3 # $Id: cf.data.pre,v 1.264 2002/06/23 13:32:24 hno Exp $
6 # SQUID Web Proxy Cache http://www.squid-cache.org/
7 # ----------------------------------------------------------
9 # Squid is the result of efforts by numerous individuals from
10 # the Internet community; see the CONTRIBUTORS file for full
11 # details. Many organizations have provided support for Squid's
12 # development; see the SPONSORS file for full details. Squid is
13 # Copyrighted (C) 2000 by the Regents of the University of
14 # California; see the COPYRIGHT file for full details. Squid
15 # incorporates software developed and/or copyrighted by other
16 # sources; see the CREDITS file for full details.
18 # This program is free software; you can redistribute it and/or modify
19 # it under the terms of the GNU General Public License as published by
20 # the Free Software Foundation; either version 2 of the License, or
21 # (at your option) any later version.
23 # This program is distributed in the hope that it will be useful,
24 # but WITHOUT ANY WARRANTY; without even the implied warranty of
25 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
26 # GNU General Public License for more details.
28 # You should have received a copy of the GNU General Public License
29 # along with this program; if not, write to the Free Software
30 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA.
37 This is the default Squid configuration file. You may wish
38 to look at the Squid home page (http://www.squid-cache.org/)
39 for the FAQ and other documentation.
41 The default Squid config file shows what the defaults for
42 various options happen to be. If you don't need to change the
43 default, you shouldn't uncomment the line. Doing so may cause
44 run-time problems. In some cases "none" refers to no default
45 setting at all, while in other cases it refers to a valid
46 option - the comments for that keyword indicate if this is the
53 -----------------------------------------------------------------------------
56 NAME: http_port ascii_port
57 TYPE: sockaddr_in_list
59 LOC: Config.Sockaddr.http
65 The socket addresses where Squid will listen for HTTP client
66 requests. You may specify multiple socket addresses.
67 There are three forms: port alone, hostname with port, and
68 IP address with port. If you specify a hostname or IP
69 address, then Squid binds the socket to that specific
70 address. This replaces the old 'tcp_incoming_address'
71 option. Most likely, you do not need to bind to a specific
72 address, so you can use the port number alone.
74 If you are running Squid in accelerator mode, then you
75 probably want to listen on port 80 also, or instead.
77 The -a command line option will override the *first* port
78 number listed here. That option will NOT override an IP
81 You may specify multiple socket addresses on multiple lines.
83 If you run Squid on a dual-homed machine with an internal
84 and an external interface then we recommend you to specify the
85 internal address:port in http_port. This way Squid will only be
86 visible on the internal address.
89 # Squid normally listens to port 3128
98 LOC: Config.Sockaddr.https
100 Usage: [ip:]port cert=certificate.pem [key=key.pem] [options...]
102 The socket address where Squid will listen for HTTPS client
105 This is really only useful for situations where you are running
106 squid in accelerator mode and you want to do the SSL work at the
109 You may specify multiple socket addresses on multiple lines,
110 each with their own SSL certificate and/or options.
114 cert= Path to SSL certificate (PEM format)
116 key= Path to SSL private key file (PEM format)
117 if not specified, the certificate file is
118 assumed to be a combined certificate and
121 version= The version of SSL/TLS supported
122 1 automatic (default)
127 cipher= Colon separated list of supported ciphers
129 options= Varions SSL engine options. The most important
131 NO_SSLv2 Disallow the use of SSLv2
132 NO_SSLv3 Disallow the use of SSLv3
133 NO_TLSv1 Disallow the use of TLSv1
134 See src/ssl_support.c or OpenSSL documentation
135 for a more complete list.
138 NAME: ssl_unclean_shutdown
142 LOC: Config.SSL.unclean_shutdown
144 Some browsers (especially MSIE) bugs out on SSL shutdown
148 NAME: icp_port udp_port
153 The port number where Squid sends and receives ICP queries to
154 and from neighbor caches. Default is 3130. To disable use
155 "0". May be overridden with -u on the command line.
162 LOC: Config.Port.htcp
164 The port number where Squid sends and receives HTCP queries to
165 and from neighbor caches. Default is 4827. To disable use
172 LOC: Config.mcast_group_list
175 This tag specifies a list of multicast groups which your server
176 should join to receive multicasted ICP queries.
178 NOTE! Be very careful what you put here! Be sure you
179 understand the difference between an ICP _query_ and an ICP
180 _reply_. This option is to be set only if you want to RECEIVE
181 multicast queries. Do NOT set this option to SEND multicast
182 ICP (use cache_peer for that). ICP replies are always sent via
183 unicast, so this option does not affect whether or not you will
184 receive replies from multicast group members.
186 You must be very careful to NOT use a multicast address which
187 is already in use by another group of caches.
189 If you are unsure about multicast, please read the Multicast
190 chapter in the Squid FAQ (http://www.squid-cache.org/FAQ/).
192 Usage: mcast_groups 239.128.16.128 224.0.1.20
194 By default, Squid doesn't listen on any multicast groups.
198 NAME: udp_incoming_address
200 LOC:Config.Addrs.udp_incoming
204 NAME: udp_outgoing_address
206 LOC: Config.Addrs.udp_outgoing
207 DEFAULT: 255.255.255.255
209 udp_incoming_address is used for the ICP socket receiving packets
211 udp_outgoing_address is used for ICP packets sent out to other
214 The default behavior is to not bind to any specific address.
216 A udp_incoming_address value of 0.0.0.0 indicates that Squid should
217 listen for UDP messages on all available interfaces.
219 If udp_outgoing_address is set to 255.255.255.255 (the default)
220 then it will use the same socket as udp_incoming_address. Only
221 change this if you want to have ICP queries sent using another
222 address than where this Squid listens for ICP queries from other
225 NOTE, udp_incoming_address and udp_outgoing_address can not
226 have the same value since they both use port 3130.
230 OPTIONS WHICH AFFECT THE NEIGHBOR SELECTION ALGORITHM
231 -----------------------------------------------------------------------------
239 To specify other caches in a hierarchy, use the format:
241 cache_peer hostname type http_port icp_port
246 # hostname type port port options
247 # -------------------- -------- ----- ----- -----------
248 cache_peer parent.foo.net parent 3128 3130 [proxy-only]
249 cache_peer sib1.foo.net sibling 3128 3130 [proxy-only]
250 cache_peer sib2.foo.net sibling 3128 3130 [proxy-only]
252 type: either 'parent', 'sibling', or 'multicast'.
254 proxy_port: The port number where the cache listens for proxy
257 icp_port: Used for querying neighbor caches about
258 objects. To have a non-ICP neighbor
259 specify '7' for the ICP port and make sure the
260 neighbor machine has the UDP echo port
261 enabled in its /etc/inetd.conf file.
274 login=user:password | PASS | *:password
280 use 'proxy-only' to specify that objects fetched
281 from this cache should not be saved locally.
283 use 'weight=n' to specify a weighted parent.
284 The weight must be an integer. The default weight
285 is 1, larger weights are favored more.
287 use 'ttl=n' to specify a IP multicast TTL to use
288 when sending an ICP queries to this address.
289 Only useful when sending to a multicast group.
290 Because we don't accept ICP replies from random
291 hosts, you must configure other group members as
292 peers with the 'multicast-responder' option below.
294 use 'no-query' to NOT send ICP queries to this
297 use 'default' if this is a parent cache which can
298 be used as a "last-resort." You should probably
299 only use 'default' in situations where you cannot
300 use ICP with your parent cache(s).
302 use 'round-robin' to define a set of parents which
303 should be used in a round-robin fashion in the
304 absence of any ICP queries.
306 'multicast-responder' indicates that the named peer
307 is a member of a multicast group. ICP queries will
308 not be sent directly to the peer, but ICP replies
309 will be accepted from it.
