOC: Config.max_open_disk_fds
3126 To avoid having disk as the I/O bottleneck Squid can optionally
3127 bypass the on-disk cache if more than this amount of disk file
3128 descriptors are open.
3130 A value of 0 indicates no limit.
3133 NAME: minimum_object_size
3137 LOC: Config.Store.minObjectSize
3139 Objects smaller than this size will NOT be saved on disk. The
3140 value is specified in kilobytes, and the default is 0 KB, which
3141 means there is no minimum.
3144 NAME: maximum_object_size
3148 LOC: Config.Store.maxObjectSize
3150 Objects larger than this size will NOT be saved on disk. The
3151 value is specified in kilobytes, and the default is 4MB. If
3152 you wish to get a high BYTES hit ratio, you should probably
3153 increase this (one 32 MB object hit counts for 3200 10KB
3154 hits). If you wish to increase speed more than your want to
3155 save bandwidth you should leave this low.
3157 NOTE: if using the LFUDA replacement policy you should increase
3158 this value to maximize the byte hit rate improvement of LFUDA!
3159 See replacement_policy below for a discussion of this policy.
3162 NAME: cache_swap_low
3163 COMMENT: (percent, 0-100)
3166 LOC: Config.Swap.lowWaterMark
3169 NAME: cache_swap_high
3170 COMMENT: (percent, 0-100)
3173 LOC: Config.Swap.highWaterMark
3176 The low- and high-water marks for cache object replacement.
3177 Replacement begins when the swap (disk) usage is above the
3178 low-water mark and attempts to maintain utilization near the
3179 low-water mark. As swap utilization gets close to high-water
3180 mark object eviction becomes more aggressive. If utilization is
3181 close to the low-water mark less replacement is done each time.
3183 Defaults are 90% and 95%. If you have a large cache, 5% could be
3184 hundreds of MB. If this is the case you may wish to set these
3185 numbers closer together.
3190 -----------------------------------------------------------------------------
3200 logformat <name> <format specification>
3202 Defines an access log format.
3204 The <format specification> is a string with embedded % format codes
3206 % format codes all follow the same basic structure where all but
3207 the formatcode is optional. Output strings are automatically escaped
3208 as required according to their context and the output format
3209 modifiers are usually not needed, but can be specified if an explicit
3210 output format is desired.
3212 % ["|[|'|#] [-] [[0]width] [{argument}] formatcode
3214 " output in quoted string format
3215 [ output in squid text log format as used by log_mime_hdrs
3216 # output in URL quoted format
3221 width minimum and/or maximum field width:
3222 [width_min][.width_max]
3223 When minimum starts with 0, the field is zero-padded.
3224 String values exceeding maximum width are truncated.
3226 {arg} argument such as header name etc
3230 % a literal % character
3231 sn Unique sequence number per log line entry
3232 err_code The ID of an error response served by Squid or
3233 a similar internal error identifier.
3234 err_detail Additional err_code-dependent error information.
3236 Connection related format codes:
3238 >a Client source IP address
3240 >p Client source port
3241 >eui Client source EUI (MAC address, EUI-48 or EUI-64 identifier)
3242 >la Local IP address the client connected to
3243 >lp Local port number the client connected to
3245 la Local listening IP address the client connection was connected to.
3246 lp Local listening port number the client connection was connected to.
3248 <a Server IP address of the last server or peer connection
3249 <A Server FQDN or peer name
3250 <p Server port number of the last server or peer connection
3251 <la Local IP address of the last server or peer connection
3252 <lp Local port number of the last server or peer connection
3254 Time related format codes:
3256 ts Seconds since epoch
3257 tu subsecond time (milliseconds)
3258 tl Local time. Optional strftime format argument
3259 default %d/%b/%Y:%H:%M:%S %z
3260 tg GMT time. Optional strftime format argument
3261 default %d/%b/%Y:%H:%M:%S %z
3262 tr Response time (milliseconds)
3263 dt Total time spent making DNS lookups (milliseconds)
3265 Access Control related format codes:
3267 et Tag returned by external acl
3268 ea Log string returned by external acl
3269 un User name (any available)
3270 ul User name from authentication
3271 ue User name from external acl helper
3272 ui User name from ident
3273 us User name from SSL
3275 HTTP related format codes:
3277 [http::]>h Original request header. Optional header name argument
3278 on the format header[:[separator]element]
3279 [http::]>ha The HTTP request headers after adaptation and redirection.
3280 Optional header name argument as for >h
3281 [http::]<h Reply header. Optional header name argument
3283 [http::]>Hs HTTP status code sent to the client
3284 [http::]<Hs HTTP status code received from the next hop
3285 [http::]<bs Number of HTTP-equivalent message body bytes
3286 received from the next hop, excluding chunked
3287 transfer encoding and control messages.
3288 Generated FTP/Gopher listings are treated as
3290 [http::]mt MIME content type
3291 [http::]rm Request method (GET/POST etc)
3292 [http::]>rm Request method from client
3293 [http::]<rm Request method sent to server or peer
3294 [http::]ru Request URL from client (historic, filtered for logging)
3295 [http::]>ru Request URL from client
3296 [http::]<ru Request URL sent to server or peer
3297 [http::]rp Request URL-Path excluding hostname
3298 [http::]>rp Request URL-Path excluding hostname from client
3299 [http::]<rp Request URL-Path excluding hostname sento to server or peer
3300 [http::]rv Request protocol version
3301 [http::]>rv Request protocol version from client
3302 [http::]<rv Request protocol version sent to server or peer
3303 [http::]<st Sent reply size including HTTP headers
3304 [http::]>st Received request size including HTTP headers. In the
3305 case of chunked requests the chunked encoding metadata
3307 [http::]>sh Received HTTP request headers size
3308 [http::]<sh Sent HTTP reply headers size
3309 [http::]st Request+Reply size including HTTP headers
3310 [http::]<sH Reply high offset sent
3311 [http::]<sS Upstream object size
3312 [http::]<pt Peer response time in milliseconds. The timer starts
3313 when the last request byte is sent to the next hop
3314 and stops when the last response byte is received.
3315 [http::]<tt Total server-side time in milliseconds. The timer
3316 starts with the first connect request (or write I/O)
3317 sent to the first selected peer. The timer stops
3318 with the last I/O with the last peer.
3320 Squid handling related format codes:
3322 Ss Squid request status (TCP_MISS etc)
3323 Sh Squid hierarchy status (DEFAULT_PARENT etc)
3325 If ICAP is enabled, the following code becomes available (as
3326 well as ICAP log codes documented with the icap_log option):
3328 icap::tt Total ICAP processing time for the HTTP
3329 transaction. The timer ticks when ICAP
3330 ACLs are checked and when ICAP
3331 transaction is in progress.
3333 If adaptation is enabled the following three codes become available:
3335 adapt::<last_h The header of the last ICAP response or
3336 meta-information from the last eCAP
3337 transaction related to the HTTP transaction.
3338 Like <h, accepts an optional header name
3341 adapt::sum_trs Summed adaptation transaction response
3342 times recorded as a comma-separated list in
3343 the order of transaction start time. Each time
3344 value is recorded as an integer number,
3345 representing response time of one or more
3346 adaptation (ICAP or eCAP) transaction in
3347 milliseconds. When a failed transaction is
3348 being retried or repeated, its time is not
3349 logged individually but added to the
3350 replacement (next) transaction. See also:
3353 adapt::all_trs All adaptation transaction response times.
3354 Same as adaptation_strs but response times of
3355 individual transactions are never added
3356 together. Instead, all transaction response
3357 times are recorded individually.
3359 You can prefix adapt::*_trs format codes with adaptation
3360 service name in curly braces to record response time(s) specific
3361 to that service. For example: %{my_service}adapt::sum_trs
3363 The default formats available (which do not need re-defining) are:
3365 logformat squid %ts.%03tu %6tr %>a %Ss/%03>Hs %<st %rm %ru %[un %Sh/%<a %mt
3366 logformat common %>a %[ui %[un [%tl] "%rm %ru HTTP/%rv" %>Hs %<st %Ss:%Sh
3367 logformat combined %>a %[ui %[un [%tl] "%rm %ru HTTP/%rv" %>Hs %<st "%{Referer}>h" "%{User-Agent}>h" %Ss:%Sh
3368 logformat referrer %ts.%03tu %>a %{Referer}>h %ru
3369 logformat useragent %>a [%tl] "%{User-Agent}>h"
3371 NOTE: When the log_mime_hdrs directive is set to ON.
3372 The squid, common and combined formats have a safely encoded copy
3373 of the mime headers appended to each line within a pair of brackets.
3375 NOTE: The common and combined formats are not quite true to the Apache definition.
3376 The logs from Squid contain an extra status and hierarchy code appended.
3380 NAME: access_log cache_access_log
3382 LOC: Config.Log.accesslogs
3383 DEFAULT_IF_NONE: daemon:@DEFAULT_ACCESS_LOG@ squid
3385 These files log client request activities. Has a line every HTTP or
3386 ICP request. The format is:
3387 access_log <module>:<place> [<logformat name> [acl acl ...]]
3388 access_log none [acl acl ...]]
3390 Will log to the specified module:place using the specified format (which
3391 must be defined in a logformat directive) those entries which match
3392 ALL the acl's specified (which must be defined in acl clauses).
3393 If no acl is specified, all requests will be logged to this destination.
3395 ===== Modules Currently available =====
3397 none Do not log any requests matching these ACL.
3398 Do not specify Place or logformat name.
3400 stdio Write each log line to disk immediately at the completion of
3402 Place: the filename and path to be written.
3404 daemon Very similar to stdio. But instead of writing to disk the log
3405 line is passed to a daemon helper for asychronous handling instead.
3406 Place: varies depending on the daemon.
3408 log_file_daemon Place: the file name and path to be written.
3410 syslog To log each request via syslog facility.
3411 Place: The syslog facility and priority level for these entries.
3412 Place Format: facility.priority
3414 where facility could be any of:
3415 authpriv, daemon, local0 ... local7 or user.
3417 And priority could be any of:
3418 err, warning, notice, info, debug.
3420 udp To send each log line as text data to a UDP receiver.
3421 Place: The destination host name or IP and port.
3422 Place Format: //host:port
3424 tcp To send each log line as text data to a TCP receiver.
3425 Place: The destination host name or IP and port.
3426 Place Format: //host:port
3429 access_log daemon:@DEFAULT_ACCESS_LOG@ squid
3435 LOC: Config.Log.icaplogs
3438 ICAP log files record ICAP transaction summaries, one line per
3441 The icap_log option format is:
3442 icap_log <filepath> [<logformat name> [acl acl ...]]
3443 icap_log none [acl acl ...]]
3445 Please see access_log option documentation for details. The two
3446 kinds of logs share the overall configuration approach and many
3449 ICAP processing of a single HTTP message or transaction may
3450 require multiple ICAP transactions. In such cases, multiple
3451 ICAP transaction log lines will correspond to a single access
3454 ICAP log uses logformat codes that make sense for an ICAP
3455 transaction. Header-related codes are applied to the HTTP header
3456 embedded in an ICAP server response, with the following caveats:
3457 For REQMOD, there is no HTTP response header unless the ICAP
3458 server performed request satisfaction. For RESPMOD, the HTTP
3459 request header is the header sent to the ICAP server. For
3460 OPTIONS, there are no HTTP headers.
3462 The following format codes are also available for ICAP logs:
3464 icap::<A ICAP server IP address. Similar to <A.
3466 icap::<service_name ICAP service name from the icap_service
3467 option in Squid configuration file.
3469 icap::ru ICAP Request-URI. Similar to ru.
3471 icap::rm ICAP request method (REQMOD, RESPMOD, or
3472 OPTIONS). Similar to existing rm.
3474 icap::>st Bytes sent to the ICAP server (TCP payload
3475 only; i.e., what Squid writes to the socket).
3477 icap::<st Bytes received from the ICAP server (TCP
3478 payload only; i.e., what Squid reads from
3481 icap::<bs Number of message body bytes received from the
3482 ICAP server. ICAP message body, if any, usually
3483 includes encapsulated HTTP message headers and
3484 possibly encapsulated HTTP message body. The
3485 HTTP body part is dechunked before its size is
3488 icap::tr Transaction response time (in
3489 milliseconds). The timer starts when
3490 the ICAP transaction is created and
3491 stops when the transaction is completed.
3494 icap::tio Transaction I/O time (in milliseconds). The
3495 timer starts when the first ICAP request
3496 byte is scheduled for sending. The timers
3497 stops when the last byte of the ICAP response
3500 icap::to Transaction outcome: ICAP_ERR* for all
3501 transaction errors, ICAP_OPT for OPTION
3502 transactions, ICAP_ECHO for 204
3503 responses, ICAP_MOD for message
3504 modification, and ICAP_SAT for request
3505 satisfaction. Similar to Ss.
3507 icap::Hs ICAP response status code. Similar to Hs.
3509 icap::>h ICAP request header(s). Similar to >h.
3511 icap::<h ICAP response header(s). Similar to <h.
3513 The default ICAP log format, which can be used without an explicit
3514 definition, is called icap_squid:
3516 logformat icap_squid %ts.%03tu %6icap::tr %>a %icap::to/%03icap::Hs %icap::<size %icap::rm %icap::ru% %un -/%icap::<A -
3518 See also: logformat, log_icap, and %adapt::<last_h
3521 NAME: logfile_daemon
3523 DEFAULT: @DEFAULT_LOGFILED@
3524 LOC: Log::TheConfig.logfile_daemon
3526 Specify the path to the logfile-writing daemon. This daemon is
3527 used to write the access and store logs, if configured.
3529 Squid sends a number of commands to the log daemon:
3530 L<data>\n - logfile data
3535 r<n>\n - set rotate count to <n>
3536 b<n>\n - 1 = buffer output, 0 = don't buffer output
3538 No responses is expected.
3543 LOC: Config.accessList.log
3545 COMMENT: allow|deny acl acl...
3547 This options allows you to control which requests gets logged
3548 to access.log (see access_log directive). Requests denied for
3549 logging will also not be accounted for in performance counters.
3551 This clause only supports fast acl types.
3552 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
3558 LOC: Config.accessList.icap
3561 This options allows you to control which requests get logged
3562 to icap.log. See the icap_log directive for ICAP log details.
3565 NAME: cache_store_log
3568 LOC: Config.Log.store
3570 Logs the activities of the storage manager. Shows which
3571 objects are ejected from the cache, and which objects are
3572 saved and for how long. To disable, enter "none" or remove the line.
3573 There are not really utilities to analyze this data, so you can safely
3577 cache_store_log @DEFAULT_STORE_LOG@
3580 NAME: cache_swap_state cache_swap_log
3582 LOC: Config.Log.swap
3585 Location for the cache "swap.state" file. This index file holds
3586 the metadata of objects saved on disk. It is used to rebuild
3587 the cache during startup. Normally this file resides in each
3588 'cache_dir' directory, but you may specify an alternate
3589 pathname here. Note you must give a full filename, not just
3590 a directory. Since this is the index for the whole object
3591 list you CANNOT periodically rotate it!
3593 If %s can be used in the file name it will be replaced with a
3594 a representation of the cache_dir name where each / is replaced
3595 with '.'. This is needed to allow adding/removing cache_dir
3596 lines when cache_swap_log is being used.
3598 If have more than one 'cache_dir', and %s is not used in the name
3599 these swap logs will have names such as:
3605 The numbered extension (which is added automatically)
3606 corresponds to the order of the 'cache_dir' lines in this
3607 configuration file. If you change the order of the 'cache_dir'
3608 lines in this file, these index files will NOT correspond to
3609 the correct 'cache_dir' entry (unless you manually rename
3610 them). We recommend you do NOT use this option. It is
3611 better to keep these index files in each 'cache_dir' directory.
3614 NAME: logfile_rotate
3617 LOC: Config.Log.rotateNumber
3619 Specifies the number of logfile rotations to make when you
3620 type 'squid -k rotate'. The default is 10, which will rotate
3621 with extensions 0 through 9. Setting logfile_rotate to 0 will
3622 disable the file name rotation, but the logfiles are still closed
3623 and re-opened. This will enable you to rename the logfiles
3624 yourself just before sending the rotate signal.
3626 Note, the 'squid -k rotate' command normally sends a USR1
3627 signal to the running squid process. In certain situations
3628 (e.g. on Linux with Async I/O), USR1 is used for other
3629 purposes, so -k rotate uses another signal. It is best to get
3630 in the habit of using 'squid -k rotate' instead of 'kill -USR1
3633 Note, from Squid-3.1 this option has no effect on the cache.log,
3634 that log can be rotated separately by using debug_options
3637 NAME: emulate_httpd_log
3640 Replace this with an access_log directive using the format 'common' or 'combined'.
3643 NAME: log_ip_on_direct
3646 Remove this option from your config. To log server or peer names use %<A in the log format.
3651 DEFAULT: @DEFAULT_MIME_TABLE@
3652 LOC: Config.mimeTablePathname
3654 Pathname to Squid's MIME table. You shouldn't need to change
3655 this, but the default file contains examples and formatting
3656 information if you do.
3662 LOC: Config.onoff.log_mime_hdrs
3665 The Cache can record both the request and the response MIME
3666 headers for each HTTP transaction. The headers are encoded
3667 safely and will appear as two bracketed fields at the end of
3668 the access log (for either the native or httpd-emulated log
3669 formats). To enable this logging set log_mime_hdrs to 'on'.
3675 Replace this with an access_log directive using the format 'useragent'.
3678 NAME: referer_log referrer_log
3681 Replace this with an access_log directive using the format 'referrer'.
3686 DEFAULT: @DEFAULT_PID_FILE@
3687 LOC: Config.pidFilename
3689 A filename to write the process-id to. To disable, enter "none".
3695 Remove this option from your config. To log FQDN use %>A in the log format.
3698 NAME: client_netmask
3700 LOC: Config.Addrs.client_netmask
3703 A netmask for client addresses in logfiles and cachemgr output.
3704 Change this to protect the privacy of your cache clients.
3705 A netmask of 255.255.255.0 will log all IP's in that range with
3706 the last digit set to '0'.
3712 Use a regular access.log with ACL limiting it to MISS events.
3715 NAME: strip_query_terms
3717 LOC: Config.onoff.strip_query_terms
3720 By default, Squid strips query terms from requested URLs before
3721 logging. This protects your user's privacy.
3728 LOC: Config.onoff.buffered_logs
3730 cache.log log file is written with stdio functions, and as such
3731 it can be buffered or unbuffered. By default it will be unbuffered.
3732 Buffering it can speed up the writing slightly (though you are
3733 unlikely to need to worry unless you run with tons of debugging
3734 enabled in which case performance will suffer badly anyway..).
3737 NAME: netdb_filename
3739 DEFAULT: stdio:@DEFAULT_NETDB_FILE@
3740 LOC: Config.netdbFilename
3743 A filename where Squid stores it's netdb state between restarts.
3744 To disable, enter "none".
3748 OPTIONS FOR TROUBLESHOOTING
3749 -----------------------------------------------------------------------------
3754 DEFAULT_IF_NONE: @DEFAULT_CACHE_LOG@
3755 LOC: Debug::cache_log
3757 Cache logging file. This is where general information about
3758 your cache's behavior goes. You can increase the amount of data
3759 logged to this file and how often its rotated with "debug_options"
3765 LOC: Debug::debugOptions
3767 Logging options are set as section,level where each source file
3768 is assigned a unique section. Lower levels result in less
3769 output, Full debugging (level 9) can result in a very large
3770 log file, so be careful.
3772 The magic word "ALL" sets debugging levels for all sections.
3773 We recommend normally running with "ALL,1".
3775 The rotate=N option can be used to keep more or less of these logs
3776 than would otherwise be kept by logfile_rotate.
3777 For most uses a single log should be enough to monitor current
3778 events affecting Squid.
3783 LOC: Config.coredump_dir
3784 DEFAULT_IF_NONE: none
3786 By default Squid leaves core files in the directory from where
3787 it was started. If you set 'coredump_dir' to a directory
3788 that exists, Squid will chdir() to that directory at startup
3789 and coredump files will be left there.
3793 # Leave coredumps in the first cache dir
3794 coredump_dir @DEFAULT_SWAP_DIR@
3800 OPTIONS FOR FTP GATEWAYING
3801 -----------------------------------------------------------------------------
3807 LOC: Config.Ftp.anon_user
3809 If you want the anonymous login password to be more informative
3810 (and enable the use of picky ftp servers), set this to something
3811 reasonable for your domain, like wwwuser@somewhere.net
3813 The reason why this is domainless by default is the
3814 request can be made on the behalf of a user in any domain,
3815 depending on how the cache is used.
3816 Some ftp server also validate the email address is valid
3817 (for example perl.com).
3823 LOC: Config.Ftp.passive
3825 If your firewall does not allow Squid to use passive
3826 connections, turn off this option.
3828 Use of ftp_epsv_all option requires this to be ON.
3834 LOC: Config.Ftp.epsv_all
3836 FTP Protocol extensions permit the use of a special "EPSV ALL" command.
3838 NATs may be able to put the connection on a "fast path" through the
3839 translator, as the EPRT command will never be used and therefore,
3840 translation of the data portion of the segments will never be needed.
3842 When a client only expects to do two-way FTP transfers this may be
3844 If squid finds that it must do a three-way FTP transfer after issuing
3845 an EPSV ALL command, the FTP session will fail.
3847 If you have any doubts about this option do not use it.
3848 Squid will nicely attempt all other connection methods.
3850 Requires ftp_passive to be ON (default) for any effect.
3856 LOC: Config.Ftp.epsv
3858 FTP Protocol extensions permit the use of a special "EPSV" command.
3860 NATs may be able to put the connection on a "fast path" through the
3861 translator using EPSV, as the EPRT command will never be used
3862 and therefore, translation of the data portion of the segments
3863 will never be needed.
3865 Turning this OFF will prevent EPSV being attempted.
3866 WARNING: Doing so will convert Squid back to the old behavior with all
3867 the related problems with external NAT devices/layers.
3869 Requires ftp_passive to be ON (default) for any effect.
3875 LOC: Config.Ftp.eprt
3877 FTP Protocol extensions permit the use of a special "EPRT" command.
3879 This extension provides a protocol neutral alternative to the
3880 IPv4-only PORT command. When supported it enables active FTP data
3881 channels over IPv6 and efficient NAT handling.
3883 Turning this OFF will prevent EPRT being attempted and will skip
3884 straight to using PORT for IPv4 servers.
3886 Some devices are known to not handle this extension correctly and
3887 may result in crashes. Devices which suport EPRT enough to fail
3888 cleanly will result in Squid attempting PORT anyway. This directive
3889 should only be disabled when EPRT results in device failures.
3891 WARNING: Doing so will convert Squid back to the old behavior with all
3892 the related problems with external NAT devices/layers and IPv4-only FTP.
3895 NAME: ftp_sanitycheck
3898 LOC: Config.Ftp.sanitycheck
3900 For security and data integrity reasons Squid by default performs
3901 sanity checks of the addresses of FTP data connections ensure the
3902 data connection is to the requested server. If you need to allow
3903 FTP connections to servers using another IP address for the data
3904 connection turn this off.
3907 NAME: ftp_telnet_protocol
3910 LOC: Config.Ftp.telnet
3912 The FTP protocol is officially defined to use the telnet protocol
3913 as transport channel for the control connection. However, many
3914 implementations are broken and does not respect this aspect of
3917 If you have trouble accessing files with ASCII code 255 in the
3918 path or similar problems involving this ASCII code you can
3919 try setting this directive to off. If that helps, report to the
3920 operator of the FTP server in question that their FTP server
3921 is broken and does not follow the FTP standard.
3925 OPTIONS FOR EXTERNAL SUPPORT PROGRAMS
3926 -----------------------------------------------------------------------------
3931 DEFAULT: @DEFAULT_DISKD@
3932 LOC: Config.Program.diskd
3934 Specify the location of the diskd executable.
3935 Note this is only useful if you have compiled in
3936 diskd as one of the store io modules.
3939 NAME: unlinkd_program
3942 DEFAULT: @DEFAULT_UNLINKD@
3943 LOC: Config.Program.unlinkd
3945 Specify the location of the executable for file deletion process.
3948 NAME: pinger_program
3950 DEFAULT: @DEFAULT_PINGER@
3951 LOC: Config.pinger.program
3954 Specify the location of the executable for the pinger process.
3960 LOC: Config.pinger.enable
3963 Control whether the pinger is active at run-time.
3964 Enables turning ICMP pinger on and off with a simple
3965 squid -k reconfigure.
3970 OPTIONS FOR URL REWRITING
3971 -----------------------------------------------------------------------------
3974 NAME: url_rewrite_program redirect_program
3976 LOC: Config.Program.redirect
3979 Specify the location of the executable URL rewriter to use.
3980 Since they can perform almost any function there isn't one included.
3982 For each requested URL, the rewriter will receive on line with the format
3984 URL <SP> client_ip "/" fqdn <SP> user <SP> method [<SP> kvpairs]<NL>
3986 In the future, the rewriter interface will be extended with
3987 key=value pairs ("kvpairs" shown above). Rewriter programs
3988 should be prepared to receive and possibly ignore additional
3989 whitespace-separated tokens on each input line.
3991 And the rewriter may return a rewritten URL. The other components of
3992 the request line does not need to be returned (ignored if they are).
3994 The rewriter can also indicate that a client-side redirect should
3995 be performed to the new URL. This is done by prefixing the returned
3996 URL with "301:" (moved permanently) or 302: (moved temporarily), etc.
3998 By default, a URL rewriter is not used.
4001 NAME: url_rewrite_children redirect_children
4002 TYPE: HelperChildConfig
4003 DEFAULT: 20 startup=0 idle=1 concurrency=0
4004 LOC: Config.redirectChildren
4006 The maximum number of redirector processes to spawn. If you limit
4007 it too few Squid will have to wait for them to process a backlog of
4008 URLs, slowing it down. If you allow too many they will use RAM
4009 and other system resources noticably.
4011 The startup= and idle= options allow some measure of skew in your
4016 Sets a minimum of how many processes are to be spawned when Squid
4017 starts or reconfigures. When set to zero the first request will
4018 cause spawning of the first child process to handle it.
4020 Starting too few will cause an initial slowdown in traffic as Squid
4021 attempts to simultaneously spawn enough processes to cope.
4025 Sets a minimum of how many processes Squid is to try and keep available
4026 at all times. When traffic begins to rise above what the existing
4027 processes can handle this many more will be spawned up to the maximum
4028 configured. A minimum setting of 1 is required.
4032 The number of requests each redirector helper can handle in
4033 parallel. Defaults to 0 which indicates the redirector
4034 is a old-style single threaded redirector.
4036 When this directive is set to a value >= 1 then the protocol
4037 used to communicate with the helper is modified to include
4038 a request ID in front of the request/response. The request
4039 ID from the request must be echoed back with the response
4043 NAME: url_rewrite_host_header redirect_rewrites_host_header
4046 LOC: Config.onoff.redir_rewrites_host
4048 To preserve same-origin security policies in browsers and
4049 prevent Host: header forgery by redirectors Squid rewrites
4050 any Host: header in redirected requests.
4052 If you are running an accelerator this may not be a wanted
4053 effect of a redirector. This directive enables you disable
4054 Host: alteration in reverse-proxy traffic.
4056 WARNING: Entries are cached on the result of the URL rewriting
4057 process, so be careful if you have domain-virtual hosts.
4059 WARNING: Squid and other software verifies the URL and Host
4060 are matching, so be careful not to relay through other proxies
4061 or inspecting firewalls with this disabled.
4064 NAME: url_rewrite_access redirector_access
4067 LOC: Config.accessList.redirector
4069 If defined, this access list specifies which requests are
4070 sent to the redirector processes. By default all requests
4073 This clause supports both fast and slow acl types.
4074 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
4077 NAME: url_rewrite_bypass redirector_bypass
4079 LOC: Config.onoff.redirector_bypass
4082 When this is 'on', a request will not go through the
4083 redirector if all redirectors are busy. If this is 'off'
4084 and the redirector queue grows too large, Squid will exit
4085 with a FATAL error and ask you to increase the number of
4086 redirectors. You should only enable this if the redirectors
4087 are not critical to your caching system. If you use
4088 redirectors for access control, and you enable this option,
4089 users may have access to pages they should not
4090 be allowed to request.
4094 OPTIONS FOR TUNING THE CACHE
4095 -----------------------------------------------------------------------------
4098 NAME: cache no_cache
4101 LOC: Config.accessList.noCache
4103 A list of ACL elements which, if matched and denied, cause the request to
4104 not be satisfied from the cache and the reply to not be cached.
4105 In other words, use this to force certain objects to never be cached.
4107 You must use the words 'allow' or 'deny' to indicate whether items
4108 matching the ACL should be allowed or denied into the cache.
4110 Default is to allow all to be cached.
4112 This clause supports both fast and slow acl types.
4113 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
4119 LOC: Config.maxStale
4122 This option puts an upper limit on how stale content Squid
4123 will serve from the cache if cache validation fails.
4124 Can be overriden by the refresh_pattern max-stale option.
4127 NAME: refresh_pattern
4128 TYPE: refreshpattern
4132 usage: refresh_pattern [-i] regex min percent max [options]
4134 By default, regular expressions are CASE-SENSITIVE. To make
4135 them case-insensitive, use the -i option.
4137 'Min' is the time (in minutes) an object without an explicit
4138 expiry time should be considered fresh. The recommended
4139 value is 0, any higher values may cause dynamic applications
4140 to be erroneously cached unless the application designer
4141 has taken the appropriate actions.
4143 'Percent' is a percentage of the objects age (time since last
4144 modification age) an object without explicit expiry time
4145 will be considered fresh.
4147 'Max' is an upper limit on how long objects without an explicit
4148 expiry time will be considered fresh.
4150 options: override-expire
4156 ignore-must-revalidate
4163 override-expire enforces min age even if the server
4164 sent an explicit expiry time (e.g., with the
4165 Expires: header or Cache-Control: max-age). Doing this
4166 VIOLATES the HTTP standard. Enabling this feature
4167 could make you liable for problems which it causes.
4169 Note: override-expire does not enforce staleness - it only extends
4170 freshness / min. If the server returns a Expires time which
4171 is longer than your max time, Squid will still consider
4172 the object fresh for that period of time.
4174 override-lastmod enforces min age even on objects
4175 that were modified recently.
4177 reload-into-ims changes client no-cache or ``reload''
4178 to If-Modified-Since requests. Doing this VIOLATES the
4179 HTTP standard. Enabling this feature could make you
4180 liable for problems which it causes.
4182 ignore-reload ignores a client no-cache or ``reload''
4183 header. Doing this VIOLATES the HTTP standard. Enabling
4184 this feature could make you liable for problems which
4187 ignore-no-cache ignores any ``Pragma: no-cache'' and
4188 ``Cache-control: no-cache'' headers received from a server.
4189 The HTTP RFC never allows the use of this (Pragma) header
4190 from a server, only a client, though plenty of servers
4193 ignore-no-store ignores any ``Cache-control: no-store''
4194 headers received from a server. Doing this VIOLATES
4195 the HTTP standard. Enabling this feature could make you
4196 liable for problems which it causes.
4198 ignore-must-revalidate ignores any ``Cache-Control: must-revalidate``
4199 headers received from a server. Doing this VIOLATES
4200 the HTTP standard. Enabling this feature could make you
4201 liable for problems which it causes.
4203 ignore-private ignores any ``Cache-control: private''
4204 headers received from a server. Doing this VIOLATES
4205 the HTTP standard. Enabling this feature could make you
4206 liable for problems which it causes.
4208 ignore-auth caches responses to requests with authorization,
4209 as if the originserver had sent ``Cache-control: public''
4210 in the response header. Doing this VIOLATES the HTTP standard.
4211 Enabling this feature could make you liable for problems which
4214 refresh-ims causes squid to contact the origin server
4215 when a client issues an If-Modified-Since request. This
4216 ensures that the client will receive an updated version
4217 if one is available.
4219 store-stale stores responses even if they don't have explicit
4220 freshness or a validator (i.e., Last-Modified or an ETag)
4221 present, or if they're already stale. By default, Squid will
4222 not cache such responses because they usually can't be
4223 reused. Note that such responses will be stale by default.
4225 max-stale=NN provide a maximum staleness factor. Squid won't
4226 serve objects more stale than this even if it failed to
4227 validate the object. Default: use the max_stale global limit.
4229 Basically a cached object is:
4231 FRESH if expires < now, else STALE
4233 FRESH if lm-factor < percent, else STALE
4237 The refresh_pattern lines are checked in the order listed here.
4238 The first entry which matches is used. If none of the entries
4239 match the default will be used.
4241 Note, you must uncomment all the default lines if you want
4242 to change one. The default setting is only active if none is
4247 # Add any of your own refresh_pattern entries above these.
4248 refresh_pattern ^ftp: 1440 20% 10080
4249 refresh_pattern ^gopher: 1440 0% 1440
4250 refresh_pattern -i (/cgi-bin/|\?) 0 0% 0
4251 refresh_pattern . 0 20% 4320
4255 NAME: quick_abort_min
4259 LOC: Config.quickAbort.min
4262 NAME: quick_abort_max
4266 LOC: Config.quickAbort.max
4269 NAME: quick_abort_pct
4273 LOC: Config.quickAbort.pct
4275 The cache by default continues downloading aborted requests
4276 which are almost completed (less than 16 KB remaining). This
4277 may be undesirable on slow (e.g. SLIP) links and/or very busy
4278 caches. Impatient users may tie up file descriptors and
4279 bandwidth by repeatedly requesting and immediately aborting
4282 When the user aborts a request, Squid will check the
4283 quick_abort values to the amount of data transfered until
4286 If the transfer has less than 'quick_abort_min' KB remaining,
4287 it will finish the retrieval.