311 'closest-only' indicates that, for ICP_OP_MISS
312 replies, we'll only forward CLOSEST_PARENT_MISSes
313 and never FIRST_PARENT_MISSes.
315 use 'no-digest' to NOT request cache digests from
318 'no-netdb-exchange' disables requesting ICMP
319 RTT database (NetDB) from the neighbor.
321 use 'no-delay' to prevent access to this neighbor
322 from influencing the delay pools.
324 use 'login=user:password' if this is a personal/workgroup
325 proxy and your parent requires proxy authentication.
326 Note: The string can include URL escapes (i.e. %20 for
327 spaces). This also means that % must be written as %%.
329 use 'login=PASS' if users must authenticate against
330 the upstream proxy. This will pass the users credentials
331 as they are to the peer proxy. This only works for the
332 Basic HTTP authentication sheme. Note: To combine this
333 with proxy_auth both proxies must share the same user
334 database as HTTP only allows for one proxy login.
335 Also be warned that this will expose your users proxy
336 password to the peer. USE WITH CAUTION
338 use 'login=*:password' to pass the username to the
339 upstream cache, but with a fixed password. This is meant
340 to be used when the peer is in another administrative
341 domain, but it is still needed to identify each user.
342 The star can optionally be followed by some extra
343 information which is added to the username. This can
344 be used to identify this proxy to the peer, similar to
345 the login=username:password option above.
347 use 'connect-timeout=nn' to specify a peer
348 specific connect timeout (also see the
349 peer_connect_timeout directive)
351 use 'digest-url=url' to tell Squid to fetch the cache
352 digest (if digests are enabled) for this host from
353 the specified URL rather than the Squid default
356 use 'allow-miss' to disable Squid's use of only-if-cached
357 when forwarding requests to siblings. This is primarily
358 useful when icp_hit_stale is used by the sibling. To
359 extensive use of this option may result in forwarding
360 loops, and you should avoid having two-way peerings
361 with this option. (for example to deny peer usage on
362 requests from peer by denying cache_peer_access if the
365 use 'max-conn' to limit the amount of connections Squid
366 may open to this peer.
368 NOTE: non-ICP neighbors must be specified as 'parent'.
371 NAME: cache_peer_domain cache_host_domain
376 Use to limit the domains for which a neighbor cache will be
379 cache_peer_domain cache-host domain [domain ...]
380 cache_peer_domain cache-host !domain
382 For example, specifying
384 cache_peer_domain parent.foo.net .edu
386 has the effect such that UDP query packets are sent to
387 'bigserver' only when the requested object exists on a
388 server in the .edu domain. Prefixing the domainname
389 with '!' means that the cache will be queried for objects
392 NOTE: * Any number of domains may be given for a cache-host,
393 either on the same or separate lines.
394 * When multiple domains are given for a particular
395 cache-host, the first matched domain is applied.
396 * Cache hosts with no domain restrictions are queried
398 * There are no defaults.
399 * There is also a 'cache_peer_access' tag in the ACL
404 NAME: neighbor_type_domain
409 usage: neighbor_type_domain parent|sibling domain domain ...
411 Modifying the neighbor type for specific domains is now
412 possible. You can treat some domains differently than the the
413 default neighbor type specified on the 'cache_peer' line.
414 Normally it should only be necessary to list domains which
415 should be treated differently because the default neighbor type
416 applies for hostnames which do not match domains listed here.
419 cache_peer parent cache.foo.org 3128 3130
420 neighbor_type_domain cache.foo.org sibling .com .net
421 neighbor_type_domain cache.foo.org sibling .au .de
424 NAME: icp_query_timeout
428 LOC: Config.Timeout.icp_query
430 Normally Squid will automatically determine an optimal ICP
431 query timeout value based on the round-trip-time of recent ICP
432 queries. If you want to override the value determined by
433 Squid, set this 'icp_query_timeout' to a non-zero value. This
434 value is specified in MILLISECONDS, so, to use a 2-second
435 timeout (the old default), you would write:
437 icp_query_timeout 2000
440 NAME: maximum_icp_query_timeout
444 LOC: Config.Timeout.icp_query_max
446 Normally the ICP query timeout is determined dynamically. But
447 sometimes it can lead to very large values (say 5 seconds).
448 Use this option to put an upper limit on the dynamic timeout
449 value. Do NOT use this option to always use a fixed (instead
450 of a dynamic) timeout value. To set a fixed timeout see the
451 'icp_query_timeout' directive.
454 NAME: minimum_icp_query_timeout
458 LOC: Config.Timeout.icp_query_min
460 Normally the ICP query timeout is determined dynamically. But
461 sometimes it can lead to very small timeouts, even lower than
462 the normal latency variance on your link due to traffic.
463 Use this option to put an lower limit on the dynamic timeout
464 value. Do NOT use this option to always use a fixed (instead
465 of a dynamic) timeout value. To set a fixed timeout see the
466 'icp_query_timeout' directive.
469 NAME: mcast_icp_query_timeout
473 LOC: Config.Timeout.mcast_icp_query
475 For Multicast peers, Squid regularly sends out ICP "probes" to
476 count how many other peers are listening on the given multicast
477 address. This value specifies how long Squid should wait to
478 count all the replies. The default is 2000 msec, or 2
482 NAME: dead_peer_timeout
486 LOC: Config.Timeout.deadPeer
488 This controls how long Squid waits to declare a peer cache
489 as "dead." If there are no ICP replies received in this
490 amount of time, Squid will declare the peer dead and not
491 expect to receive any further ICP replies. However, it
492 continues to send ICP queries, and will mark the peer as
493 alive upon receipt of the first subsequent ICP reply.
495 This timeout also affects when Squid expects to receive ICP
496 replies from peers. If more than 'dead_peer' seconds have
497 passed since the last ICP reply was received, Squid will not
498 expect to receive an ICP reply on the next query. Thus, if
499 your time between requests is greater than this timeout, you
500 will see a lot of requests sent DIRECT to origin servers
501 instead of to your parents.
505 NAME: hierarchy_stoplist
508 LOC: Config.hierarchy_stoplist
510 A list of words which, if found in a URL, cause the object to
511 be handled directly by this cache. In other words, use this
512 to not query neighbor caches for certain objects. You may
513 list this option multiple times.
515 #We recommend you to use at least the following line.
516 hierarchy_stoplist cgi-bin ?
524 LOC: Config.accessList.noCache
526 A list of ACL elements which, if matched, cause the request to
527 not be satisfied from the cache and the reply to not be cached.
528 In other words, use this to force certain objects to never be cached.
530 You must use the word 'DENY' to indicate the ACL names which should
534 #We recommend you to use the following two lines.
535 acl QUERY urlpath_regex cgi-bin \?
542 OPTIONS WHICH AFFECT THE CACHE SIZE
543 -----------------------------------------------------------------------------
550 LOC: Config.memMaxSize
552 NOTE: THIS PARAMETER DOES NOT SPECIFY THE MAXIMUM PROCESS SIZE.
553 IT ONLY PLACES A LIMIT ON HOW MUCH ADDITIONAL MEMORY SQUID WILL
554 USE AS A MEMORY CACHE OF OBJECTS. SQUID USES MEMORY FOR OTHER
555 THINGS AS WELL. SEE THE SQUID FAQ SECTION 8 FOR DETAILS.
557 'cache_mem' specifies the ideal amount of memory to be used
561 * Negative-Cached objects
563 Data for these objects are stored in 4 KB blocks. This
564 parameter specifies the ideal upper limit on the total size of
565 4 KB blocks allocated. In-Transit objects take the highest
568 In-transit objects have priority over the others. When
569 additional space is needed for incoming data, negative-cached
570 and hot objects will be released. In other words, the
571 negative-cached and hot objects will fill up any unused space
572 not needed for in-transit objects.