4289 If the transfer has more than 'quick_abort_max' KB remaining,
4290 it will abort the retrieval.
4292 If more than 'quick_abort_pct' of the transfer has completed,
4293 it will finish the retrieval.
4295 If you do not want any retrieval to continue after the client
4296 has aborted, set both 'quick_abort_min' and 'quick_abort_max'
4299 If you want retrievals to always continue if they are being
4300 cached set 'quick_abort_min' to '-1 KB'.
4303 NAME: read_ahead_gap
4304 COMMENT: buffer-size
4306 LOC: Config.readAheadGap
4309 The amount of data the cache will buffer ahead of what has been
4310 sent to the client when retrieving an object from another server.
4314 IFDEF: USE_HTTP_VIOLATIONS
4317 LOC: Config.negativeTtl
4320 Set the Default Time-to-Live (TTL) for failed requests.
4321 Certain types of failures (such as "connection refused" and
4322 "404 Not Found") are able to be negatively-cached for a short time.
4323 Modern web servers should provide Expires: header, however if they
4324 do not this can provide a minimum TTL.
4325 The default is not to cache errors with unknown expiry details.
4327 Note that this is different from negative caching of DNS lookups.
4329 WARNING: Doing this VIOLATES the HTTP standard. Enabling
4330 this feature could make you liable for problems which it
4334 NAME: positive_dns_ttl
4337 LOC: Config.positiveDnsTtl
4340 Upper limit on how long Squid will cache positive DNS responses.
4341 Default is 6 hours (360 minutes). This directive must be set
4342 larger than negative_dns_ttl.
4345 NAME: negative_dns_ttl
4348 LOC: Config.negativeDnsTtl
4351 Time-to-Live (TTL) for negative caching of failed DNS lookups.
4352 This also sets the lower cache limit on positive lookups.
4353 Minimum value is 1 second, and it is not recommendable to go
4354 much below 10 seconds.
4357 NAME: range_offset_limit
4358 COMMENT: size [acl acl...]
4360 LOC: Config.rangeOffsetLimit
4363 usage: (size) [units] [[!]aclname]
4365 Sets an upper limit on how far (number of bytes) into the file
4366 a Range request may be to cause Squid to prefetch the whole file.
4367 If beyond this limit, Squid forwards the Range request as it is and
4368 the result is NOT cached.
4370 This is to stop a far ahead range request (lets say start at 17MB)
4371 from making Squid fetch the whole object up to that point before
4372 sending anything to the client.
4374 Multiple range_offset_limit lines may be specified, and they will
4375 be searched from top to bottom on each request until a match is found.
4376 The first match found will be used. If no line matches a request, the
4377 default limit of 0 bytes will be used.
4379 'size' is the limit specified as a number of units.
4381 'units' specifies whether to use bytes, KB, MB, etc.
4382 If no units are specified bytes are assumed.
4384 A size of 0 causes Squid to never fetch more than the
4385 client requested. (default)
4387 A size of 'none' causes Squid to always fetch the object from the
4388 beginning so it may cache the result. (2.0 style)
4390 'aclname' is the name of a defined ACL.
4392 NP: Using 'none' as the byte value here will override any quick_abort settings
4393 that may otherwise apply to the range request. The range request will
4394 be fully fetched from start to finish regardless of the client
4395 actions. This affects bandwidth usage.
4398 NAME: minimum_expiry_time
4401 LOC: Config.minimum_expiry_time
4404 The minimum caching time according to (Expires - Date)
4405 Headers Squid honors if the object can't be revalidated
4406 defaults to 60 seconds. In reverse proxy environments it
4407 might be desirable to honor shorter object lifetimes. It
4408 is most likely better to make your server return a
4409 meaningful Last-Modified header however. In ESI environments
4410 where page fragments often have short lifetimes, this will
4411 often be best set to 0.
4414 NAME: store_avg_object_size
4418 LOC: Config.Store.avgObjectSize
4420 Average object size, used to estimate number of objects your
4421 cache can hold. The default is 13 KB.
4424 NAME: store_objects_per_bucket
4427 LOC: Config.Store.objectsPerBucket
4429 Target number of objects per bucket in the store hash table.
4430 Lowering this value increases the total number of buckets and
4431 also the storage maintenance rate. The default is 20.
4436 -----------------------------------------------------------------------------
4439 NAME: request_header_max_size
4443 LOC: Config.maxRequestHeaderSize
4445 This specifies the maximum size for HTTP headers in a request.
4446 Request headers are usually relatively small (about 512 bytes).
4447 Placing a limit on the request header size will catch certain
4448 bugs (for example with persistent connections) and possibly
4449 buffer-overflow or denial-of-service attacks.
4452 NAME: reply_header_max_size
4456 LOC: Config.maxReplyHeaderSize
4458 This specifies the maximum size for HTTP headers in a reply.
4459 Reply headers are usually relatively small (about 512 bytes).
4460 Placing a limit on the reply header size will catch certain
4461 bugs (for example with persistent connections) and possibly
4462 buffer-overflow or denial-of-service attacks.
4465 NAME: request_body_max_size
4469 LOC: Config.maxRequestBodySize
4471 This specifies the maximum size for an HTTP request body.
4472 In other words, the maximum size of a PUT/POST request.
4473 A user who attempts to send a request with a body larger
4474 than this limit receives an "Invalid Request" error message.
4475 If you set this parameter to a zero (the default), there will
4476 be no limit imposed.
4479 NAME: client_request_buffer_max_size
4483 LOC: Config.maxRequestBufferSize
4485 This specifies the maximum buffer size of a client request.
4486 It prevents squid eating too much memory when somebody uploads
4490 NAME: chunked_request_body_max_size
4494 LOC: Config.maxChunkedRequestBodySize
4496 A broken or confused HTTP/1.1 client may send a chunked HTTP
4497 request to Squid. Squid does not have full support for that
4498 feature yet. To cope with such requests, Squid buffers the
4499 entire request and then dechunks request body to create a
4500 plain HTTP/1.0 request with a known content length. The plain
4501 request is then used by the rest of Squid code as usual.
4503 The option value specifies the maximum size of the buffer used
4504 to hold the request before the conversion. If the chunked
4505 request size exceeds the specified limit, the conversion
4506 fails, and the client receives an "unsupported request" error,
4507 as if dechunking was disabled.
4509 Dechunking is enabled by default. To disable conversion of
4510 chunked requests, set the maximum to zero.
4512 Request dechunking feature and this option in particular are a
4513 temporary hack. When chunking requests and responses are fully
4514 supported, there will be no need to buffer a chunked request.
4518 IFDEF: USE_HTTP_VIOLATIONS
4521 LOC: Config.accessList.brokenPosts
4523 A list of ACL elements which, if matched, causes Squid to send
4524 an extra CRLF pair after the body of a PUT/POST request.
4526 Some HTTP servers has broken implementations of PUT/POST,
4527 and rely on an extra CRLF pair sent by some WWW clients.
4529 Quote from RFC2616 section 4.1 on this matter:
4531 Note: certain buggy HTTP/1.0 client implementations generate an
4532 extra CRLF's after a POST request. To restate what is explicitly
4533 forbidden by the BNF, an HTTP/1.1 client must not preface or follow
4534 a request with an extra CRLF.
4536 This clause only supports fast acl types.
4537 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
4540 acl buggy_server url_regex ^http://....
4541 broken_posts allow buggy_server
4544 NAME: adaptation_uses_indirect_client icap_uses_indirect_client
4547 IFDEF: FOLLOW_X_FORWARDED_FOR&&USE_ADAPTATION
4549 LOC: Adaptation::Config::use_indirect_client
4551 Controls whether the indirect client IP address (instead of the direct
4552 client IP address) is passed to adaptation services.
4554 See also: follow_x_forwarded_for adaptation_send_client_ip
4558 IFDEF: USE_HTTP_VIOLATIONS
4562 LOC: Config.onoff.via
4564 If set (default), Squid will include a Via header in requests and
4565 replies as required by RFC2616.
4571 LOC: Config.onoff.ie_refresh
4574 Microsoft Internet Explorer up until version 5.5 Service
4575 Pack 1 has an issue with transparent proxies, wherein it
4576 is impossible to force a refresh. Turning this on provides
4577 a partial fix to the problem, by causing all IMS-REFRESH
4578 requests from older IE versions to check the origin server
4579 for fresh content. This reduces hit ratio by some amount
4580 (~10% in my experience), but allows users to actually get
4581 fresh content when they want it. Note because Squid
4582 cannot tell if the user is using 5.5 or 5.5SP1, the behavior
4583 of 5.5 is unchanged from old versions of Squid (i.e. a
4584 forced refresh is impossible). Newer versions of IE will,
4585 hopefully, continue to have the new behavior and will be
4586 handled based on that assumption. This option defaults to
4587 the old Squid behavior, which is better for hit ratios but
4588 worse for clients using IE, if they need to be able to
4589 force fresh content.
4592 NAME: vary_ignore_expire
4595 LOC: Config.onoff.vary_ignore_expire
4598 Many HTTP servers supporting Vary gives such objects
4599 immediate expiry time with no cache-control header
4600 when requested by a HTTP/1.0 client. This option
4601 enables Squid to ignore such expiry times until
4602 HTTP/1.1 is fully implemented.
4604 WARNING: If turned on this may eventually cause some
4605 varying objects not intended for caching to get cached.
4608 NAME: request_entities
4610 LOC: Config.onoff.request_entities
4613 Squid defaults to deny GET and HEAD requests with request entities,
4614 as the meaning of such requests are undefined in the HTTP standard
4615 even if not explicitly forbidden.
4617 Set this directive to on if you have clients which insists
4618 on sending request entities in GET or HEAD requests. But be warned
4619 that there is server software (both proxies and web servers) which
4620 can fail to properly process this kind of request which may make you
4621 vulnerable to cache pollution attacks if enabled.
4624 NAME: request_header_access
4625 IFDEF: USE_HTTP_VIOLATIONS
4626 TYPE: http_header_access[]
4627 LOC: Config.request_header_access
4630 Usage: request_header_access header_name allow|deny [!]aclname ...
4632 WARNING: Doing this VIOLATES the HTTP standard. Enabling
4633 this feature could make you liable for problems which it
4636 This option replaces the old 'anonymize_headers' and the
4637 older 'http_anonymizer' option with something that is much
4638 more configurable. This new method creates a list of ACLs
4639 for each header, allowing you very fine-tuned header
4642 This option only applies to request headers, i.e., from the
4643 client to the server.
4645 You can only specify known headers for the header name.
4646 Other headers are reclassified as 'Other'. You can also
4647 refer to all the headers with 'All'.
4649 For example, to achieve the same behavior as the old
4650 'http_anonymizer standard' option, you should use:
4652 request_header_access From deny all
4653 request_header_access Referer deny all
4654 request_header_access Server deny all
4655 request_header_access User-Agent deny all
4656 request_header_access WWW-Authenticate deny all
4657 request_header_access Link deny all
4659 Or, to reproduce the old 'http_anonymizer paranoid' feature
4662 request_header_access Allow allow all
4663 request_header_access Authorization allow all
4664 request_header_access WWW-Authenticate allow all
4665 request_header_access Proxy-Authorization allow all
4666 request_header_access Proxy-Authenticate allow all
4667 request_header_access Cache-Control allow all
4668 request_header_access Content-Encoding allow all
4669 request_header_access Content-Length allow all
4670 request_header_access Content-Type allow all
4671 request_header_access Date allow all
4672 request_header_access Expires allow all
4673 request_header_access Host allow all
4674 request_header_access If-Modified-Since allow all
4675 request_header_access Last-Modified allow all
4676 request_header_access Location allow all
4677 request_header_access Pragma allow all
4678 request_header_access Accept allow all
4679 request_header_access Accept-Charset allow all
4680 request_header_access Accept-Encoding allow all
4681 request_header_access Accept-Language allow all
4682 request_header_access Content-Language allow all
4683 request_header_access Mime-Version allow all
4684 request_header_access Retry-After allow all
4685 request_header_access Title allow all
4686 request_header_access Connection allow all
4687 request_header_access All deny all
4689 although many of those are HTTP reply headers, and so should be
4690 controlled with the reply_header_access directive.
4692 By default, all headers are allowed (no anonymizing is
4696 NAME: reply_header_access
4697 IFDEF: USE_HTTP_VIOLATIONS
4698 TYPE: http_header_access[]
4699 LOC: Config.reply_header_access
4702 Usage: reply_header_access header_name allow|deny [!]aclname ...
4704 WARNING: Doing this VIOLATES the HTTP standard. Enabling
4705 this feature could make you liable for problems which it
4708 This option only applies to reply headers, i.e., from the
4709 server to the client.
4711 This is the same as request_header_access, but in the other
4714 This option replaces the old 'anonymize_headers' and the
4715 older 'http_anonymizer' option with something that is much
4716 more configurable. This new method creates a list of ACLs
4717 for each header, allowing you very fine-tuned header
4720 You can only specify known headers for the header name.
4721 Other headers are reclassified as 'Other'. You can also
4722 refer to all the headers with 'All'.
4724 For example, to achieve the same behavior as the old
4725 'http_anonymizer standard' option, you should use:
4727 reply_header_access From deny all
4728 reply_header_access Referer deny all
4729 reply_header_access Server deny all
4730 reply_header_access User-Agent deny all
4731 reply_header_access WWW-Authenticate deny all
4732 reply_header_access Link deny all
4734 Or, to reproduce the old 'http_anonymizer paranoid' feature
4737 reply_header_access Allow allow all
4738 reply_header_access Authorization allow all
4739 reply_header_access WWW-Authenticate allow all
4740 reply_header_access Proxy-Authorization allow all
4741 reply_header_access Proxy-Authenticate allow all
4742 reply_header_access Cache-Control allow all
4743 reply_header_access Content-Encoding allow all
4744 reply_header_access Content-Length allow all
4745 reply_header_access Content-Type allow all
4746 reply_header_access Date allow all
4747 reply_header_access Expires allow all
4748 reply_header_access Host allow all
4749 reply_header_access If-Modified-Since allow all
4750 reply_header_access Last-Modified allow all
4751 reply_header_access Location allow all
4752 reply_header_access Pragma allow all
4753 reply_header_access Accept allow all
4754 reply_header_access Accept-Charset allow all
4755 reply_header_access Accept-Encoding allow all
4756 reply_header_access Accept-Language allow all
4757 reply_header_access Content-Language allow all
4758 reply_header_access Mime-Version allow all
4759 reply_header_access Retry-After allow all
4760 reply_header_access Title allow all
4761 reply_header_access Connection allow all
4762 reply_header_access All deny all
4764 although the HTTP request headers won't be usefully controlled
4765 by this directive -- see request_header_access for details.
4767 By default, all headers are allowed (no anonymizing is
4771 NAME: request_header_replace header_replace
4772 IFDEF: USE_HTTP_VIOLATIONS
4773 TYPE: http_header_replace[]
4774 LOC: Config.request_header_access
4777 Usage: request_header_replace header_name message
4778 Example: request_header_replace User-Agent Nutscrape/1.0 (CP/M; 8-bit)
4780 This option allows you to change the contents of headers
4781 denied with request_header_access above, by replacing them
4782 with some fixed string. This replaces the old fake_user_agent
4785 This only applies to request headers, not reply headers.
4787 By default, headers are removed if denied.
4790 NAME: reply_header_replace
4791 IFDEF: USE_HTTP_VIOLATIONS
4792 TYPE: http_header_replace[]
4793 LOC: Config.reply_header_access
4796 Usage: reply_header_replace header_name message
4797 Example: reply_header_replace Server Foo/1.0
4799 This option allows you to change the contents of headers
4800 denied with reply_header_access above, by replacing them
4801 with some fixed string.
4803 This only applies to reply headers, not request headers.
4805 By default, headers are removed if denied.
4808 NAME: relaxed_header_parser
4809 COMMENT: on|off|warn
4811 LOC: Config.onoff.relaxed_header_parser
4814 In the default "on" setting Squid accepts certain forms
4815 of non-compliant HTTP messages where it is unambiguous
4816 what the sending application intended even if the message
4817 is not correctly formatted. The messages is then normalized
4818 to the correct form when forwarded by Squid.
4820 If set to "warn" then a warning will be emitted in cache.log
4821 each time such HTTP error is encountered.
4823 If set to "off" then such HTTP errors will cause the request
4824 or response to be rejected.
4829 -----------------------------------------------------------------------------
4832 NAME: forward_timeout
4835 LOC: Config.Timeout.forward
4838 This parameter specifies how long Squid should at most attempt in
4839 finding a forwarding path for the request before giving up.
4842 NAME: connect_timeout
4845 LOC: Config.Timeout.connect
4848 This parameter specifies how long to wait for the TCP connect to
4849 the requested server or peer to complete before Squid should
4850 attempt to find another path where to forward the request.
4853 NAME: peer_connect_timeout
4856 LOC: Config.Timeout.peer_connect
4859 This parameter specifies how long to wait for a pending TCP
4860 connection to a peer cache. The default is 30 seconds. You
4861 may also set different timeout values for individual neighbors
4862 with the 'connect-timeout' option on a 'cache_peer' line.
4868 LOC: Config.Timeout.read
4871 The read_timeout is applied on server-side connections. After
4872 each successful read(), the timeout will be extended by this
4873 amount. If no data is read again after this amount of time,
4874 the request is aborted and logged with ERR_READ_TIMEOUT. The
4875 default is 15 minutes.
4881 LOC: Config.Timeout.write
4884 This timeout is tracked for all connections that have data
4885 available for writing and are waiting for the socket to become
4886 ready. After each successful write, the timeout is extended by
4887 the configured amount. If Squid has data to write but the
4888 connection is not ready for the configured duration, the
4889 transaction associated with the connection is terminated. The
4890 default is 15 minutes.
4893 NAME: request_timeout
4895 LOC: Config.Timeout.request
4898 How long to wait for complete HTTP request headers after initial
4899 connection establishment.
4902 NAME: client_idle_pconn_timeout persistent_request_timeout
4904 LOC: Config.Timeout.clientIdlePconn
4907 How long to wait for the next HTTP request on a persistent
4908 client connection after the previous request completes.
4911 NAME: client_lifetime
4914 LOC: Config.Timeout.lifetime
4917 The maximum amount of time a client (browser) is allowed to
4918 remain connected to the cache process. This protects the Cache
4919 from having a lot of sockets (and hence file descriptors) tied up
4920 in a CLOSE_WAIT state from remote clients that go away without
4921 properly shutting down (either because of a network failure or
4922 because of a poor client implementation). The default is one
4925 NOTE: The default value is intended to be much larger than any
4926 client would ever need to be connected to your cache. You
4927 should probably change client_lifetime only as a last resort.
4928 If you seem to have many client connections tying up
4929 filedescriptors, we recommend first tuning the read_timeout,
4930 request_timeout, persistent_request_timeout and quick_abort values.
4933 NAME: half_closed_clients
4935 LOC: Config.onoff.half_closed_clients
4938 Some clients may shutdown the sending side of their TCP
4939 connections, while leaving their receiving sides open. Sometimes,
4940 Squid can not tell the difference between a half-closed and a
4941 fully-closed TCP connection.
4943 By default, Squid will immediately close client connections when
4944 read(2) returns "no more data to read."
4946 Change this option to 'on' and Squid will keep open connections
4947 until a read(2) or write(2) on the socket returns an error.
4948 This may show some benefits for reverse proxies. But if not
4949 it is recommended to leave OFF.
4952 NAME: server_idle_pconn_timeout pconn_timeout
4954 LOC: Config.Timeout.serverIdlePconn
4957 Timeout for idle persistent connections to servers and other
4964 LOC: Ident::TheConfig.timeout
4967 Maximum time to wait for IDENT lookups to complete.
4969 If this is too high, and you enabled IDENT lookups from untrusted
4970 users, you might be susceptible to denial-of-service by having
4971 many ident requests going at once.
4974 NAME: shutdown_lifetime
4977 LOC: Config.shutdownLifetime
4980 When SIGTERM or SIGHUP is received, the cache is put into
4981 "shutdown pending" mode until all active sockets are closed.
4982 This value is the lifetime to set for all open descriptors
4983 during shutdown mode. Any active clients after this many
4984 seconds will receive a 'timeout' message.
4988 ADMINISTRATIVE PARAMETERS
4989 -----------------------------------------------------------------------------
4995 LOC: Config.adminEmail
4997 Email-address of local cache manager who will receive
4998 mail if the cache dies. The default is "webmaster."
5004 LOC: Config.EmailFrom
5006 From: email-address for mail sent when the cache dies.
5007 The default is to use 'appname@unique_hostname'.
5008 Default appname value is "squid", can be changed into
5009 src/globals.h before building squid.
5015 LOC: Config.EmailProgram
5017 Email program used to send mail if the cache dies.
5018 The default is "mail". The specified program must comply
5019 with the standard Unix mail syntax:
5020 mail-program recipient < mailfile
5022 Optional command line options can be specified.
5025 NAME: cache_effective_user
5027 DEFAULT: @DEFAULT_CACHE_EFFECTIVE_USER@
5028 LOC: Config.effectiveUser
5030 If you start Squid as root, it will change its effective/real
5031 UID/GID to the user specified below. The default is to change
5032 to UID of @DEFAULT_CACHE_EFFECTIVE_USER@.
5033 see also; cache_effective_group
5036 NAME: cache_effective_group
5039 LOC: Config.effectiveGroup
5041 Squid sets the GID to the effective user's default group ID
5042 (taken from the password file) and supplementary group list
5043 from the groups membership.
5045 If you want Squid to run with a specific GID regardless of
5046 the group memberships of the effective user then set this
5047 to the group (or GID) you want Squid to run as. When set
5048 all other group privileges of the effective user are ignored
5049 and only this GID is effective. If Squid is not started as
5050 root the user starting Squid MUST be member of the specified
5053 This option is not recommended by the Squid Team.
5054 Our preference is for administrators to configure a secure
5055 user account for squid with UID/GID matching system policies.
5058 NAME: httpd_suppress_version_string
5062 LOC: Config.onoff.httpd_suppress_version_string
5064 Suppress Squid version string info in HTTP headers and HTML error pages.
5067 NAME: visible_hostname
5069 LOC: Config.visibleHostname
5072 If you want to present a special hostname in error messages, etc,
5073 define this. Otherwise, the return value of gethostname()
5074 will be used. If you have multiple caches in a cluster and
5075 get errors about IP-forwarding you must set them to have individual
5076 names with this setting.
5079 NAME: unique_hostname
5081 LOC: Config.uniqueHostname
5084 If you want to have multiple machines with the same
5085 'visible_hostname' you must give each machine a different
5086 'unique_hostname' so forwarding loops can be detected.
5089 NAME: hostname_aliases
5091 LOC: Config.hostnameAliases
5094 A list of other DNS names your cache has.
5102 Minimum umask which should be enforced while the proxy
5103 is running, in addition to the umask set at startup.
5105 For a traditional octal representation of umasks, start
5110 OPTIONS FOR THE CACHE REGISTRATION SERVICE
5111 -----------------------------------------------------------------------------
5113 This section contains parameters for the (optional) cache
5114 announcement service. This service is provided to help
5115 cache administrators locate one another in order to join or
5116 create cache hierarchies.
5118 An 'announcement' message is sent (via UDP) to the registration
5119 service by Squid. By default, the announcement message is NOT
5120 SENT unless you enable it with 'announce_period' below.
5122 The announcement message includes your hostname, plus the
5123 following information from this configuration file:
5129 All current information is processed regularly and made
5130 available on the Web at http://www.ircache.net/Cache/Tracker/.
5133 NAME: announce_period
5135 LOC: Config.Announce.period
5138 This is how frequently to send cache announcements. The
5139 default is `0' which disables sending the announcement
5142 To enable announcing your cache, just set an announce period.
5145 announce_period 1 day
5150 DEFAULT: tracker.ircache.net
5151 LOC: Config.Announce.host
5157 LOC: Config.Announce.file
5163 LOC: Config.Announce.port
5165 announce_host and announce_port set the hostname and port
5166 number where the registration message will be sent.
5168 Hostname will default to 'tracker.ircache.net' and port will
5169 default default to 3131. If the 'filename' argument is given,
5170 the contents of that file will be included in the announce
5175 HTTPD-ACCELERATOR OPTIONS
5176 -----------------------------------------------------------------------------
5179 NAME: httpd_accel_surrogate_id
5182 LOC: Config.Accel.surrogate_id
5184 Surrogates (http://www.esi.org/architecture_spec_1.0.html)
5185 need an identification token to allow control targeting. Because
5186 a farm of surrogates may all perform the same tasks, they may share
5187 an identification token.
5189 The default ID is the visible_hostname
5192 NAME: http_accel_surrogate_remote
5196 LOC: Config.onoff.surrogate_is_remote
5198 Remote surrogates (such as those in a CDN) honour Surrogate-Control: no-store-remote.
5199 Set this to on to have squid behave as a remote surrogate.
5203 IFDEF: USE_SQUID_ESI
5204 COMMENT: libxml2|expat|custom
5206 LOC: ESIParser::Type
5209 ESI markup is not strictly XML compatible. The custom ESI parser
5210 will give higher performance, but cannot handle non ASCII character
5215 DELAY POOL PARAMETERS
5216 -----------------------------------------------------------------------------
5220 TYPE: delay_pool_count
5222 IFDEF: USE_DELAY_POOLS
5225 This represents the number of delay pools to be used. For example,
5226 if you have one class 2 delay pool and one class 3 delays pool, you
5227 have a total of 2 delay pools.
5231 TYPE: delay_pool_class
5233 IFDEF: USE_DELAY_POOLS
5236 This defines the class of each delay pool. There must be exactly one
5237 delay_class line for each delay pool. For example, to define two
5238 delay pools, one of class 2 and one of class 3, the settings above
5242 delay_pools 4 # 4 delay pools
5243 delay_class 1 2 # pool 1 is a class 2 pool
5244 delay_class 2 3 # pool 2 is a class 3 pool
5245 delay_class 3 4 # pool 3 is a class 4 pool
5246 delay_class 4 5 # pool 4 is a class 5 pool
5248 The delay pool classes are:
5250 class 1 Everything is limited by a single aggregate
5253 class 2 Everything is limited by a single aggregate
5254 bucket as well as an "individual" bucket chosen
5255 from bits 25 through 32 of the IPv4 address.
5257 class 3 Everything is limited by a single aggregate
5258 bucket as well as a "network" bucket chosen
5259 from bits 17 through 24 of the IP address and a
5260 "individual" bucket chosen from bits 17 through
5261 32 of the IPv4 address.
5263 class 4 Everything in a class 3 delay pool, with an
5264 additional limit on a per user basis. This
5265 only takes effect if the username is established
5266 in advance - by forcing authentication in your
5269 class 5 Requests are grouped according their tag (see
5270 external_acl's tag= reply).
5273 Each pool also requires a delay_parameters directive to configure the pool size
5274 and speed limits used whenever the pool is applied to a request. Along with
5275 a set of delay_access directives to determine when it is used.
5277 NOTE: If an IP address is a.b.c.d
5278 -> bits 25 through 32 are "d"
5279 -> bits 17 through 24 are "c"
5280 -> bits 17 through 32 are "c * 256 + d"
5282 NOTE-2: Due to the use of bitmasks in class 2,3,4 pools they only apply to
5283 IPv4 traffic. Class 1 and 5 pools may be used with IPv6 traffic.
5287 TYPE: delay_pool_access
5289 IFDEF: USE_DELAY_POOLS
5292 This is used to determine which delay pool a request falls into.
5294 delay_access is sorted per pool and the matching starts with pool 1,
5295 then pool 2, ..., and finally pool N. The first delay pool where the
5296 request is allowed is selected for the request. If it does not allow
5297 the request to any pool then the request is not delayed (default).
5299 For example, if you want some_big_clients in delay
5300 pool 1 and lotsa_little_clients in delay pool 2:
5303 delay_access 1 allow some_big_clients
5304 delay_access 1 deny all
5305 delay_access 2 allow lotsa_little_clients
5306 delay_access 2 deny all
5307 delay_access 3 allow authenticated_clients
5310 NAME: delay_parameters
5311 TYPE: delay_pool_rates
5313 IFDEF: USE_DELAY_POOLS
5316 This defines the parameters for a delay pool. Each delay pool has
5317 a number of "buckets" associated with it, as explained in the
5318 description of delay_class.
5320 For a class 1 delay pool, the syntax is:
5322 delay_parameters pool aggregate
5324 For a class 2 delay pool:
5326 delay_parameters pool aggregate individual
5328 For a class 3 delay pool:
5330 delay_parameters pool aggregate network individual
5332 For a class 4 delay pool:
5334 delay_parameters pool aggregate network individual user
5336 For a class 5 delay pool:
5338 delay_parameters pool tagrate
5340 The option variables are:
5342 pool a pool number - ie, a number between 1 and the
5343 number specified in delay_pools as used in
5346 aggregate the speed limit parameters for the aggregate bucket
5349 individual the speed limit parameters for the individual
5350 buckets (class 2, 3).
5352 network the speed limit parameters for the network buckets
5355 user the speed limit parameters for the user buckets
5358 tagrate the speed limit parameters for the tag buckets
5361 A pair of delay parameters is written restore/maximum, where restore is
5362 the number of bytes (not bits - modem and network speeds are usually
5363 quoted in bits) per second placed into the bucket, and maximum is the
5364 maximum number of bytes which can be in the bucket at any time.
5366 There must be one delay_parameters line for each delay pool.
5369 For example, if delay pool number 1 is a class 2 delay pool as in the
5370 above example, and is being used to strictly limit each host to 64Kbit/sec
5371 (plus overheads), with no overall limit, the line is:
5373 delay_parameters 1 -1/-1 8000/8000
5375 Note that 8 x 8000 KByte/sec -> 64Kbit/sec.
5377 Note that the figure -1 is used to represent "unlimited".
5380 And, if delay pool number 2 is a class 3 delay pool as in the above
5381 example, and you want to limit it to a total of 256Kbit/sec (strict limit)
5382 with each 8-bit network permitted 64Kbit/sec (strict limit) and each
5383 individual host permitted 4800bit/sec with a bucket maximum size of 64Kbits
5384 to permit a decent web page to be downloaded at a decent speed
5385 (if the network is not being limited due to overuse) but slow down
5386 large downloads more significantly:
5388 delay_parameters 2 32000/32000 8000/8000 600/8000
5390 Note that 8 x 32000 KByte/sec -> 256Kbit/sec.
5391 8 x 8000 KByte/sec -> 64Kbit/sec.
5392 8 x 600 Byte/sec -> 4800bit/sec.
5395 Finally, for a class 4 delay pool as in the example - each user will
5396 be limited to 128Kbits/sec no matter how many workstations they are logged into.:
5398 delay_parameters 4 32000/32000 8000/8000 600/64000 16000/16000
5401 NAME: delay_initial_bucket_level
5402 COMMENT: (percent, 0-100)
5405 IFDEF: USE_DELAY_POOLS
5406 LOC: Config.Delay.initial
5408 The initial bucket percentage is used to determine how much is put
5409 in each bucket when squid starts, is reconfigured, or first notices
5410 a host accessing it (in class 2 and class 3, individual hosts and
5411 networks only have buckets associated with them once they have been
5416 CLIENT DELAY POOL PARAMETERS
5417 -----------------------------------------------------------------------------
5420 NAME: client_delay_pools
5421 TYPE: client_delay_pool_count
5423 IFDEF: USE_DELAY_POOLS
5424 LOC: Config.ClientDelay
5426 This option specifies the number of client delay pools used. It must
5427 preceed other client_delay_* options.
5430 client_delay_pools 2
5433 NAME: client_delay_initial_bucket_level
5434 COMMENT: (percent, 0-no_limit)
5437 IFDEF: USE_DELAY_POOLS
5438 LOC: Config.ClientDelay.initial
5440 This option determines the initial bucket size as a percentage of
5441 max_bucket_size from client_delay_parameters. Buckets are created
5442 at the time of the "first" connection from the matching IP. Idle
5443 buckets are periodically deleted up.
5445 You can specify more than 100 percent but note that such "oversized"
5446 buckets are not refilled until their size goes down to max_bucket_size
5447 from client_delay_parameters.
5450 client_delay_initial_bucket_level 50
5453 NAME: client_delay_parameters
5454 TYPE: client_delay_pool_rates
5456 IFDEF: USE_DELAY_POOLS
5457 LOC: Config.ClientDelay
5460 This option configures client-side bandwidth limits using the
5463 client_delay_parameters pool speed_limit max_bucket_size
5465 pool is an integer ID used for client_delay_access matching.
5467 speed_limit is bytes added to the bucket per second.
5469 max_bucket_size is the maximum size of a bucket, enforced after any
5470 speed_limit additions.