574 If circumstances require, this limit will be exceeded.
575 Specifically, if your incoming request rate requires more than
576 'cache_mem' of memory to hold in-transit objects, Squid will
577 exceed this limit to satisfy the new requests. When the load
578 decreases, blocks will be freed until the high-water mark is
579 reached. Thereafter, blocks will be used to store hot
585 COMMENT: (percent, 0-100)
588 LOC: Config.Swap.lowWaterMark
591 NAME: cache_swap_high
592 COMMENT: (percent, 0-100)
595 LOC: Config.Swap.highWaterMark
598 The low- and high-water marks for cache object replacement.
599 Replacement begins when the swap (disk) usage is above the
600 low-water mark and attempts to maintain utilization near the
601 low-water mark. As swap utilization gets close to high-water
602 mark object eviction becomes more aggressive. If utilization is
603 close to the low-water mark less replacement is done each time.
605 Defaults are 90% and 95%. If you have a large cache, 5% could be
606 hundreds of MB. If this is the case you may wish to set these
607 numbers closer together.
610 NAME: maximum_object_size
614 LOC: Config.Store.maxObjectSize
616 Objects larger than this size will NOT be saved on disk. The
617 value is specified in kilobytes, and the default is 4MB. If
618 you wish to get a high BYTES hit ratio, you should probably
619 increase this (one 32 MB object hit counts for 3200 10KB
620 hits). If you wish to increase speed more than your want to
621 save bandwidth you should leave this low.
623 NOTE: if using the LFUDA replacement policy you should increase
624 this value to maximize the byte hit rate improvement of LFUDA!
625 See replacement_policy below for a discussion of this policy.
628 NAME: minimum_object_size
632 LOC: Config.Store.minObjectSize
634 Objects smaller than this size will NOT be saved on disk. The
635 value is specified in kilobytes, and the default is 0 KB, which
636 means there is no minimum.
639 NAME: maximum_object_size_in_memory
643 LOC: Config.Store.maxInMemObjSize
645 Objects greater than this size will not be attempted to kept in
646 the memory cache. This should be set high enough to keep objects
647 accessed frequently in memory to improve performance whilst low
648 enough to keep larger objects from hoarding cache_mem .
652 COMMENT: (number of entries)
655 LOC: Config.ipcache.size
662 LOC: Config.ipcache.low
669 LOC: Config.ipcache.high
671 The size, low-, and high-water marks for the IP cache.
675 COMMENT: (number of entries)
678 LOC: Config.fqdncache.size
680 Maximum number of FQDN cache entries.
683 NAME: cache_replacement_policy
685 LOC: Config.replPolicy
688 The cache replacement policy parameter determines which
689 objects are evicted (replaced) when disk space is needed.
691 lru : Squid's original list based LRU policy
692 heap GDSF : Greedy-Dual Size Frequency
693 heap LFUDA: Least Frequently Used with Dynamic Aging
694 heap LRU : LRU policy implemented using a heap
696 Applies to any cache_dir lines listed below this.
698 The LRU policies keeps recently referenced objects.
700 The heap GDSF policy optimizes object hit rate by keeping smaller
701 popular objects in cache so it has a better chance of getting a
702 hit. It achieves a lower byte hit rate than LFUDA though since
703 it evicts larger (possibly popular) objects.
705 The heap LFUDA policy keeps popular objects in cache regardless of
706 their size and thus optimizes byte hit rate at the expense of
707 hit rate since one large, popular object will prevent many
708 smaller, slightly less popular objects from being cached.
710 Both policies utilize a dynamic aging mechanism that prevents
711 cache pollution that can otherwise occur with frequency-based
712 replacement policies.
714 NOTE: if using the LFUDA replacement policy you should increase
715 the value of maximum_object_size above its default of 4096 KB to
716 to maximize the potential byte hit rate improvement of LFUDA.
718 For more information about the GDSF and LFUDA cache replacement
719 policies see http://www.hpl.hp.com/techreports/1999/HPL-1999-69.html
720 and http://fog.hpl.external.hp.com/techreports/98/HPL-98-173.html.
723 NAME: memory_replacement_policy
725 LOC: Config.memPolicy
728 The memory replacement policy parameter determines which
729 objects are purged from memory when memory space is needed.
731 See cache_replacement_policy for details.
736 LOGFILE PATHNAMES AND CACHE DIRECTORIES
737 -----------------------------------------------------------------------------
743 DEFAULT_IF_NONE: ufs @DEFAULT_SWAP_DIR@ 100 16 256
744 LOC: Config.cacheSwap
748 cache_dir Type Directory-Name Fs-specific-data [options]
750 cache_dir diskd Maxobjsize Directory-Name MB L1 L2 Q1 Q2
752 You can specify multiple cache_dir lines to spread the
753 cache among different disk partitions.
755 Type specifies the kind of storage system to use. Only "ufs"
756 is built by default. To eanble any of the other storage systems
757 see the --enable-storeio configure option.
759 'Directory' is a top-level directory where cache swap
760 files will be stored. If you want to use an entire disk
761 for caching, then this can be the mount-point directory.
762 The directory must exist and be writable by the Squid
763 process. Squid will NOT create this directory for you.
767 "ufs" is the old well-known Squid storage format that has always
770 cache_dir ufs Directory-Name Mbytes L1 L2 [options]
772 'Mbytes' is the amount of disk space (MB) to use under this
773 directory. The default is 100 MB. Change this to suit your
776 'Level-1' is the number of first-level subdirectories which
777 will be created under the 'Directory'. The default is 16.
779 'Level-2' is the number of second-level subdirectories which
780 will be created under each first-level directory. The default
785 "aufs" uses the same storage format as "ufs", utilizing
786 POSIX-threads to avoid blocking the main Squid process on
787 disk-I/O. This was formerly known in Squid as async-io.
789 cache_dir aufs Directory-Name Mbytes L1 L2 [options]
791 see argument descriptions under ufs above
793 The diskd store type:
795 "diskd" uses the same storage format as "ufs", utilizing a
796 separate process to avoid blocking the main Squid process on
799 cache_dir diskd Directory-Name Mbytes L1 L2 [options] [Q1=n] [Q2=n]
801 see argument descriptions under ufs above
803 Q1 specifies the number of unacknowledged I/O requests when Squid
804 stops opening new files. If this many messages are in the queues,
805 Squid won't open new files. Default is 64
807 Q2 specifies the number of unacknowledged messages when Squid
808 starts blocking. If this many messages are in the queues,
809 Squid blocks until it recevies some replies. Default is 72
813 read-only, this cache_dir is read only.
815 max-size=n, refers to the max object size this storedir supports.
816 It is used to initially choose the storedir to dump the object.
817 Note: To make optimal use of the max-size limits you should order
818 the cache_dir lines with the smallest max-size value first and the
819 ones with no max-size specification last.
823 NAME: cache_access_log
825 DEFAULT: @DEFAULT_ACCESS_LOG@
826 LOC: Config.Log.access
828 Logs the client request activity. Contains an entry for
829 every HTTP and ICP queries received. To disable, enter "none".
835 DEFAULT: @DEFAULT_CACHE_LOG@
838 Cache logging file. This is where general information about
839 your cache's behavior goes. You can increase the amount of data
840 logged to this file with the "debug_options" tag below.
844 NAME: cache_store_log
846 DEFAULT: @DEFAULT_STORE_LOG@
847 LOC: Config.Log.store
849 Logs the activities of the storage manager. Shows which
850 objects are ejected from the cache, and which objects are
851 saved and for how long. To disable, enter "none". There are
852 not really utilities to analyze this data, so you can safely
862 Location for the cache "swap.log." This log file holds the
863 metadata of objects saved on disk. It is used to rebuild the
864 cache during startup. Normally this file resides in each
865 'cache_dir' directory, but you may specify an alternate
866 pathname here. Note you must give a full filename, not just
867 a directory. Since this is the index for the whole object
868 list you CANNOT periodically rotate it!