5472 Please see the delay_parameters option for more information and
5476 client_delay_parameters 1 1024 2048
5477 client_delay_parameters 2 51200 16384
5480 NAME: client_delay_access
5481 TYPE: client_delay_pool_access
5483 IFDEF: USE_DELAY_POOLS
5484 LOC: Config.ClientDelay
5487 This option determines the client-side delay pool for the
5490 client_delay_access pool_ID allow|deny acl_name
5492 All client_delay_access options are checked in their pool ID
5493 order, starting with pool 1. The first checked pool with allowed
5494 request is selected for the request. If no ACL matches or there
5495 are no client_delay_access options, the request bandwidth is not
5498 The ACL-selected pool is then used to find the
5499 client_delay_parameters for the request. Client-side pools are
5500 not used to aggregate clients. Clients are always aggregated
5501 based on their source IP addresses (one bucket per source IP).
5503 Please see delay_access for more examples.
5506 client_delay_access 1 allow low_rate_network
5507 client_delay_access 2 allow vips_network
5511 WCCPv1 AND WCCPv2 CONFIGURATION OPTIONS
5512 -----------------------------------------------------------------------------
5517 LOC: Config.Wccp.router
5521 Use this option to define your WCCP ``home'' router for
5524 wccp_router supports a single WCCP(v1) router
5526 wccp2_router supports multiple WCCPv2 routers
5528 only one of the two may be used at the same time and defines
5529 which version of WCCP to use.
5533 TYPE: IpAddress_list
5534 LOC: Config.Wccp2.router
5538 Use this option to define your WCCP ``home'' router for
5541 wccp_router supports a single WCCP(v1) router
5543 wccp2_router supports multiple WCCPv2 routers
5545 only one of the two may be used at the same time and defines
5546 which version of WCCP to use.
5551 LOC: Config.Wccp.version
5555 This directive is only relevant if you need to set up WCCP(v1)
5556 to some very old and end-of-life Cisco routers. In all other
5557 setups it must be left unset or at the default setting.
5558 It defines an internal version in the WCCP(v1) protocol,
5559 with version 4 being the officially documented protocol.
5561 According to some users, Cisco IOS 11.2 and earlier only
5562 support WCCP version 3. If you're using that or an earlier
5563 version of IOS, you may need to change this value to 3, otherwise
5564 do not specify this parameter.
5567 NAME: wccp2_rebuild_wait
5569 LOC: Config.Wccp2.rebuildwait
5573 If this is enabled Squid will wait for the cache dir rebuild to finish
5574 before sending the first wccp2 HereIAm packet
5577 NAME: wccp2_forwarding_method
5579 LOC: Config.Wccp2.forwarding_method
5583 WCCP2 allows the setting of forwarding methods between the
5584 router/switch and the cache. Valid values are as follows:
5586 gre - GRE encapsulation (forward the packet in a GRE/WCCP tunnel)
5587 l2 - L2 redirect (forward the packet using Layer 2/MAC rewriting)
5589 Currently (as of IOS 12.4) cisco routers only support GRE.
5590 Cisco switches only support the L2 redirect assignment method.
5593 NAME: wccp2_return_method
5595 LOC: Config.Wccp2.return_method
5599 WCCP2 allows the setting of return methods between the
5600 router/switch and the cache for packets that the cache
5601 decides not to handle. Valid values are as follows:
5603 gre - GRE encapsulation (forward the packet in a GRE/WCCP tunnel)
5604 l2 - L2 redirect (forward the packet using Layer 2/MAC rewriting)
5606 Currently (as of IOS 12.4) cisco routers only support GRE.
5607 Cisco switches only support the L2 redirect assignment.
5609 If the "ip wccp redirect exclude in" command has been
5610 enabled on the cache interface, then it is still safe for
5611 the proxy server to use a l2 redirect method even if this
5612 option is set to GRE.
5615 NAME: wccp2_assignment_method
5617 LOC: Config.Wccp2.assignment_method
5621 WCCP2 allows the setting of methods to assign the WCCP hash
5622 Valid values are as follows:
5624 hash - Hash assignment
5625 mask - Mask assignment
5627 As a general rule, cisco routers support the hash assignment method
5628 and cisco switches support the mask assignment method.
5633 LOC: Config.Wccp2.info
5634 DEFAULT_IF_NONE: standard 0
5637 WCCP2 allows for multiple traffic services. There are two
5638 types: "standard" and "dynamic". The standard type defines
5639 one service id - http (id 0). The dynamic service ids can be from
5640 51 to 255 inclusive. In order to use a dynamic service id
5641 one must define the type of traffic to be redirected; this is done
5642 using the wccp2_service_info option.
5644 The "standard" type does not require a wccp2_service_info option,
5645 just specifying the service id will suffice.
5647 MD5 service authentication can be enabled by adding
5648 "password=<password>" to the end of this service declaration.
5652 wccp2_service standard 0 # for the 'web-cache' standard service
5653 wccp2_service dynamic 80 # a dynamic service type which will be
5654 # fleshed out with subsequent options.
5655 wccp2_service standard 0 password=foo
5658 NAME: wccp2_service_info
5659 TYPE: wccp2_service_info
5660 LOC: Config.Wccp2.info
5664 Dynamic WCCPv2 services require further information to define the
5665 traffic you wish to have diverted.
5669 wccp2_service_info <id> protocol=<protocol> flags=<flag>,<flag>..
5670 priority=<priority> ports=<port>,<port>..
5672 The relevant WCCPv2 flags:
5673 + src_ip_hash, dst_ip_hash
5674 + source_port_hash, dst_port_hash
5675 + src_ip_alt_hash, dst_ip_alt_hash
5676 + src_port_alt_hash, dst_port_alt_hash
5679 The port list can be one to eight entries.
5683 wccp2_service_info 80 protocol=tcp flags=src_ip_hash,ports_source
5684 priority=240 ports=80
5686 Note: the service id must have been defined by a previous
5687 'wccp2_service dynamic <id>' entry.
5692 LOC: Config.Wccp2.weight
5696 Each cache server gets assigned a set of the destination
5697 hash proportional to their weight.
5702 LOC: Config.Wccp.address
5709 LOC: Config.Wccp2.address
5713 Use this option if you require WCCP to use a specific
5716 The default behavior is to not bind to any specific address.
5720 PERSISTENT CONNECTION HANDLING
5721 -----------------------------------------------------------------------------
5723 Also see "pconn_timeout" in the TIMEOUTS section
5726 NAME: client_persistent_connections
5728 LOC: Config.onoff.client_pconns
5732 NAME: server_persistent_connections
5734 LOC: Config.onoff.server_pconns
5737 Persistent connection support for clients and servers. By
5738 default, Squid uses persistent connections (when allowed)
5739 with its clients and servers. You can use these options to
5740 disable persistent connections with clients and/or servers.
5743 NAME: persistent_connection_after_error
5745 LOC: Config.onoff.error_pconns
5748 With this directive the use of persistent connections after
5749 HTTP errors can be disabled. Useful if you have clients
5750 who fail to handle errors on persistent connections proper.
5753 NAME: detect_broken_pconn
5755 LOC: Config.onoff.detect_broken_server_pconns
5758 Some servers have been found to incorrectly signal the use
5759 of HTTP/1.0 persistent connections even on replies not
5760 compatible, causing significant delays. This server problem
5761 has mostly been seen on redirects.
5763 By enabling this directive Squid attempts to detect such
5764 broken replies and automatically assume the reply is finished
5765 after 10 seconds timeout.
5769 CACHE DIGEST OPTIONS
5770 -----------------------------------------------------------------------------
5773 NAME: digest_generation
5774 IFDEF: USE_CACHE_DIGESTS
5776 LOC: Config.onoff.digest_generation
5779 This controls whether the server will generate a Cache Digest
5780 of its contents. By default, Cache Digest generation is
5781 enabled if Squid is compiled with --enable-cache-digests defined.
5784 NAME: digest_bits_per_entry
5785 IFDEF: USE_CACHE_DIGESTS
5787 LOC: Config.digest.bits_per_entry
5790 This is the number of bits of the server's Cache Digest which
5791 will be associated with the Digest entry for a given HTTP
5792 Method and URL (public key) combination. The default is 5.
5795 NAME: digest_rebuild_period
5796 IFDEF: USE_CACHE_DIGESTS
5799 LOC: Config.digest.rebuild_period
5802 This is the wait time between Cache Digest rebuilds.
5805 NAME: digest_rewrite_period
5807 IFDEF: USE_CACHE_DIGESTS
5809 LOC: Config.digest.rewrite_period
5812 This is the wait time between Cache Digest writes to
5816 NAME: digest_swapout_chunk_size
5819 IFDEF: USE_CACHE_DIGESTS
5820 LOC: Config.digest.swapout_chunk_size
5823 This is the number of bytes of the Cache Digest to write to
5824 disk at a time. It defaults to 4096 bytes (4KB), the Squid
5828 NAME: digest_rebuild_chunk_percentage
5829 COMMENT: (percent, 0-100)
5830 IFDEF: USE_CACHE_DIGESTS
5832 LOC: Config.digest.rebuild_chunk_percentage
5835 This is the percentage of the Cache Digest to be scanned at a
5836 time. By default it is set to 10% of the Cache Digest.
5841 -----------------------------------------------------------------------------
5846 LOC: Config.Port.snmp
5850 The port number where Squid listens for SNMP requests. To enable
5851 SNMP support set this to a suitable port number. Port number
5852 3401 is often used for the Squid SNMP agent. By default it's
5853 set to "0" (disabled)
5861 LOC: Config.accessList.snmp
5862 DEFAULT_IF_NONE: deny all
5865 Allowing or denying access to the SNMP port.
5867 All access to the agent is denied by default.
5870 snmp_access allow|deny [!]aclname ...
5872 This clause only supports fast acl types.
5873 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
5875 snmp_access allow snmppublic localhost
5876 snmp_access deny all
5879 NAME: snmp_incoming_address
5881 LOC: Config.Addrs.snmp_incoming
5886 NAME: snmp_outgoing_address
5888 LOC: Config.Addrs.snmp_outgoing
5892 Just like 'udp_incoming_address', but for the SNMP port.
5894 snmp_incoming_address is used for the SNMP socket receiving
5895 messages from SNMP agents.
5896 snmp_outgoing_address is used for SNMP packets returned to SNMP
5899 The default snmp_incoming_address is to listen on all
5900 available network interfaces.
5902 If snmp_outgoing_address is not set it will use the same socket
5903 as snmp_incoming_address. Only change this if you want to have
5904 SNMP replies sent using another address than where this Squid
5905 listens for SNMP queries.
5907 NOTE, snmp_incoming_address and snmp_outgoing_address can not have
5908 the same value since they both use port 3401.
5913 -----------------------------------------------------------------------------
5916 NAME: icp_port udp_port
5919 LOC: Config.Port.icp
5921 The port number where Squid sends and receives ICP queries to
5922 and from neighbor caches. The standard UDP port for ICP is 3130.
5923 Default is disabled (0).
5926 icp_port @DEFAULT_ICP_PORT@
5933 LOC: Config.Port.htcp
5935 The port number where Squid sends and receives HTCP queries to
5936 and from neighbor caches. To turn it on you want to set it to
5937 4827. By default it is set to "0" (disabled).
5943 NAME: log_icp_queries
5947 LOC: Config.onoff.log_udp
5949 If set, ICP queries are logged to access.log. You may wish
5950 do disable this if your ICP load is VERY high to speed things
5951 up or to simplify log analysis.
5954 NAME: udp_incoming_address
5956 LOC:Config.Addrs.udp_incoming
5959 udp_incoming_address is used for UDP packets received from other
5962 The default behavior is to not bind to any specific address.
5964 Only change this if you want to have all UDP queries received on
5965 a specific interface/address.
5967 NOTE: udp_incoming_address is used by the ICP, HTCP, and DNS
5968 modules. Altering it will affect all of them in the same manner.
5970 see also; udp_outgoing_address
5972 NOTE, udp_incoming_address and udp_outgoing_address can not
5973 have the same value since they both use the same port.
5976 NAME: udp_outgoing_address
5978 LOC: Config.Addrs.udp_outgoing
5981 udp_outgoing_address is used for UDP packets sent out to other
5984 The default behavior is to not bind to any specific address.
5986 Instead it will use the same socket as udp_incoming_address.
5987 Only change this if you want to have UDP queries sent using another
5988 address than where this Squid listens for UDP queries from other
5991 NOTE: udp_outgoing_address is used by the ICP, HTCP, and DNS
5992 modules. Altering it will affect all of them in the same manner.
5994 see also; udp_incoming_address
5996 NOTE, udp_incoming_address and udp_outgoing_address can not
5997 have the same value since they both use the same port.
6004 LOC: Config.onoff.icp_hit_stale
6006 If you want to return ICP_HIT for stale cache objects, set this
6007 option to 'on'. If you have sibling relationships with caches
6008 in other administrative domains, this should be 'off'. If you only
6009 have sibling relationships with caches under your control,
6010 it is probably okay to set this to 'on'.
6011 If set to 'on', your siblings should use the option "allow-miss"
6012 on their cache_peer lines for connecting to you.
6015 NAME: minimum_direct_hops
6018 LOC: Config.minDirectHops
6020 If using the ICMP pinging stuff, do direct fetches for sites
6021 which are no more than this many hops away.
6024 NAME: minimum_direct_rtt
6027 LOC: Config.minDirectRtt
6029 If using the ICMP pinging stuff, do direct fetches for sites
6030 which are no more than this many rtt milliseconds away.
6036 LOC: Config.Netdb.low
6042 LOC: Config.Netdb.high
6044 The low and high water marks for the ICMP measurement
6045 database. These are counts, not percents. The defaults are
6046 900 and 1000. When the high water mark is reached, database
6047 entries will be deleted until the low mark is reached.
6050 NAME: netdb_ping_period
6052 LOC: Config.Netdb.period
6055 The minimum period for measuring a site. There will be at
6056 least this much delay between successive pings to the same
6057 network. The default is five minutes.
6064 LOC: Config.onoff.query_icmp
6066 If you want to ask your peers to include ICMP data in their ICP
6067 replies, enable this option.
6069 If your peer has configured Squid (during compilation) with
6070 '--enable-icmp' that peer will send ICMP pings to origin server
6071 sites of the URLs it receives. If you enable this option the
6072 ICP replies from that peer will include the ICMP data (if available).
6073 Then, when choosing a parent cache, Squid will choose the parent with
6074 the minimal RTT to the origin server. When this happens, the
6075 hierarchy field of the access.log will be
6076 "CLOSEST_PARENT_MISS". This option is off by default.
6079 NAME: test_reachability
6083 LOC: Config.onoff.test_reachability
6085 When this is 'on', ICP MISS replies will be ICP_MISS_NOFETCH
6086 instead of ICP_MISS if the target host is NOT in the ICMP
6087 database, or has a zero RTT.
6090 NAME: icp_query_timeout
6094 LOC: Config.Timeout.icp_query
6096 Normally Squid will automatically determine an optimal ICP
6097 query timeout value based on the round-trip-time of recent ICP
6098 queries. If you want to override the value determined by
6099 Squid, set this 'icp_query_timeout' to a non-zero value. This
6100 value is specified in MILLISECONDS, so, to use a 2-second
6101 timeout (the old default), you would write:
6103 icp_query_timeout 2000
6106 NAME: maximum_icp_query_timeout
6110 LOC: Config.Timeout.icp_query_max
6112 Normally the ICP query timeout is determined dynamically. But
6113 sometimes it can lead to very large values (say 5 seconds).
6114 Use this option to put an upper limit on the dynamic timeout
6115 value. Do NOT use this option to always use a fixed (instead
6116 of a dynamic) timeout value. To set a fixed timeout see the
6117 'icp_query_timeout' directive.
6120 NAME: minimum_icp_query_timeout
6124 LOC: Config.Timeout.icp_query_min
6126 Normally the ICP query timeout is determined dynamically. But
6127 sometimes it can lead to very small timeouts, even lower than
6128 the normal latency variance on your link due to traffic.
6129 Use this option to put an lower limit on the dynamic timeout
6130 value. Do NOT use this option to always use a fixed (instead
6131 of a dynamic) timeout value. To set a fixed timeout see the
6132 'icp_query_timeout' directive.
6135 NAME: background_ping_rate
6139 LOC: Config.backgroundPingRate
6141 Controls how often the ICP pings are sent to siblings that
6142 have background-ping set.
6146 MULTICAST ICP OPTIONS
6147 -----------------------------------------------------------------------------
6152 LOC: Config.mcast_group_list
6155 This tag specifies a list of multicast groups which your server
6156 should join to receive multicasted ICP queries.
6158 NOTE! Be very careful what you put here! Be sure you
6159 understand the difference between an ICP _query_ and an ICP
6160 _reply_. This option is to be set only if you want to RECEIVE
6161 multicast queries. Do NOT set this option to SEND multicast
6162 ICP (use cache_peer for that). ICP replies are always sent via
6163 unicast, so this option does not affect whether or not you will
6164 receive replies from multicast group members.