870 If %s can be used in the file name then it will be replaced with a
871 a representation of the cache_dir name where each / is replaced
872 with '.'. This is needed to allow adding/removing cache_dir
873 lines when cache_swap_log is being used.
875 If have more than one 'cache_dir', and %s is not used in the name
876 then these swap logs will have names such as:
882 The numbered extension (which is added automatically)
883 corresponds to the order of the 'cache_dir' lines in this
884 configuration file. If you change the order of the 'cache_dir'
885 lines in this file, then these log files will NOT correspond to
886 the correct 'cache_dir' entry (unless you manually rename
887 them). We recommend that you do NOT use this option. It is
888 better to keep these log files in each 'cache_dir' directory.
892 NAME: emulate_httpd_log
896 LOC: Config.onoff.common_log
898 The Cache can emulate the log file format which many 'httpd'
899 programs use. To disable/enable this emulation, set
900 emulate_httpd_log to 'off' or 'on'. The default
901 is to use the native log format since it includes useful
902 information that Squid-specific log analyzers use.
905 NAME: log_ip_on_direct
909 LOC: Config.onoff.log_ip_on_direct
911 Log the destination IP address in the hierarchy log tag when going
912 direct. Earlier Squid versions logged the hostname here. If you
913 prefer the old way set this to off.
918 DEFAULT: @DEFAULT_MIME_TABLE@
919 LOC: Config.mimeTablePathname
921 Pathname to Squid's MIME table. You shouldn't need to change
922 this, but the default file contains examples and formatting
923 information if you do.
930 LOC: Config.onoff.log_mime_hdrs
933 The Cache can record both the request and the response MIME
934 headers for each HTTP transaction. The headers are encoded
935 safely and will appear as two bracketed fields at the end of
936 the access log (for either the native or httpd-emulated log
937 formats). To enable this logging set log_mime_hdrs to 'on'.
943 LOC: Config.Log.useragent
945 IFDEF: USE_USERAGENT_LOG
947 Squid will write the User-Agent field from HTTP requests
948 to the filename specified here. By default useragent_log
955 LOC: Config.Log.referer
957 IFDEF: USE_REFERER_LOG
959 Squid will write the Referer field from HTTP requests to the
960 filename specified here. By default referer_log is disabled.
966 DEFAULT: @DEFAULT_PID_FILE@
967 LOC: Config.pidFilename
969 A filename to write the process-id to. To disable, enter "none".
976 LOC: Config.debugOptions
978 Logging options are set as section,level where each source file
979 is assigned a unique section. Lower levels result in less
980 output, Full debugging (level 9) can result in a very large
981 log file, so be careful. The magic word "ALL" sets debugging
982 levels for all sections. We recommend normally running with
991 LOC: Config.onoff.log_fqdn
993 Turn this on if you wish to log fully qualified domain names
994 in the access.log. To do this Squid does a DNS lookup of all
995 IP's connecting to it. This can (in some situations) increase
996 latency, which makes your cache seem slower for interactive
1001 NAME: client_netmask
1003 LOC: Config.Addrs.client_netmask
1004 DEFAULT: 255.255.255.255
1006 A netmask for client addresses in logfiles and cachemgr output.
1007 Change this to protect the privacy of your cache clients.
1008 A netmask of 255.255.255.0 will log all IP's in that range with
1009 the last digit set to '0'.
1014 OPTIONS FOR EXTERNAL SUPPORT PROGRAMS
1015 -----------------------------------------------------------------------------
1021 LOC: Config.Ftp.anon_user
1023 If you want the anonymous login password to be more informative
1024 (and enable the use of picky ftp servers), set this to something
1025 reasonable for your domain, like wwwuser@somewhere.net
1027 The reason why this is domainless by default is that the
1028 request can be made on the behalf of a user in any domain,
1029 depending on how the cache is used.
1030 Some ftp server also validate that the email address is valid
1031 (for example perl.com).
1034 NAME: ftp_list_width
1037 LOC: Config.Ftp.list_width
1039 Sets the width of ftp listings. This should be set to fit in
1040 the width of a standard browser. Setting this too small
1041 can cut off long filenames when browsing ftp sites.
1047 LOC: Config.Ftp.passive
1049 If your firewall does not allow Squid to use passive
1050 connections, then turn off this option.
1053 NAME: ftp_sanitycheck
1056 LOC: Config.Ftp.sanitycheck
1058 For security and data integrity reasons Squid by default performs
1059 sanity checks of the addresses of FTP data connections ensure the
1060 data connection is to the requested server. If you need to allow
1061 FTP connections to servers using another IP address for the data
1062 connection then turn this off.
1065 NAME: cache_dns_program
1067 IFDEF: USE_DNSSERVERS
1068 DEFAULT: @DEFAULT_DNSSERVER@
1069 LOC: Config.Program.dnsserver
1071 Specify the location of the executable for dnslookup process.
1076 IFDEF: USE_DNSSERVERS
1078 LOC: Config.dnsChildren
1080 The number of processes spawn to service DNS name lookups.
1081 For heavily loaded caches on large servers, you should
1082 probably increase this value to at least 10. The maximum
1083 is 32. The default is 5.
1085 You must have at least one dnsserver process.
1088 NAME: dns_retransmit_interval
1091 LOC: Config.Timeout.idns_retransmit
1092 IFDEF: !USE_DNSSERVERS
1094 Initial retransmit interval for DNS queries. The interval is
1095 doubled each time all configured DNS servers have been tried.
1102 LOC: Config.Timeout.idns_query
1103 IFDEF: !USE_DNSSERVERS
1105 DNS Query timeout. If no response is received to a DNS query
1106 within this time then all DNS servers for the queried domain
1107 is assumed to be unavailable.
1112 IFDEF: USE_DNSSERVERS
1115 LOC: Config.onoff.res_defnames
1116 IFDEF: USE_DNSSERVERS
1118 Normally the 'dnsserver' disables the RES_DEFNAMES resolver
1119 option (see res_init(3)). This prevents caches in a hierarchy
1120 from interpreting single-component hostnames locally. To allow
1121 dnsserver to handle single-component names, enable this
1125 NAME: dns_nameservers
1128 LOC: Config.dns_nameservers
1130 Use this if you want to specify a list of DNS name servers
1131 (IP addresses) to use instead of those given in your
1132 /etc/resolv.conf file.
1134 Example: dns_nameservers 10.0.0.1 192.172.0.4
1140 LOC: Config.etcHostsPath
1142 Location of the host-local IP name-address associations
1143 database. Most Operating Systems have such a file: under
1144 Un*X it's by default in /etc/hosts MS-Windows NT/2000 places
1145 that in %SystemRoot%(by default
1146 c:\winnt)\system32\drivers\etc\hosts, while Windows 9x/ME
1147 places that in %windir%(usually c:\windows)\hosts
1149 The file contains newline-separated definitions, in the
1150 form ip_address_in_dotted_form name [name ...] names are
1151 whitespace-separated. lines beginnng with an hash (#)
1152 character are comments.
1154 The file is checked at startup and upon configuration. If
1155 set to 'none', it won't be checked. If append_domain is
1156 used, that domain will be added to domain-local (i.e. not
1157 containing any dot character) host definitions.
1162 DEFAULT: @DEFAULT_DISKD@
1163 LOC: Config.Program.diskd
1165 Specify the location of the diskd executable.
1166 Note that this is only useful if you have compiled in
1167 diskd as one of the store io modules.