6166 You must be very careful to NOT use a multicast address which
6167 is already in use by another group of caches.
6169 If you are unsure about multicast, please read the Multicast
6170 chapter in the Squid FAQ (http://www.squid-cache.org/FAQ/).
6172 Usage: mcast_groups 239.128.16.128 224.0.1.20
6174 By default, Squid doesn't listen on any multicast groups.
6177 NAME: mcast_miss_addr
6178 IFDEF: MULTICAST_MISS_STREAM
6180 LOC: Config.mcast_miss.addr
6183 If you enable this option, every "cache miss" URL will
6184 be sent out on the specified multicast address.
6186 Do not enable this option unless you are are absolutely
6187 certain you understand what you are doing.
6190 NAME: mcast_miss_ttl
6191 IFDEF: MULTICAST_MISS_STREAM
6193 LOC: Config.mcast_miss.ttl
6196 This is the time-to-live value for packets multicasted
6197 when multicasting off cache miss URLs is enabled. By
6198 default this is set to 'site scope', i.e. 16.
6201 NAME: mcast_miss_port
6202 IFDEF: MULTICAST_MISS_STREAM
6204 LOC: Config.mcast_miss.port
6207 This is the port number to be used in conjunction with
6211 NAME: mcast_miss_encode_key
6212 IFDEF: MULTICAST_MISS_STREAM
6214 LOC: Config.mcast_miss.encode_key
6215 DEFAULT: XXXXXXXXXXXXXXXX
6217 The URLs that are sent in the multicast miss stream are
6218 encrypted. This is the encryption key.
6221 NAME: mcast_icp_query_timeout
6225 LOC: Config.Timeout.mcast_icp_query
6227 For multicast peers, Squid regularly sends out ICP "probes" to
6228 count how many other peers are listening on the given multicast
6229 address. This value specifies how long Squid should wait to
6230 count all the replies. The default is 2000 msec, or 2
6235 INTERNAL ICON OPTIONS
6236 -----------------------------------------------------------------------------
6239 NAME: icon_directory
6241 LOC: Config.icons.directory
6242 DEFAULT: @DEFAULT_ICON_DIR@
6244 Where the icons are stored. These are normally kept in
6248 NAME: global_internal_static
6250 LOC: Config.onoff.global_internal_static
6253 This directive controls is Squid should intercept all requests for
6254 /squid-internal-static/ no matter which host the URL is requesting
6255 (default on setting), or if nothing special should be done for
6256 such URLs (off setting). The purpose of this directive is to make
6257 icons etc work better in complex cache hierarchies where it may
6258 not always be possible for all corners in the cache mesh to reach
6259 the server generating a directory listing.
6262 NAME: short_icon_urls
6264 LOC: Config.icons.use_short_names
6267 If this is enabled Squid will use short URLs for icons.
6268 If disabled it will revert to the old behavior of including
6269 it's own name and port in the URL.
6271 If you run a complex cache hierarchy with a mix of Squid and
6272 other proxies you may need to disable this directive.
6277 -----------------------------------------------------------------------------
6280 NAME: error_directory
6282 LOC: Config.errorDirectory
6285 If you wish to create your own versions of the default
6286 error files to customize them to suit your company copy
6287 the error/template files to another directory and point
6290 WARNING: This option will disable multi-language support
6291 on error pages if used.
6293 The squid developers are interested in making squid available in
6294 a wide variety of languages. If you are making translations for a
6295 language that Squid does not currently provide please consider
6296 contributing your translation back to the project.
6297 http://wiki.squid-cache.org/Translations
6299 The squid developers working on translations are happy to supply drop-in
6300 translated error files in exchange for any new language contributions.
6303 NAME: error_default_language
6304 IFDEF: USE_ERR_LOCALES
6306 LOC: Config.errorDefaultLanguage
6309 Set the default language which squid will send error pages in
6310 if no existing translation matches the clients language
6313 If unset (default) generic English will be used.
6315 The squid developers are interested in making squid available in
6316 a wide variety of languages. If you are interested in making
6317 translations for any language see the squid wiki for details.
6318 http://wiki.squid-cache.org/Translations
6321 NAME: error_log_languages
6322 IFDEF: USE_ERR_LOCALES
6324 LOC: Config.errorLogMissingLanguages
6327 Log to cache.log what languages users are attempting to
6328 auto-negotiate for translations.
6330 Successful negotiations are not logged. Only failures
6331 have meaning to indicate that Squid may need an upgrade
6332 of its error page translations.
6335 NAME: err_page_stylesheet
6337 LOC: Config.errorStylesheet
6338 DEFAULT: @DEFAULT_CONFIG_DIR@/errorpage.css
6340 CSS Stylesheet to pattern the display of Squid default error pages.
6342 For information on CSS see http://www.w3.org/Style/CSS/
6347 LOC: Config.errHtmlText
6350 HTML text to include in error messages. Make this a "mailto"
6351 URL to your admin address, or maybe just a link to your
6352 organizations Web page.
6354 To include this in your error messages, you must rewrite
6355 the error template files (found in the "errors" directory).
6356 Wherever you want the 'err_html_text' line to appear,
6357 insert a %L tag in the error template file.
6360 NAME: email_err_data
6363 LOC: Config.onoff.emailErrData
6366 If enabled, information about the occurred error will be
6367 included in the mailto links of the ERR pages (if %W is set)
6368 so that the email body contains the data.
6369 Syntax is <A HREF="mailto:%w%W">%w</A>
6374 LOC: Config.denyInfoList
6377 Usage: deny_info err_page_name acl
6378 or deny_info http://... acl
6379 or deny_info TCP_RESET acl
6381 This can be used to return a ERR_ page for requests which
6382 do not pass the 'http_access' rules. Squid remembers the last
6383 acl it evaluated in http_access, and if a 'deny_info' line exists
6384 for that ACL Squid returns a corresponding error page.
6386 The acl is typically the last acl on the http_access deny line which
6387 denied access. The exceptions to this rule are:
6388 - When Squid needs to request authentication credentials. It's then
6389 the first authentication related acl encountered
6390 - When none of the http_access lines matches. It's then the last
6391 acl processed on the last http_access line.
6392 - When the decision to deny access was made by an adaptation service,
6393 the acl name is the corresponding eCAP or ICAP service_name.
6395 NP: If providing your own custom error pages with error_directory
6396 you may also specify them by your custom file name:
6397 Example: deny_info ERR_CUSTOM_ACCESS_DENIED bad_guys
6399 By defaut Squid will send "403 Forbidden". A different 4xx or 5xx
6400 may be specified by prefixing the file name with the code and a colon.
6401 e.g. 404:ERR_CUSTOM_ACCESS_DENIED
6403 Alternatively you can tell Squid to reset the TCP connection
6404 by specifying TCP_RESET.
6406 Or you can specify an error URL or URL pattern. The browsers will
6407 get redirected to the specified URL after formatting tags have
6408 been replaced. Redirect will be done with 302 or 307 according to
6409 HTTP/1.1 specs. A different 3xx code may be specified by prefixing
6410 the URL. e.g. 303:http://example.com/
6413 %a - username (if available. Password NOT included)
6416 %E - Error description
6418 %H - Request domain name
6419 %i - Client IP Address
6421 %o - Message result from external ACL helper
6422 %p - Request Port number
6423 %P - Request Protocol name
6424 %R - Request URL path
6425 %T - Timestamp in RFC 1123 format
6426 %U - Full canonical URL from client
6427 (HTTPS URLs terminate with *)
6428 %u - Full canonical URL from client
6429 %w - Admin email from squid.conf
6431 %% - Literal percent (%) code
6436 OPTIONS INFLUENCING REQUEST FORWARDING
6437 -----------------------------------------------------------------------------
6440 NAME: nonhierarchical_direct
6442 LOC: Config.onoff.nonhierarchical_direct
6445 By default, Squid will send any non-hierarchical requests
6446 (matching hierarchy_stoplist or not cacheable request type) direct
6449 If you set this to off, Squid will prefer to send these
6450 requests to parents.
6452 Note that in most configurations, by turning this off you will only
6453 add latency to these request without any improvement in global hit
6456 If you are inside an firewall see never_direct instead of
6462 LOC: Config.onoff.prefer_direct
6465 Normally Squid tries to use parents for most requests. If you for some
6466 reason like it to first try going direct and only use a parent if
6467 going direct fails set this to on.
6469 By combining nonhierarchical_direct off and prefer_direct on you
6470 can set up Squid to use a parent as a backup path if going direct
6473 Note: If you want Squid to use parents for all requests see
6474 the never_direct directive. prefer_direct only modifies how Squid
6475 acts on cacheable requests.
6480 LOC: Config.accessList.AlwaysDirect
6483 Usage: always_direct allow|deny [!]aclname ...
6485 Here you can use ACL elements to specify requests which should
6486 ALWAYS be forwarded by Squid to the origin servers without using
6487 any peers. For example, to always directly forward requests for
6488 local servers ignoring any parents or siblings you may have use
6491 acl local-servers dstdomain my.domain.net
6492 always_direct allow local-servers
6494 To always forward FTP requests directly, use
6497 always_direct allow FTP
6499 NOTE: There is a similar, but opposite option named
6500 'never_direct'. You need to be aware that "always_direct deny
6501 foo" is NOT the same thing as "never_direct allow foo". You
6502 may need to use a deny rule to exclude a more-specific case of
6503 some other rule. Example:
6505 acl local-external dstdomain external.foo.net
6506 acl local-servers dstdomain .foo.net
6507 always_direct deny local-external
6508 always_direct allow local-servers
6510 NOTE: If your goal is to make the client forward the request
6511 directly to the origin server bypassing Squid then this needs
6512 to be done in the client configuration. Squid configuration
6513 can only tell Squid how Squid should fetch the object.
6515 NOTE: This directive is not related to caching. The replies
6516 is cached as usual even if you use always_direct. To not cache
6517 the replies see the 'cache' directive.
6519 This clause supports both fast and slow acl types.
6520 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6525 LOC: Config.accessList.NeverDirect
6528 Usage: never_direct allow|deny [!]aclname ...
6530 never_direct is the opposite of always_direct. Please read
6531 the description for always_direct if you have not already.
6533 With 'never_direct' you can use ACL elements to specify
6534 requests which should NEVER be forwarded directly to origin
6535 servers. For example, to force the use of a proxy for all
6536 requests, except those in your local domain use something like:
6538 acl local-servers dstdomain .foo.net
6539 never_direct deny local-servers
6540 never_direct allow all
6542 or if Squid is inside a firewall and there are local intranet
6543 servers inside the firewall use something like:
6545 acl local-intranet dstdomain .foo.net
6546 acl local-external dstdomain external.foo.net
6547 always_direct deny local-external
6548 always_direct allow local-intranet
6549 never_direct allow all
6551 This clause supports both fast and slow acl types.
6552 See http://wiki.squid-cache.org/SquidFaq/SquidAcl for details.
6556 ADVANCED NETWORKING OPTIONS
6557 -----------------------------------------------------------------------------
6560 NAME: incoming_icp_average
6563 LOC: Config.comm_incoming.icp_average
6566 NAME: incoming_http_average
6569 LOC: Config.comm_incoming.http_average
6572 NAME: incoming_dns_average
6575 LOC: Config.comm_incoming.dns_average
6578 NAME: min_icp_poll_cnt
6581 LOC: Config.comm_incoming.icp_min_poll
6584 NAME: min_dns_poll_cnt
6587 LOC: Config.comm_incoming.dns_min_poll
6590 NAME: min_http_poll_cnt
6593 LOC: Config.comm_incoming.http_min_poll
6595 Heavy voodoo here. I can't even believe you are reading this.
6596 Are you crazy? Don't even think about adjusting these unless
6597 you understand the algorithms in comm_select.c first!
6603 LOC: Config.accept_filter
6607 The name of an accept(2) filter to install on Squid's
6608 listen socket(s). This feature is perhaps specific to
6609 FreeBSD and requires support in the kernel.
6611 The 'httpready' filter delays delivering new connections
6612 to Squid until a full HTTP request has been received.
6613 See the accf_http(9) man page for details.
6615 The 'dataready' filter delays delivering new connections
6616 to Squid until there is some data to process.
6617 See the accf_dataready(9) man page for details.
6621 The 'data' filter delays delivering of new connections
6622 to Squid until there is some data to process by TCP_ACCEPT_DEFER.
6623 You may optionally specify a number of seconds to wait by
6624 'data=N' where N is the number of seconds. Defaults to 30
6625 if not specified. See the tcp(7) man page for details.
6628 accept_filter httpready
6633 NAME: client_ip_max_connections
6635 LOC: Config.client_ip_max_connections
6638 Set an absolute limit on the number of connections a single
6639 client IP can use. Any more than this and Squid will begin to drop
6640 new connections from the client until it closes some links.
6642 Note that this is a global limit. It affects all HTTP, HTCP, Gopher and FTP
6643 connections from the client. For finer control use the ACL access controls.
6645 Requires client_db to be enabled (the default).
6647 WARNING: This may noticably slow down traffic received via external proxies
6648 or NAT devices and cause them to rebound error messages back to their clients.
6651 NAME: tcp_recv_bufsize
6655 LOC: Config.tcpRcvBufsz
6657 Size of receive buffer to set for TCP sockets. Probably just
6658 as easy to change your kernel's default. Set to zero to use
6659 the default buffer size.
6664 -----------------------------------------------------------------------------
6671 LOC: Adaptation::Icap::TheConfig.onoff
6674 If you want to enable the ICAP module support, set this to on.
6677 NAME: icap_connect_timeout
6680 LOC: Adaptation::Icap::TheConfig.connect_timeout_raw
6683 This parameter specifies how long to wait for the TCP connect to
6684 the requested ICAP server to complete before giving up and either
6685 terminating the HTTP transaction or bypassing the failure.
6687 The default for optional services is peer_connect_timeout.
6688 The default for essential services is connect_timeout.
6689 If this option is explicitly set, its value applies to all services.
6692 NAME: icap_io_timeout
6696 LOC: Adaptation::Icap::TheConfig.io_timeout_raw
6699 This parameter specifies how long to wait for an I/O activity on
6700 an established, active ICAP connection before giving up and
6701 either terminating the HTTP transaction or bypassing the
6704 The default is read_timeout.
6707 NAME: icap_service_failure_limit
6708 COMMENT: limit [in memory-depth time-units]
6709 TYPE: icap_service_failure_limit
6711 LOC: Adaptation::Icap::TheConfig
6714 The limit specifies the number of failures that Squid tolerates
6715 when establishing a new TCP connection with an ICAP service. If
6716 the number of failures exceeds the limit, the ICAP service is
6717 not used for new ICAP requests until it is time to refresh its
6720 A negative value disables the limit. Without the limit, an ICAP
6721 service will not be considered down due to connectivity failures
6722 between ICAP OPTIONS requests.
6724 Squid forgets ICAP service failures older than the specified
6725 value of memory-depth. The memory fading algorithm
6726 is approximate because Squid does not remember individual
6727 errors but groups them instead, splitting the option
6728 value into ten time slots of equal length.
6730 When memory-depth is 0 and by default this option has no
6731 effect on service failure expiration.
6733 Squid always forgets failures when updating service settings
6734 using an ICAP OPTIONS transaction, regardless of this option
6738 # suspend service usage after 10 failures in 5 seconds:
6739 icap_service_failure_limit 10 in 5 seconds
6742 NAME: icap_service_revival_delay
6745 LOC: Adaptation::Icap::TheConfig.service_revival_delay
6748 The delay specifies the number of seconds to wait after an ICAP
6749 OPTIONS request failure before requesting the options again. The
6750 failed ICAP service is considered "down" until fresh OPTIONS are
6753 The actual delay cannot be smaller than the hardcoded minimum
6754 delay of 30 seconds.
6757 NAME: icap_preview_enable
6761 LOC: Adaptation::Icap::TheConfig.preview_enable
6764 The ICAP Preview feature allows the ICAP server to handle the
6765 HTTP message by looking only at the beginning of the message body
6766 or even without receiving the body at all. In some environments,
6767 previews greatly speedup ICAP processing.
6769 During an ICAP OPTIONS transaction, the server may tell Squid what
6770 HTTP messages should be previewed and how big the preview should be.
6771 Squid will not use Preview if the server did not request one.
6773 To disable ICAP Preview for all ICAP services, regardless of
6774 individual ICAP server OPTIONS responses, set this option to "off".
6776 icap_preview_enable off
6779 NAME: icap_preview_size
6782 LOC: Adaptation::Icap::TheConfig.preview_size
6785 The default size of preview data to be sent to the ICAP server.
6786 -1 means no preview. This value might be overwritten on a per server
6787 basis by OPTIONS requests.