1170 NAME: unlinkd_program
1173 DEFAULT: @DEFAULT_UNLINKD@
1174 LOC: Config.Program.unlinkd
1176 Specify the location of the executable for file deletion process.
1179 NAME: pinger_program
1181 DEFAULT: @DEFAULT_PINGER@
1182 LOC: Config.Program.pinger
1185 Specify the location of the executable for the pinger process.
1189 NAME: redirect_program
1191 LOC: Config.Program.redirect
1194 Specify the location of the executable for the URL redirector.
1195 Since they can perform almost any function there isn't one included.
1196 See the FAQ (section 15) for information on how to write one.
1197 By default, a redirector is not used.
1201 NAME: redirect_children
1204 LOC: Config.redirectChildren
1206 The number of redirector processes to spawn. If you start
1207 too few Squid will have to wait for them to process a backlog of
1208 URLs, slowing it down. If you start too many they will use RAM
1209 and other system resources.
1212 NAME: redirect_rewrites_host_header
1215 LOC: Config.onoff.redir_rewrites_host
1217 By default Squid rewrites any Host: header in redirected
1218 requests. If you are running an accelerator then this may
1219 not be a wanted effect of a redirector.
1222 NAME: redirector_access
1225 LOC: Config.accessList.redirector
1227 If defined, this access list specifies which requests are
1228 sent to the redirector processes. By default all requests
1234 LOC: Config.authConfig
1237 This is used to pass parameters to the various authentication
1239 format: auth_param scheme parameter [setting]
1241 auth_param basic program @DEFAULT_PREFIX@/bin/ncsa_auth @DEFAULT_PREFIX@/etc/passwd
1242 would tell the basic authentication scheme it's program parameter.
1244 The order that authentication prompts are presented to the client_agent
1245 is dependant on the order the scheme first appears in config file.
1246 IE has a bug (it's not rfc 2617 compliant) in that it will use the basic
1247 scheme if basic is the first entry presented, even if more secure schemes
1248 are presented. For now use the order in the file below. If other browsers
1249 have difficulties (don't recognise the schemes offered even if you are using
1250 basic) then either put basic first, or disable the other schemes (by commenting
1251 out their program entry).
1253 Once an authentication scheme is fully configured, it can only be shutdown
1254 by shutting squid down and restarting. Changes can be made on the fly and
1255 activated with a reconfigure. I.E. You can change to a different helper,
1256 but not unconfigure the helper completely.
1258 === Parameters for the basic scheme follow. ===
1261 Specify the command for the external authenticator. Such a
1262 program reads a line containing "username password" and replies
1263 "OK" or "ERR" in an endless loop. If you use an authenticator,
1264 make sure you have 1 acl of type proxy_auth. By default, the
1265 authenticate_program is not used.
1267 If you want to use the traditional proxy authentication,
1268 jump over to the ../auth_modules/NCSA directory and
1273 Then, set this line to something like
1275 auth_param basic program @DEFAULT_PREFIX@/bin/ncsa_auth @DEFAULT_PREFIX@/etc/passwd
1277 "children" numberofchildren
1278 The number of authenticator processes to spawn (no default). If you
1279 start too few Squid will have to wait for them to process a backlog
1280 of usercode/password verifications, slowing it down. When password
1281 verifications are done via a (slow) network you are likely to need
1282 lots of authenticator processes.
1283 auth_param basic children 5
1286 Specifies the realm name which is to be reported to the client for
1287 the basic proxy authentication scheme (part of the text the user will
1288 see when prompted their username and password). Their is no default.
1289 auth_param basic realm Squid proxy-caching web server
1291 "credentialsttl" timetolive
1292 Specifies how long squid assumes an externally validated username:password
1293 pair is valid for - in other words how often the helper program is called
1294 for that user. Set this low to force revalidation with short lived passwords.
1295 Note that setting this high does not impact your susceptability to replay
1296 attacks unless you are using an one-time password system (such as SecureID).
1297 If you are using such a system, you will be vulnerable to replay attacks
1298 unless you also enable the IP ttl is strict option.
1300 === Parameters for the digest scheme follow ===
1303 Specify the command for the external authenticator. Such a
1304 program reads a line containing "username":"realm" and replies
1305 with the appropriate H(A1) value base64 encoded. See rfc 2616 for
1306 the definition of H(A1). If you use an authenticator,
1307 make sure you have 1 acl of type proxy_auth. By default,
1308 authentication is not used.
1310 If you want to use build an authenticator,
1311 jump over to the ../digest_auth_modules directory and choose the
1312 authenticator to use. It it's directory type
1316 Then, set this line to something like
1318 auth_param digest program @DEFAULT_PREFIX@/bin/digest_auth_pw @DEFAULT_PREFIX@/etc/digpass
1321 "children" numberofchildren
1322 The number of authenticator processes to spawn (no default). If you
1323 start too few Squid will have to wait for them to process a backlog
1324 of H(A1) calculations, slowing it down. When the H(A1) calculations
1325 are done via a (slow) network you are likely to need lots of
1326 authenticator processes.
1327 auth_param digest children 5
1330 Specifies the realm name which is to be reported to the client for
1331 the digest proxy authentication scheme (part of the text the user will
1332 see when prompted their username and password). There is no default.
1333 auth_param digest realm Squid proxy-caching web server
1335 "nonce_garbage_interval" timeinterval
1336 Specifies the interval that nonces that have been issued to client_agent's
1337 are checked for validity.
1339 "nonce_max_duration" timeinterval
1340 Specifies the maximum length of time a given nonce will be valid for.
1342 "nonce_max_count" number
1343 Specifies the maximum number of times a given nonce can be used.
1345 "nonce_strictness" on|off
1346 Determines if squid requires increment-by-1 behaviour for nonce counts
1347 (on - the default), or strictly incrementing (off - for use when useragents
1348 generate nonce counts that occasionally miss 1 (ie, 1,2,4,6)).
1350 === NTLM scheme options follow ===
1353 Specify the command for the external ntlm authenticator. Such a
1354 program reads a line containing the uuencoded NEGOTIATE and replies
1355 with the ntlm CHALLENGE, then waits for the response and answers with
1356 "OK" or "ERR" in an endless loop. If you use an ntlm authenticator,
1357 make sure you have 1 acl of type proxy_auth. By default, the
1358 ntlm authenticator_program is not used.
1360 auth_param ntlm program @DEFAULT_PREFIX@/bin/ntlm_auth
1362 "children" numberofchildren
1363 The number of authenticator processes to spawn (no default). If you
1364 start too few Squid will have to wait for them to process a backlog
1365 of credential verifications, slowing it down. When crendential
1366 verifications are done via a (slow) network you are likely to need
1367 lots of authenticator processes.
1368 auth_param ntlm children 5
1370 "max_challenge_reuses" number
1371 The maximum number of times a challenge given by a ntlm authentication
1372 helper can be reused. Increasing this number increases your exposure
1373 to replay attacks on your network. 0 means use the challenge only once.
1374 (disable challenge caching)
1375 See max_ntlm_challenge_lifetime for more information.
1376 auth_param ntlm max_challenge_reuses 0
1378 "max_challenge_lifetime" timespan
1379 The maximum time period that a ntlm challenge is reused over.
1380 The actual period will be the minimum of this time AND the number of
1382 auth_param ntlm max_challenge_lifetime 2 minutes
1385 #Recommended minimum configuration:
1386 #auth_param digest program <uncomment and complete this line>
1387 #auth_param digest children 5
1388 #auth_param digest realm Squid proxy-caching web server
1389 #auth_param digest nonce_garbage_interval 5 minutes
1390 #auth_param digest nonce_max_duration 30 minutes
1391 #auth_param digest nonce_max_count 50
1392 #auth_param ntlm program <uncomment and complete this line to activate>
1393 #auth_param ntlm children 5
1394 #auth_param ntlm max_challenge_reuses 0
1395 #auth_param ntlm max_challenge_lifetime 2 minutes
1396 #auth_param basic program <uncomment and complete this line>
1397 auth_param basic children 5
1398 auth_param basic realm Squid proxy-caching web server
1399 auth_param basic credentialsttl 2 hours
1403 NAME: authenticate_cache_garbage_interval
1406 LOC: Config.authenticateGCInterval
1408 The time period between garbage collection across the username cache.
1409 This is a tradeoff between memory utilisation (long intervals - say
1410 2 days) and CPU (short intervals - say 1 minute). Only change if
1411 you have good reason to.
1414 NAME: authenticate_ttl
1417 LOC: Config.authenticateTTL
1419 The time a user & their credentials stay in the logged in user cache
1420 since their last request. When the garbage interval passes, all
1421 user credentials that have passed their TTL are removed from memory.
1424 NAME: authenticate_ip_ttl
1426 LOC: Config.authenticateIpTTL
1429 With this option you control how long a proxy authentication
1430 will be bound to a specific IP address. If a request using
1431 the same user name is received during this time then access
1432 will be denied and both users are required to reauthenticate
1433 them selves. The idea behind this is to make it annoying
1434 for people to share their password to their friends, but
1435 yet allow a dialup user to reconnect on a different dialup
1438 The default is 0 to disable the check. Recommended value
1439 if you have dialup users are no more than 60 seconds to allow
1440 the user to redial without hassle. If all your users are
1441 stationary then higher values may be used.
1443 See also the acl max_user_ip. The max_user_ip acl replaces
1444 the authenticate_ip_ttl_is_strict option found in earlier
1448 NAME: external_acl_type
1449 TYPE: externalAclHelper
1450 LOC: Config.externalAclHelperList
1453 This option defines external acl classes using a helper program
1454 to look up the status
1456 external_acl_type name [options] FORMAT.. /path/to/helper [helper arguments..]
1460 ttl=n TTL in seconds for cached results (defaults to 3600
1463 TTL for cached negative lookups (default same
1465 concurrenty=n Concurrency level / number of processes spawn
1466 to service external acl lookups of this type.
1467 cache=n result cache size, 0 is unbounded (default)
1469 The helper receives lines per the FORMAT specification, and returns
1470 OK or ERR indicating the validity of the request.
1472 FORMAT specifications
1474 %LOGIN Authenticated user login name
1475 %IDENT Ident user name
1478 %PROTO Requested protocol
1479 %PORT Requested port
1480 %METHOD Request method
1481 %{Header} HTTP request header
1482 %{Hdr:member} HTTP request header list member
1484 HTTP request header list member using ; as
1485 list separator. ; can be any non-alphanumeric
1488 In addition, any string specified in the referencing acl will also be
1489 included in the helper request line, after the specified formats
1490 (see the acl directive)
1495 OPTIONS FOR TUNING THE CACHE
1496 -----------------------------------------------------------------------------
1499 NAME: wais_relay_host
1502 LOC: Config.Wais.relayHost
1505 NAME: wais_relay_port
1508 LOC: Config.Wais.relayPort
1510 Relay WAIS request to host (1st arg) at port (2 arg).
1514 NAME: request_header_max_size
1518 LOC: Config.maxRequestHeaderSize
1520 This specifies the maximum size for HTTP headers in a request.
1521 Request headers are usually relatively small (about 512 bytes).
1522 Placing a limit on the request header size will catch certain
1523 bugs (for example with persistent connections) and possibly
1524 buffer-overflow or denial-of-service attacks.
1527 NAME: request_body_max_size
1531 LOC: Config.maxRequestBodySize
1533 This specifies the maximum size for an HTTP request body.
1534 In other words, the maximum size of a PUT/POST request.
1535 A user who attempts to send a request with a body larger
1536 than this limit receives an "Invalid Request" error message.
1537 If you set this parameter to a zero (the default), there will
1538 be no limit imposed.
1541 NAME: refresh_pattern
1542 TYPE: refreshpattern
1546 usage: refresh_pattern [-i] regex min percent max [options]
1548 By default, regular expressions are CASE-SENSITIVE. To make
1549 them case-insensitive, use the -i option.
1551 'Min' is the time (in minutes) an object without an explicit
1552 expiry time should be considered fresh. The recommended
1553 value is 0, any higher values may cause dynamic applications
1554 to be erroneously cached unless the application designer
1555 has taken the appropriate actions.
1557 'Percent' is a percentage of the objects age (time since last
1558 modification age) an object without explicit expiry time
1559 will be considered fresh.
1561 'Max' is an upper limit on how long objects without an explicit
1562 expiry time will be considered fresh.
1564 options: override-expire
1569 override-expire enforces min age even if the server
1570 sent a Expires: header. Doing this VIOLATES the HTTP
1571 standard. Enabling this feature could make you liable
1572 for problems which it causes.
1574 override-lastmod enforces min age even on objects
1575 that was modified recently.
1577 reload-into-ims changes client no-cache or ``reload''
1578 to If-Modified-Since requests. Doing this VIOLATES the
1579 HTTP standard. Enabling this feature could make you
1580 liable for problems which it causes.
1582 ignore-reload ignores a client no-cache or ``reload''
1583 header. Doing this VIOLATES the HTTP standard. Enabling
1584 this feature could make you liable for problems which
1587 Please see the file doc/Release-Notes-1.1.txt for a full
1588 description of Squid's refresh algorithm. Basically a
1589 cached object is: (the order is changed from 1.1.X)
1591 FRESH if expires < now, else STALE
1593 FRESH if lm-factor < percent, else STALE
1597 The refresh_pattern lines are checked in the order listed here.
1598 The first entry which matches is used. If none of the entries
1599 match, then the default will be used.
1601 Note, you must uncomment all the default lines if you want
1602 to change one. The default setting is only active if none is
1607 refresh_pattern ^ftp: 1440 20% 10080
1608 refresh_pattern ^gopher: 1440 0% 1440
1609 refresh_pattern . 0 20% 4320
1615 LOC: Config.referenceAge
1618 As a part of normal operation, Squid performs Least Recently
1619 Used removal of cached objects. The LRU age for removal is
1620 computed dynamically, based on the amount of disk space in
1621 use. The dynamic value can be seen in the Cache Manager 'info'
1624 The 'reference_age' parameter defines the maximum LRU age. For
1625 example, setting reference_age to '1 week' will cause objects
1626 to be removed if they have not been accessed for a week or
1627 more. The default value is one year.
1629 Specify a number here, followed by units of time. For example:
1635 NOTE: this parameter is not used when using the enhanced
1636 replacement policies, GDSH or LFUDA.
1640 NAME: quick_abort_min
1644 LOC: Config.quickAbort.min
1647 NAME: quick_abort_max
1651 LOC: Config.quickAbort.max
1654 NAME: quick_abort_pct
1658 LOC: Config.quickAbort.pct
1660 The cache by default continues downloading aborted requests
1661 which are almost completed (less than 16 KB remaining). This
1662 may be undesirable on slow (e.g. SLIP) links and/or very busy
1663 caches. Impatient users may tie up file descriptors and
1664 bandwidth by repeatedly requesting and immediately aborting
1667 When the user aborts a request, Squid will check the
1668 quick_abort values to the amount of data transfered until
1671 If the transfer has less than 'quick_abort_min' KB remaining,
1672 it will finish the retrieval.
1674 If the transfer has more than 'quick_abort_max' KB remaining,
1675 it will abort the retrieval.
1677 If more than 'quick_abort_pct' of the transfer has completed,
1678 it will finish the retrieval.
1680 If you do not want any retrieval to continue after the client
1681 has aborted, set both 'quick_abort_min' and 'quick_abort_max'
1684 If you want retrievals to always continue if they are being
1685 cached then set 'quick_abort_min' to '-1 KB'.