6790 NAME: icap_206_enable
6794 LOC: Adaptation::Icap::TheConfig.allow206_enable
6797 206 (Partial Content) responses is an ICAP extension that allows the
6798 ICAP agents to optionally combine adapted and original HTTP message
6799 content. The decision to combine is postponed until the end of the
6800 ICAP response. Squid supports Partial Content extension by default.
6802 Activation of the Partial Content extension is negotiated with each
6803 ICAP service during OPTIONS exchange. Most ICAP servers should handle
6804 negotation correctly even if they do not support the extension, but
6805 some might fail. To disable Partial Content support for all ICAP
6806 services and to avoid any negotiation, set this option to "off".
6812 NAME: icap_default_options_ttl
6815 LOC: Adaptation::Icap::TheConfig.default_options_ttl
6818 The default TTL value for ICAP OPTIONS responses that don't have
6819 an Options-TTL header.
6822 NAME: icap_persistent_connections
6826 LOC: Adaptation::Icap::TheConfig.reuse_connections
6829 Whether or not Squid should use persistent connections to
6833 NAME: adaptation_send_client_ip icap_send_client_ip
6835 IFDEF: USE_ADAPTATION
6837 LOC: Adaptation::Config::send_client_ip
6840 If enabled, Squid shares HTTP client IP information with adaptation
6841 services. For ICAP, Squid adds the X-Client-IP header to ICAP requests.
6842 For eCAP, Squid sets the libecap::metaClientIp transaction option.
6844 See also: adaptation_uses_indirect_client
6847 NAME: adaptation_send_username icap_send_client_username
6849 IFDEF: USE_ADAPTATION
6851 LOC: Adaptation::Config::send_username
6854 This sends authenticated HTTP client username (if available) to
6855 the adaptation service.
6857 For ICAP, the username value is encoded based on the
6858 icap_client_username_encode option and is sent using the header
6859 specified by the icap_client_username_header option.
6862 NAME: icap_client_username_header
6865 LOC: Adaptation::Icap::TheConfig.client_username_header
6866 DEFAULT: X-Client-Username
6868 ICAP request header name to use for adaptation_send_username.
6871 NAME: icap_client_username_encode
6875 LOC: Adaptation::Icap::TheConfig.client_username_encode
6878 Whether to base64 encode the authenticated client username.
6882 TYPE: icap_service_type
6884 LOC: Adaptation::Icap::TheConfig
6887 Defines a single ICAP service using the following format:
6889 icap_service id vectoring_point uri [option ...]
6892 an opaque identifier or name which is used to direct traffic to
6893 this specific service. Must be unique among all adaptation
6894 services in squid.conf.
6896 vectoring_point: reqmod_precache|reqmod_postcache|respmod_precache|respmod_postcache
6897 This specifies at which point of transaction processing the
6898 ICAP service should be activated. *_postcache vectoring points
6899 are not yet supported.
6901 uri: icap://servername:port/servicepath
6902 ICAP server and service location.
6904 ICAP does not allow a single service to handle both REQMOD and RESPMOD
6905 transactions. Squid does not enforce that requirement. You can specify
6906 services with the same service_url and different vectoring_points. You
6907 can even specify multiple identical services as long as their
6908 service_names differ.
6911 Service options are separated by white space. ICAP services support
6912 the following name=value options:
6915 If set to 'on' or '1', the ICAP service is treated as
6916 optional. If the service cannot be reached or malfunctions,
6917 Squid will try to ignore any errors and process the message as
6918 if the service was not enabled. No all ICAP errors can be
6919 bypassed. If set to 0, the ICAP service is treated as
6920 essential and all ICAP errors will result in an error page
6921 returned to the HTTP client.
6923 Bypass is off by default: services are treated as essential.
6926 If set to 'on' or '1', the ICAP service is allowed to
6927 dynamically change the current message adaptation plan by
6928 returning a chain of services to be used next. The services
6929 are specified using the X-Next-Services ICAP response header
6930 value, formatted as a comma-separated list of service names.
6931 Each named service should be configured in squid.conf. Other
6932 services are ignored. An empty X-Next-Services value results
6933 in an empty plan which ends the current adaptation.
6935 Dynamic adaptation plan may cross or cover multiple supported
6936 vectoring points in their natural processing order.
6938 Routing is not allowed by default: the ICAP X-Next-Services
6939 response header is ignored.
6942 Only has effect on split-stack systems. The default on those systems
6943 is to use IPv4-only connections. When set to 'on' this option will
6944 make Squid use IPv6-only connections to contact this ICAP service.
6946 on-overload=block|bypass|wait|force
6947 If the service Max-Connections limit has been reached, do
6948 one of the following for each new ICAP transaction:
6949 * block: send an HTTP error response to the client
6950 * bypass: ignore the "over-connected" ICAP service
6951 * wait: wait (in a FIFO queue) for an ICAP connection slot
6952 * force: proceed, ignoring the Max-Connections limit
6954 In SMP mode with N workers, each worker assumes the service
6955 connection limit is Max-Connections/N, even though not all
6956 workers may use a given service.
6958 The default value is "bypass" if service is bypassable,
6959 otherwise it is set to "wait".
6963 Use the given number as the Max-Connections limit, regardless
6964 of the Max-Connections value given by the service, if any.
6966 Older icap_service format without optional named parameters is
6967 deprecated but supported for backward compatibility.
6970 icap_service svcBlocker reqmod_precache icap://icap1.mydomain.net:1344/reqmod bypass=0
6971 icap_service svcLogger reqmod_precache icap://icap2.mydomain.net:1344/respmod routing=on
6975 TYPE: icap_class_type
6980 This deprecated option was documented to define an ICAP service
6981 chain, even though it actually defined a set of similar, redundant
6982 services, and the chains were not supported.
6984 To define a set of redundant services, please use the
6985 adaptation_service_set directive. For service chains, use
6986 adaptation_service_chain.
6990 TYPE: icap_access_type
6995 This option is deprecated. Please use adaptation_access, which
6996 has the same ICAP functionality, but comes with better
6997 documentation, and eCAP support.
7002 -----------------------------------------------------------------------------
7009 LOC: Adaptation::Ecap::TheConfig.onoff
7012 Controls whether eCAP support is enabled.
7016 TYPE: ecap_service_type
7018 LOC: Adaptation::Ecap::TheConfig
7021 Defines a single eCAP service
7023 ecap_service id vectoring_point uri [option ...]
7026 an opaque identifier or name which is used to direct traffic to
7027 this specific service. Must be unique among all adaptation
7028 services in squid.conf.
7030 vectoring_point: reqmod_precache|reqmod_postcache|respmod_precache|respmod_postcache
7031 This specifies at which point of transaction processing the
7032 eCAP service should be activated. *_postcache vectoring points
7033 are not yet supported.
7035 uri: ecap://vendor/service_name?custom&cgi=style¶meters=optional
7036 Squid uses the eCAP service URI to match this configuration
7037 line with one of the dynamically loaded services. Each loaded
7038 eCAP service must have a unique URI. Obtain the right URI from
7039 the service provider.
7042 Service options are separated by white space. eCAP services support
7043 the following name=value options:
7046 If set to 'on' or '1', the eCAP service is treated as optional.
7047 If the service cannot be reached or malfunctions, Squid will try
7048 to ignore any errors and process the message as if the service
7049 was not enabled. No all eCAP errors can be bypassed.
7050 If set to 'off' or '0', the eCAP service is treated as essential
7051 and all eCAP errors will result in an error page returned to the
7054 Bypass is off by default: services are treated as essential.
7057 If set to 'on' or '1', the eCAP service is allowed to
7058 dynamically change the current message adaptation plan by
7059 returning a chain of services to be used next.
7061 Dynamic adaptation plan may cross or cover multiple supported
7062 vectoring points in their natural processing order.
7064 Routing is not allowed by default.
7066 Older ecap_service format without optional named parameters is
7067 deprecated but supported for backward compatibility.
7071 ecap_service s1 reqmod_precache ecap://filters.R.us/leakDetector?on_error=block bypass=off
7072 ecap_service s2 respmod_precache ecap://filters.R.us/virusFilter config=/etc/vf.cfg bypass=on
7075 NAME: loadable_modules
7077 IFDEF: USE_LOADABLE_MODULES
7078 LOC: Config.loadable_module_names
7081 Instructs Squid to load the specified dynamic module(s) or activate
7082 preloaded module(s).
7084 loadable_modules @DEFAULT_PREFIX@/lib/MinimalAdapter.so
7088 MESSAGE ADAPTATION OPTIONS
7089 -----------------------------------------------------------------------------
7092 NAME: adaptation_service_set
7093 TYPE: adaptation_service_set_type
7094 IFDEF: USE_ADAPTATION
7099 Configures an ordered set of similar, redundant services. This is
7100 useful when hot standby or backup adaptation servers are available.
7102 adaptation_service_set set_name service_name1 service_name2 ...
7104 The named services are used in the set declaration order. The first
7105 applicable adaptation service from the set is used first. The next
7106 applicable service is tried if and only if the transaction with the
7107 previous service fails and the message waiting to be adapted is still
7110 When adaptation starts, broken services are ignored as if they were
7111 not a part of the set. A broken service is a down optional service.
7113 The services in a set must be attached to the same vectoring point
7114 (e.g., pre-cache) and use the same adaptation method (e.g., REQMOD).
7116 If all services in a set are optional then adaptation failures are
7117 bypassable. If all services in the set are essential, then a
7118 transaction failure with one service may still be retried using
7119 another service from the set, but when all services fail, the master
7120 transaction fails as well.
7122 A set may contain a mix of optional and essential services, but that
7123 is likely to lead to surprising results because broken services become
7124 ignored (see above), making previously bypassable failures fatal.
7125 Technically, it is the bypassability of the last failed service that
7128 See also: adaptation_access adaptation_service_chain
7131 adaptation_service_set svcBlocker urlFilterPrimary urlFilterBackup
7132 adaptation service_set svcLogger loggerLocal loggerRemote
7135 NAME: adaptation_service_chain
7136 TYPE: adaptation_service_chain_type
7137 IFDEF: USE_ADAPTATION
7142 Configures a list of complementary services that will be applied
7143 one-by-one, forming an adaptation chain or pipeline. This is useful
7144 when Squid must perform different adaptations on the same message.
7146 adaptation_service_chain chain_name service_name1 svc_name2 ...
7148 The named services are used in the chain declaration order. The first
7149 applicable adaptation service from the chain is used first. The next
7150 applicable service is applied to the successful adaptation results of
7151 the previous service in the chain.
7153 When adaptation starts, broken services are ignored as if they were
7154 not a part of the chain. A broken service is a down optional service.
7156 Request satisfaction terminates the adaptation chain because Squid
7157 does not currently allow declaration of RESPMOD services at the
7158 "reqmod_precache" vectoring point (see icap_service or ecap_service).
7160 The services in a chain must be attached to the same vectoring point
7161 (e.g., pre-cache) and use the same adaptation method (e.g., REQMOD).
7163 A chain may contain a mix of optional and essential services. If an
7164 essential adaptation fails (or the failure cannot be bypassed for
7165 other reasons), the master transaction fails. Otherwise, the failure
7166 is bypassed as if the failed adaptation service was not in the chain.
7168 See also: adaptation_access adaptation_service_set
7171 adaptation_service_chain svcRequest requestLogger urlFilter leakDetector
7174 NAME: adaptation_access
7175 TYPE: adaptation_access_type
7176 IFDEF: USE_ADAPTATION
7180 Sends an HTTP transaction to an ICAP or eCAP adaptation service.
7182 adaptation_access service_name allow|deny [!]aclname...
7183 adaptation_access set_name allow|deny [!]aclname...
7185 At each supported vectoring point, the adaptation_access
7186 statements are processed in the order they appear in this
7187 configuration file. Statements pointing to the following services
7188 are ignored (i.e., skipped without checking their ACL):
7190 - services serving different vectoring points
7191 - "broken-but-bypassable" services
7192 - "up" services configured to ignore such transactions
7193 (e.g., based on the ICAP Transfer-Ignore header).
7195 When a set_name is used, all services in the set are checked
7196 using the same rules, to find the first applicable one. See
7197 adaptation_service_set for details.
7199 If an access list is checked and there is a match, the
7200 processing stops: For an "allow" rule, the corresponding
7201 adaptation service is used for the transaction. For a "deny"
7202 rule, no adaptation service is activated.
7204 It is currently not possible to apply more than one adaptation
7205 service at the same vectoring point to the same HTTP transaction.
7207 See also: icap_service and ecap_service
7210 adaptation_access service_1 allow all
7213 NAME: adaptation_service_iteration_limit
7215 IFDEF: USE_ADAPTATION
7216 LOC: Adaptation::Config::service_iteration_limit
7219 Limits the number of iterations allowed when applying adaptation
7220 services to a message. If your longest adaptation set or chain
7221 may have more than 16 services, increase the limit beyond its
7222 default value of 16. If detecting infinite iteration loops sooner
7223 is critical, make the iteration limit match the actual number
7224 of services in your longest adaptation set or chain.
7226 Infinite adaptation loops are most likely with routing services.
7228 See also: icap_service routing=1
7231 NAME: adaptation_masterx_shared_names
7233 IFDEF: USE_ADAPTATION
7234 LOC: Adaptation::Config::masterx_shared_name
7237 For each master transaction (i.e., the HTTP request and response
7238 sequence, including all related ICAP and eCAP exchanges), Squid
7239 maintains a table of metadata. The table entries are (name, value)
7240 pairs shared among eCAP and ICAP exchanges. The table is destroyed
7241 with the master transaction.
7243 This option specifies the table entry names that Squid must accept
7244 from and forward to the adaptation transactions.
7246 An ICAP REQMOD or RESPMOD transaction may set an entry in the
7247 shared table by returning an ICAP header field with a name
7248 specified in adaptation_masterx_shared_names.
7250 An eCAP REQMOD or RESPMOD transaction may set an entry in the
7251 shared table by implementing the libecap::visitEachOption() API
7252 to provide an option with a name specified in
7253 adaptation_masterx_shared_names.
7255 Squid will store and forward the set entry to subsequent adaptation
7256 transactions within the same master transaction scope.
7258 Only one shared entry name is supported at this time.
7261 # share authentication information among ICAP services
7262 adaptation_masterx_shared_names X-Subscriber-ID
7265 NAME: adaptation_meta
7266 TYPE: adaptation_meta_type
7267 IFDEF: USE_ADAPTATION
7268 LOC: Adaptation::Config::metaHeaders
7271 This option allows Squid administrator to add custom ICAP request
7272 headers or eCAP options to Squid ICAP requests or eCAP transactions.
7273 Use it to pass custom authentication tokens and other
7274 transaction-state related meta information to an ICAP/eCAP service.
7276 The addition of a meta header is ACL-driven:
7277 adaptation_meta name value [!]aclname ...
7279 Processing for a given header name stops after the first ACL list match.
7280 Thus, it is impossible to add two headers with the same name. If no ACL
7281 lists match for a given header name, no such header is added. For
7284 # do not debug transactions except for those that need debugging
7285 adaptation_meta X-Debug 1 needs_debugging
7287 # log all transactions except for those that must remain secret
7288 adaptation_meta X-Log 1 !keep_secret
7290 # mark transactions from users in the "G 1" group
7291 adaptation_meta X-Authenticated-Groups "G 1" authed_as_G1
7293 The "value" parameter may be a regular squid.conf token or a "double
7294 quoted string". Within the quoted string, use backslash (\) to escape
7295 any character, which is currently only useful for escaping backslashes
7296 and double quotes. For example,
7297 "this string has one backslash (\\) and two \"quotes\""
7303 LOC: Adaptation::Icap::TheConfig.repeat
7304 DEFAULT_IF_NONE: deny all
7306 This ACL determines which retriable ICAP transactions are
7307 retried. Transactions that received a complete ICAP response
7308 and did not have to consume or produce HTTP bodies to receive
7309 that response are usually retriable.
7311 icap_retry allow|deny [!]aclname ...
7313 Squid automatically retries some ICAP I/O timeouts and errors
7314 due to persistent connection race conditions.
7316 See also: icap_retry_limit
7319 NAME: icap_retry_limit
7322 LOC: Adaptation::Icap::TheConfig.repeat_limit
7325 Limits the number of retries allowed. When set to zero (default),
7326 no retries are allowed.
7328 Communication errors due to persistent connection race
7329 conditions are unavoidable, automatically retried, and do not
7330 count against this limit.