1688 NAME: read_ahead_gap
1689 COMMENT: buffer-size
1691 LOC: Config.readAheadGap
1694 The amount of data the cache will buffer ahead of what has been
1695 sent to the client when retrieving an object from another server.
1701 LOC: Config.negativeTtl
1704 Time-to-Live (TTL) for failed requests. Certain types of
1705 failures (such as "connection refused" and "404 Not Found") are
1706 negatively-cached for a configurable amount of time. The
1707 default is 5 minutes. Note that this is different from
1708 negative caching of DNS lookups.
1712 NAME: positive_dns_ttl
1715 LOC: Config.positiveDnsTtl
1718 Time-to-Live (TTL) for positive caching of successful DNS lookups.
1719 Default is 6 hours (360 minutes). If you want to minimize the
1720 use of Squid's ipcache, set this to 1, not 0.
1724 NAME: negative_dns_ttl
1727 LOC: Config.negativeDnsTtl
1730 Time-to-Live (TTL) for negative caching of failed DNS lookups.
1733 NAME: range_offset_limit
1736 LOC: Config.rangeOffsetLimit
1739 Sets a upper limit on how far into the the file a Range request
1740 may be to cause Squid to prefetch the whole file. If beyond this
1741 limit then Squid forwards the Range request as it is and the result
1744 This is to stop a far ahead range request (lets say start at 17MB)
1745 from making Squid fetch the whole object up to that point before
1746 sending anything to the client.
1748 A value of -1 causes Squid to always fetch the object from the
1749 beginning so that it may cache the result. (2.0 style)
1751 A value of 0 causes Squid to never fetch more than the
1752 client requested. (default)
1758 -----------------------------------------------------------------------------
1761 NAME: connect_timeout
1764 LOC: Config.Timeout.connect
1767 Some systems (notably Linux) can not be relied upon to properly
1768 time out connect(2) requests. Therefore the Squid process
1769 enforces its own timeout on server connections. This parameter
1770 specifies how long to wait for the connect to complete. The
1771 default is two minutes (120 seconds).
1774 NAME: peer_connect_timeout
1777 LOC: Config.Timeout.peer_connect
1780 This parameter specifies how long to wait for a pending TCP
1781 connection to a peer cache. The default is 30 seconds. You
1782 may also set different timeout values for individual neighbors
1783 with the 'connect-timeout' option on a 'cache_peer' line.
1786 NAME: siteselect_timeout
1789 LOC: Config.Timeout.siteSelect
1792 For URN to multiple URL's URL selection
1798 LOC: Config.Timeout.read
1801 The read_timeout is applied on server-side connections. After
1802 each successful read(), the timeout will be extended by this
1803 amount. If no data is read again after this amount of time,
1804 the request is aborted and logged with ERR_READ_TIMEOUT. The
1805 default is 15 minutes.
1809 NAME: request_timeout
1811 LOC: Config.Timeout.request
1814 How long to wait for an HTTP request after initial
1815 connection establishment.
1819 NAME: persistent_request_timeout
1821 LOC: Config.Timeout.persistent_request
1824 How long to wait for the next HTTP request on a persistent
1825 connection after the previous request completes.
1829 NAME: client_lifetime
1832 LOC: Config.Timeout.lifetime
1835 The maximum amount of time that a client (browser) is allowed to
1836 remain connected to the cache process. This protects the Cache
1837 from having a lot of sockets (and hence file descriptors) tied up
1838 in a CLOSE_WAIT state from remote clients that go away without
1839 properly shutting down (either because of a network failure or
1840 because of a poor client implementation). The default is one
1843 NOTE: The default value is intended to be much larger than any
1844 client would ever need to be connected to your cache. You
1845 should probably change client_lifetime only as a last resort.
1846 If you seem to have many client connections tying up
1847 filedescriptors, we recommend first tuning the read_timeout,
1848 request_timeout, persistent_request_timeout and quick_abort values.
1851 NAME: half_closed_clients
1853 LOC: Config.onoff.half_closed_clients
1856 Some clients may shutdown the sending side of their TCP
1857 connections, while leaving their receiving sides open. Sometimes,
1858 Squid can not tell the difference between a half-closed and a
1859 fully-closed TCP connection. By default, half-closed client
1860 connections are kept open until a read(2) or write(2) on the
1861 socket returns an error. Change this option to 'off' and Squid
1862 will immediately close client connections when read(2) returns
1863 "no more data to read."
1868 LOC: Config.Timeout.pconn
1869 DEFAULT: 120 seconds
1871 Timeout for idle persistent connections to servers and other
1878 LOC: Config.Timeout.ident
1881 Maximum time to wait for IDENT lookups to complete.
1883 If this is too high, and you enabled IDENT lookups from untrusted
1884 users, then you might be susceptible to denial-of-service by having
1885 many ident requests going at once.
1889 NAME: shutdown_lifetime
1892 LOC: Config.shutdownLifetime
1895 When SIGTERM or SIGHUP is received, the cache is put into
1896 "shutdown pending" mode until all active sockets are closed.
1897 This value is the lifetime to set for all open descriptors
1898 during shutdown mode. Any active clients after this many
1899 seconds will receive a 'timeout' message.
1904 -----------------------------------------------------------------------------
1912 Defining an Access List
1914 acl aclname acltype string1 ...
1915 acl aclname acltype "file" ...
1917 when using "file", the file should contain one item per line
1919 acltype is one of the types described below
1921 By default, regular expressions are CASE-SENSITIVE. To make
1922 them case-insensitive, use the -i option.
1924 acl aclname src ip-address/netmask ... (clients IP address)
1925 acl aclname src addr1-addr2/netmask ... (range of addresses)
1926 acl aclname dst ip-address/netmask ... (URL host's IP address)
1927 acl aclname myip ip-address/netmask ... (local socket IP address)
1929 acl aclname srcdomain .foo.com ... # reverse lookup, client IP
1930 acl aclname dstdomain .foo.com ... # Destination server from URL
1931 acl aclname srcdom_regex [-i] xxx ... # regex matching client name
1932 acl aclname dstdom_regex [-i] xxx ... # regex matching server
1933 # For dstdomain and dstdom_regex a reverse lookup is tried if a IP
1934 # based URL is used. The name "none" is used if the reverse lookup
1937 acl aclname time [day-abbrevs] [h1:m1-h2:m2]
1946 h1:m1 must be less than h2:m2
1947 acl aclname url_regex [-i] ^http:// ... # regex matching on whole URL
1948 acl aclname urlpath_regex [-i] \.gif$ ... # regex matching on URL path
1949 acl aclname port 80 70 21 ...
1950 acl aclname port 0-1024 ... # ranges allowed
1951 acl aclname myport 3128 ... # (local socket TCP port)
1952 acl aclname proto HTTP FTP ...
1953 acl aclname method GET POST ...
1954 acl aclname browser [-i] regexp ...
1955 # pattern match on User-Agent header
1956 acl aclname referer_regex [-i] regexp ...
1957 # pattern match on Referer header
1958 # Referer is highly unreliable, so use with care
1959 acl aclname ident username ...
1960 acl aclname ident_regex [-i] pattern ...
1961 # string match on ident output.
1962 # use REQUIRED to accept any non-null ident.
1963 acl aclname src_as number ...
1964 acl aclname dst_as number ...
1965 # Except for access control, AS numbers can be used for
1966 # routing of requests to specific caches. Here's an
1967 # example for routing all requests for AS#1241 and only
1968 # those to mycache.mydomain.net:
1969 # acl asexample dst_as 1241
1970 # cache_peer_access mycache.mydomain.net allow asexample
1971 # cache_peer_access mycache_mydomain.net deny all
1973 acl aclname proxy_auth username ...
1974 acl aclname proxy_auth_regex [-i] pattern ...
1975 # list of valid usernames
1976 # use REQUIRED to accept any valid username.