7332 See also: icap_retry
7338 -----------------------------------------------------------------------------
7341 NAME: check_hostnames
7344 LOC: Config.onoff.check_hostnames
7346 For security and stability reasons Squid can check
7347 hostnames for Internet standard RFC compliance. If you want
7348 Squid to perform these checks turn this directive on.
7351 NAME: allow_underscore
7354 LOC: Config.onoff.allow_underscore
7356 Underscore characters is not strictly allowed in Internet hostnames
7357 but nevertheless used by many sites. Set this to off if you want
7358 Squid to be strict about the standard.
7359 This check is performed only when check_hostnames is set to on.
7362 NAME: cache_dns_program
7364 IFDEF: USE_DNSHELPER
7365 DEFAULT: @DEFAULT_DNSSERVER@
7366 LOC: Config.Program.dnsserver
7368 Specify the location of the executable for dnslookup process.
7372 TYPE: HelperChildConfig
7373 IFDEF: USE_DNSHELPER
7374 DEFAULT: 32 startup=1 idle=1
7375 LOC: Config.dnsChildren
7377 The maximum number of processes spawn to service DNS name lookups.
7378 If you limit it too few Squid will have to wait for them to process
7379 a backlog of requests, slowing it down. If you allow too many they
7380 will use RAM and other system resources noticably.
7381 The maximum this may be safely set to is 32.
7383 The startup= and idle= options allow some measure of skew in your
7388 Sets a minimum of how many processes are to be spawned when Squid
7389 starts or reconfigures. When set to zero the first request will
7390 cause spawning of the first child process to handle it.
7392 Starting too few will cause an initial slowdown in traffic as Squid
7393 attempts to simultaneously spawn enough processes to cope.
7397 Sets a minimum of how many processes Squid is to try and keep available
7398 at all times. When traffic begins to rise above what the existing
7399 processes can handle this many more will be spawned up to the maximum
7400 configured. A minimum setting of 1 is required.
7403 NAME: dns_retransmit_interval
7406 LOC: Config.Timeout.idns_retransmit
7407 IFDEF: !USE_DNSHELPER
7409 Initial retransmit interval for DNS queries. The interval is
7410 doubled each time all configured DNS servers have been tried.
7416 LOC: Config.Timeout.idns_query
7417 IFDEF: !USE_DNSHELPER
7419 DNS Query timeout. If no response is received to a DNS query
7420 within this time all DNS servers for the queried domain
7421 are assumed to be unavailable.
7424 NAME: dns_packet_max
7427 LOC: Config.dns.packet_max
7428 IFDEF: !USE_DNSHELPER
7430 Maximum number of bytes packet size to advertise via EDNS.
7431 Set to "none" to disable EDNS large packet support.
7433 For legacy reasons DNS UDP replies will default to 512 bytes which
7434 is too small for many responses. EDNS provides a means for Squid to
7435 negotiate receiving larger responses back immediately without having
7436 to failover with repeat requests. Responses larger than this limit
7437 will retain the old behaviour of failover to TCP DNS.
7439 Squid has no real fixed limit internally, but allowing packet sizes
7440 over 1500 bytes requires network jumbogram support and is usually not
7443 WARNING: The RFC also indicates that some older resolvers will reply
7444 with failure of the whole request if the extension is added. Some
7445 resolvers have already been identified which will reply with mangled
7446 EDNS response on occasion. Usually in response to many-KB jumbogram
7447 sizes being advertised by Squid.
7448 Squid will currently treat these both as an unable-to-resolve domain
7449 even if it would be resolvable without EDNS.
7456 LOC: Config.onoff.res_defnames
7458 Normally the RES_DEFNAMES resolver option is disabled
7459 (see res_init(3)). This prevents caches in a hierarchy
7460 from interpreting single-component hostnames locally. To allow
7461 Squid to handle single-component names, enable this option.
7464 NAME: dns_nameservers
7467 LOC: Config.dns_nameservers
7469 Use this if you want to specify a list of DNS name servers
7470 (IP addresses) to use instead of those given in your
7471 /etc/resolv.conf file.
7472 On Windows platforms, if no value is specified here or in
7473 the /etc/resolv.conf file, the list of DNS name servers are
7474 taken from the Windows registry, both static and dynamic DHCP
7475 configurations are supported.
7477 Example: dns_nameservers 10.0.0.1 192.172.0.4
7482 DEFAULT: @DEFAULT_HOSTS@
7483 LOC: Config.etcHostsPath
7485 Location of the host-local IP name-address associations
7486 database. Most Operating Systems have such a file on different
7488 - Un*X & Linux: /etc/hosts
7489 - Windows NT/2000: %SystemRoot%\system32\drivers\etc\hosts
7490 (%SystemRoot% value install default is c:\winnt)
7491 - Windows XP/2003: %SystemRoot%\system32\drivers\etc\hosts
7492 (%SystemRoot% value install default is c:\windows)
7493 - Windows 9x/Me: %windir%\hosts
7494 (%windir% value is usually c:\windows)
7495 - Cygwin: /etc/hosts
7497 The file contains newline-separated definitions, in the
7498 form ip_address_in_dotted_form name [name ...] names are
7499 whitespace-separated. Lines beginning with an hash (#)
7500 character are comments.
7502 The file is checked at startup and upon configuration.
7503 If set to 'none', it won't be checked.
7504 If append_domain is used, that domain will be added to
7505 domain-local (i.e. not containing any dot character) host
7511 LOC: Config.appendDomain
7514 Appends local domain name to hostnames without any dots in
7515 them. append_domain must begin with a period.
7517 Be warned there are now Internet names with no dots in
7518 them using only top-domain names, so setting this may
7519 cause some Internet sites to become unavailable.
7522 append_domain .yourdomain.com
7525 NAME: ignore_unknown_nameservers
7527 LOC: Config.onoff.ignore_unknown_nameservers
7529 IFDEF: !USE_DNSHELPER
7531 By default Squid checks that DNS responses are received
7532 from the same IP addresses they are sent to. If they
7533 don't match, Squid ignores the response and writes a warning
7534 message to cache.log. You can allow responses from unknown
7535 nameservers by setting this option to 'off'.
7541 LOC: Config.dns.v4_first
7542 IFDEF: !USE_DNSHELPER
7544 With the IPv6 Internet being as fast or faster than IPv4 Internet
7545 for most networks Squid prefers to contact websites over IPv6.
7547 This option reverses the order of preference to make Squid contact
7548 dual-stack websites over IPv4 first. Squid will still perform both
7549 IPv6 and IPv4 DNS lookups before connecting.
7552 This option will restrict the situations under which IPv6
7553 connectivity is used (and tested). Hiding network problems
7554 which would otherwise be detected and warned about.
7558 COMMENT: (number of entries)
7561 LOC: Config.ipcache.size
7568 LOC: Config.ipcache.low
7575 LOC: Config.ipcache.high
7577 The size, low-, and high-water marks for the IP cache.
7580 NAME: fqdncache_size
7581 COMMENT: (number of entries)
7584 LOC: Config.fqdncache.size
7586 Maximum number of FQDN cache entries.
7591 -----------------------------------------------------------------------------
7598 LOC: Config.onoff.mem_pools
7600 If set, Squid will keep pools of allocated (but unused) memory
7601 available for future use. If memory is a premium on your
7602 system and you believe your malloc library outperforms Squid
7603 routines, disable this.
7606 NAME: memory_pools_limit
7610 LOC: Config.MemPools.limit
7612 Used only with memory_pools on:
7613 memory_pools_limit 50 MB
7615 If set to a non-zero value, Squid will keep at most the specified
7616 limit of allocated (but unused) memory in memory pools. All free()
7617 requests that exceed this limit will be handled by your malloc
7618 library. Squid does not pre-allocate any memory, just safe-keeps
7619 objects that otherwise would be free()d. Thus, it is safe to set
7620 memory_pools_limit to a reasonably high value even if your
7621 configuration will use less memory.
7623 If set to none, Squid will keep all memory it can. That is, there
7624 will be no limit on the total amount of memory used for safe-keeping.
7626 To disable memory allocation optimization, do not set
7627 memory_pools_limit to 0 or none. Set memory_pools to "off" instead.
7629 An overhead for maintaining memory pools is not taken into account
7630 when the limit is checked. This overhead is close to four bytes per
7631 object kept. However, pools may actually _save_ memory because of
7632 reduced memory thrashing in your malloc library.
7636 COMMENT: on|off|transparent|truncate|delete
7639 LOC: opt_forwarded_for
7641 If set to "on", Squid will append your client's IP address
7642 in the HTTP requests it forwards. By default it looks like:
7644 X-Forwarded-For: 192.1.2.3
7646 If set to "off", it will appear as
7648 X-Forwarded-For: unknown
7650 If set to "transparent", Squid will not alter the
7651 X-Forwarded-For header in any way.
7653 If set to "delete", Squid will delete the entire
7654 X-Forwarded-For header.
7656 If set to "truncate", Squid will remove all existing
7657 X-Forwarded-For entries, and place the client IP as the sole entry.
7660 NAME: cachemgr_passwd
7661 TYPE: cachemgrpasswd
7663 LOC: Config.passwd_list
7665 Specify passwords for cachemgr operations.
7667 Usage: cachemgr_passwd password action action ...
7669 Some valid actions are (see cache manager menu for a full list):
7709 * Indicates actions which will not be performed without a
7710 valid password, others can be performed if not listed here.
7712 To disable an action, set the password to "disable".
7713 To allow performing an action without a password, set the
7716 Use the keyword "all" to set the same password for all actions.
7719 cachemgr_passwd secret shutdown
7720 cachemgr_passwd lesssssssecret info stats/objects
7721 cachemgr_passwd disable all
7728 LOC: Config.onoff.client_db
7730 If you want to disable collecting per-client statistics,
7731 turn off client_db here.
7734 NAME: refresh_all_ims
7738 LOC: Config.onoff.refresh_all_ims
7740 When you enable this option, squid will always check
7741 the origin server for an update when a client sends an
7742 If-Modified-Since request. Many browsers use IMS
7743 requests when the user requests a reload, and this
7744 ensures those clients receive the latest version.
7746 By default (off), squid may return a Not Modified response
7747 based on the age of the cached version.
7750 NAME: reload_into_ims
7751 IFDEF: USE_HTTP_VIOLATIONS
7755 LOC: Config.onoff.reload_into_ims
7757 When you enable this option, client no-cache or ``reload''
7758 requests will be changed to If-Modified-Since requests.
7759 Doing this VIOLATES the HTTP standard. Enabling this
7760 feature could make you liable for problems which it
7763 see also refresh_pattern for a more selective approach.
7766 NAME: connect_retries
7768 LOC: Config.connect_retries
7771 This sets the maximum number of connection attempts made for each
7772 TCP connection. The connect_retries attempts must all still
7773 complete within the connection timeout period.
7775 The default is not to re-try if the first connection attempt fails.
7776 The (not recommended) maximum is 10 tries.
7778 A warning message will be generated if it is set to a too-high
7779 value and the configured value will be over-ridden.
7781 Note: These re-tries are in addition to forward_max_tries
7782 which limit how many different addresses may be tried to find
7786 NAME: retry_on_error
7788 LOC: Config.retry.onerror
7791 If set to ON Squid will automatically retry requests when
7792 receiving an error response with status 403 (Forbidden),
7793 500 (Internal Error), 501 or 503 (Service not available).
7794 Status 502 and 504 (Gateway errors) are always retried.
7796 This is mainly useful if you are in a complex cache hierarchy to
7797 work around access control errors.
7799 NOTE: This retry will attempt to find another working destination.
7800 Which is different from the server which just failed.
7803 NAME: as_whois_server
7805 LOC: Config.as_whois_server
7806 DEFAULT: whois.ra.net
7808 WHOIS server to query for AS numbers. NOTE: AS numbers are
7809 queried only when Squid starts up, not for every request.
7814 LOC: Config.onoff.offline
7817 Enable this option and Squid will never try to validate cached
7821 NAME: uri_whitespace
7822 TYPE: uri_whitespace
7823 LOC: Config.uri_whitespace
7826 What to do with requests that have whitespace characters in the
7829 strip: The whitespace characters are stripped out of the URL.
7830 This is the behavior recommended by RFC2396.
7831 deny: The request is denied. The user receives an "Invalid
7833 allow: The request is allowed and the URI is not changed. The
7834 whitespace characters remain in the URI. Note the
7835 whitespace is passed to redirector processes if they
7837 encode: The request is allowed and the whitespace characters are
7838 encoded according to RFC1738. This could be considered
7839 a violation of the HTTP/1.1
7840 RFC because proxies are not allowed to rewrite URI's.
7841 chop: The request is allowed and the URI is chopped at the
7842 first whitespace. This might also be considered a
7848 LOC: Config.chroot_dir
7851 Specifies a directory where Squid should do a chroot() while
7852 initializing. This also causes Squid to fully drop root
7853 privileges after initializing. This means, for example, if you
7854 use a HTTP port less than 1024 and try to reconfigure, you may
7855 get an error saying that Squid can not open the port.
7858 NAME: balance_on_multiple_ip
7860 LOC: Config.onoff.balance_on_multiple_ip
7863 Modern IP resolvers in squid sort lookup results by preferred access.
7864 By default squid will use these IP in order and only rotates to
7865 the next listed when the most preffered fails.
7867 Some load balancing servers based on round robin DNS have been
7868 found not to preserve user session state across requests
7869 to different IP addresses.
7871 Enabling this directive Squid rotates IP's per request.
7874 NAME: pipeline_prefetch
7876 LOC: Config.onoff.pipeline_prefetch
7879 To boost the performance of pipelined requests to closer
7880 match that of a non-proxied environment Squid can try to fetch
7881 up to two requests in parallel from a pipeline.
7883 Defaults to off for bandwidth management and access logging
7886 WARNING: pipelining breaks NTLM and Negotiate/Kerberos authentication.
7889 NAME: high_response_time_warning
7892 LOC: Config.warnings.high_rptm
7895 If the one-minute median response time exceeds this value,
7896 Squid prints a WARNING with debug level 0 to get the
7897 administrators attention. The value is in milliseconds.
7900 NAME: high_page_fault_warning
7902 LOC: Config.warnings.high_pf
7905 If the one-minute average page fault rate exceeds this
7906 value, Squid prints a WARNING with debug level 0 to get
7907 the administrators attention. The value is in page faults
7911 NAME: high_memory_warning
7913 LOC: Config.warnings.high_memory
7916 If the memory usage (as determined by mallinfo) exceeds
7917 this amount, Squid prints a WARNING with debug level 0 to get
7918 the administrators attention.
7921 NAME: sleep_after_fork
7922 COMMENT: (microseconds)
7924 LOC: Config.sleep_after_fork
7927 When this is set to a non-zero value, the main Squid process
7928 sleeps the specified number of microseconds after a fork()
7929 system call. This sleep may help the situation where your
7930 system reports fork() failures due to lack of (virtual)
7931 memory. Note, however, if you have a lot of child
7932 processes, these sleep delays will add up and your
7933 Squid will not service requests for some amount of time
7934 until all the child processes have been started.
7935 On Windows value less then 1000 (1 milliseconds) are
7939 NAME: windows_ipaddrchangemonitor
7940 IFDEF: _SQUID_MSWIN_
7944 LOC: Config.onoff.WIN32_IpAddrChangeMonitor
7946 On Windows Squid by default will monitor IP address changes and will
7947 reconfigure itself after any detected event. This is very useful for
7948 proxies connected to internet with dial-up interfaces.
7949 In some cases (a Proxy server acting as VPN gateway is one) it could be
7950 desiderable to disable this behaviour setting this to 'off'.
7951 Note: after changing this, Squid service must be restarted.
7956 IFDEF: USE_SQUID_EUI
7958 LOC: Eui::TheConfig.euiLookup
7960 Whether to lookup the EUI or MAC address of a connected client.
7963 NAME: max_filedescriptors max_filedesc
7966 LOC: Config.max_filedescriptors
7968 The maximum number of filedescriptors supported.
7970 The default "0" means Squid inherits the current ulimit setting.
7972 Note: Changing this requires a restart of Squid. Also
7973 not all comm loops supports large values.
7981 Number of main Squid processes or "workers" to fork and maintain.
7982 0: "no daemon" mode, like running "squid -N ..."
7983 1: "no SMP" mode, start one main Squid process daemon (default)
7984 N: start N main Squid process daemons (i.e., SMP mode)
7986 In SMP mode, each worker does nearly all what a single Squid daemon
7987 does (e.g., listen on http_port and forward HTTP requests).
7990 NAME: cpu_affinity_map
7991 TYPE: CpuAffinityMap
7992 LOC: Config.cpuAffinityMap
7995 Usage: cpu_affinity_map process_numbers=P1,P2,... cores=C1,C2,...
7997 Sets 1:1 mapping between Squid processes and CPU cores. For example,
7999 cpu_affinity_map process_numbers=1,2,3,4 cores=1,3,5,7
8001 affects processes 1 through 4 only and places them on the first
8002 four even cores, starting with core #1.
8004 CPU cores are numbered starting from 1. Requires support for
8005 sched_getaffinity(2) and sched_setaffinity(2) system calls.
8007 Multiple cpu_affinity_map options are merged.