1978 # NOTE: when a Proxy-Authentication header is sent but it is not
1979 # needed during ACL checking the username is NOT logged
1982 # NOTE: proxy_auth requires a EXTERNAL authentication program
1983 # to check username/password combinations (see
1984 # authenticate_program).
1986 # WARNING: proxy_auth can't be used in a transparent proxy. It
1987 # collides with any authentication done by origin servers. It may
1988 # seem like it works at first, but it doesn't.
1990 acl aclname snmp_community string ...
1991 # A community string to limit access to your SNMP Agent
1994 # acl snmppublic snmp_community public
1996 acl aclname maxconn number
1997 # This will be matched when the client's IP address has
1998 # more than <number> HTTP connections established.
2000 acl aclname max_user_ip [-s] number
2001 # This will be matched when the user attempts to log in from more
2002 # than <number> different ip address's. The authenticate_ip_ttl
2003 # parameter controls the timeout on the ip entries.
2004 # If -s is specified then the limit is strict, denying browsing
2005 # from any further IP addresses until the ttl has expired. Without
2006 # -s Squid will just annoy the user by "randomly" deny requests.
2007 # (the counter is then reset each time the limit is reached and a
2008 # request is denied)
2009 # NOTE: in acceleration mode or where there is mesh of child proxies,
2010 # clients may appear to come from multiple address's if they are
2011 # going through proxy farms, so a limit of 1 may cause user problems.
2013 acl aclname req_mime_type mime-type1 ...
2014 # regex match agains the mime type of the request generated
2015 # by the client. Can be used to detect file upload or some
2016 # types HTTP tunelling requests.
2017 # NOTE: This does NOT match the reply. You cannot use this
2018 # to match the returned file type.
2020 acl aclname rep_mime_type mime-type1 ...
2021 # regex match against the mime type of the reply recieved by
2022 # squid. Can be used to detect file download or some
2023 # types HTTP tunelling requests.
2024 # NOTE: This has no effect in http_access rules. It only has
2025 # effect in rules that affect the reply data stream such as
2026 # http_reply_access.
2028 acl acl_name external class_name [arguments...]
2029 # external ACL lookup via a helper class defined by the
2030 # external_acl_type directive.
2033 acl myexample dst_as 1241
2034 acl password proxy_auth REQUIRED
2035 acl fileupload req_mime_type -i ^multipart/form-data$
2036 acl javascript rep_mime_type -i ^application/x-javascript$
2039 #Recommended minimum configuration:
2040 acl all src 0.0.0.0/0.0.0.0
2041 acl manager proto cache_object
2042 acl localhost src 127.0.0.1/255.255.255.255
2043 acl to_localhost dst 127.0.0.0/8
2044 acl SSL_ports port 443 563
2045 acl Safe_ports port 80 # http
2046 acl Safe_ports port 21 # ftp
2047 acl Safe_ports port 443 563 # https, snews
2048 acl Safe_ports port 70 # gopher
2049 acl Safe_ports port 210 # wais
2050 acl Safe_ports port 1025-65535 # unregistered ports
2051 acl Safe_ports port 280 # http-mgmt
2052 acl Safe_ports port 488 # gss-http
2053 acl Safe_ports port 591 # filemaker
2054 acl Safe_ports port 777 # multiling http
2055 acl CONNECT method CONNECT
2061 LOC: Config.accessList.http
2063 DEFAULT_IF_NONE: deny all
2065 Allowing or Denying access based on defined access lists
2067 Access to the HTTP port:
2068 http_access allow|deny [!]aclname ...
2070 NOTE on default values:
2072 If there are no "access" lines present, the default is to deny
2075 If none of the "access" lines cause a match, the default is the
2076 opposite of the last line in the list. If the last line was
2077 deny, then the default is allow. Conversely, if the last line
2078 is allow, the default will be deny. For these reasons, it is a
2079 good idea to have an "deny all" or "allow all" entry at the end
2080 of your access lists to avoid potential confusion.
2083 #Recommended minimum configuration:
2085 # Only allow cachemgr access from localhost
2086 http_access allow manager localhost
2087 http_access deny manager
2088 # Deny requests to unknown ports
2089 http_access deny !Safe_ports
2090 # Deny CONNECT to other than SSL ports
2091 http_access deny CONNECT !SSL_ports
2093 # We strongly recommend to uncomment the following to protect innocent
2094 # web applications running on the proxy server who think that the only
2095 # one who can access services on "localhost" is a local user
2096 #http_access deny to_localhost
2098 # INSERT YOUR OWN RULE(S) HERE TO ALLOW ACCESS FROM YOUR CLIENTS
2100 # Exampe rule allowing access from your local networks. Adapt
2101 # to list your (internal) IP networks from where browsing should
2103 #acl our_networks src 192.168.1.0/24 192.168.2.0/24
2104 #http_access allow our_networks
2106 # And finally deny all other access to this proxy
2107 http_access deny all
2111 NAME: http_reply_access
2113 LOC: Config.accessList.reply
2115 DEFAULT_IF_NONE: allow all
2117 Allow replies to client requests. This is complementary to http_access.
2119 http_reply_access allow|deny [!] aclname ...
2121 NOTE: if there are no access lines present, the default is to allow
2124 If none of the access lines cause a match, then the opposite of the
2125 last line will apply. Thus it is good practice to end the rules
2126 with an "allow all" or "deny all" entry.
2129 #Recommended minimum configuration:
2131 # Insert your own rules here.
2134 # and finally allow by default
2135 http_reply_access allow all
2142 LOC: Config.accessList.icp
2144 DEFAULT_IF_NONE: deny all
2146 Allowing or Denying access to the ICP port based on defined
2149 icp_access allow|deny [!]aclname ...
2151 See http_access for details
2154 #Allow ICP queries from eveyone
2155 icp_access allow all
2162 LOC: Config.accessList.miss
2165 Use to force your neighbors to use you as a sibling instead of
2166 a parent. For example:
2168 acl localclients src 172.16.0.0/16
2169 miss_access allow localclients
2170 miss_access deny !localclients
2172 This means that only your local clients are allowed to fetch
2173 MISSES and all other clients can only fetch HITS.
2175 By default, allow all clients who passed the http_access rules
2176 to fetch MISSES from us.
2180 # miss_access allow all
2185 NAME: cache_peer_access
2190 Similar to 'cache_peer_domain' but provides more flexibility by
2193 cache_peer_access cache-host allow|deny [!]aclname ...
2195 The syntax is identical to 'http_access' and the other lists of
2196 ACL elements. See the comments for 'http_access' below, or
2197 the Squid FAQ (http://www.squid-cache.org/FAQ/FAQ-10.html).
2200 NAME: ident_lookup_access
2204 DEFAULT_IF_NONE: deny all
2205 LOC: Config.accessList.identLookup
2207 A list of ACL elements which, if matched, cause an ident
2208 (RFC 931) lookup to be performed for this request. For
2209 example, you might choose to always perform ident lookups
2210 for your main multi-user Unix boxes, but not for your Macs
2211 and PCs. By default, ident lookups are not performed for
2214 To enable ident lookups for specific client addresses, you
2215 can follow this example:
2217 acl ident_aware_hosts src 198.168.1.0/255.255.255.0
2218 ident_lookup_access allow ident_aware_hosts
2219 ident_lookup_access deny all
2221 Only src type ACL checks are fully supported. A src_domain
2222 ACL might work at times, but it will not always provide
2226 NAME: tcp_outgoing_tos tcp_outgoing_ds tcp_outgoing_dscp
2229 LOC: Config.accessList.outgoing_tos
2231 Allows you to select a TOS/Diffserv value to mark outgoing
2232 connections with, based on the username or source address
2235 tcp_outgoing_tos ds-field [!]aclname ...
2237 Example where normal_service_net uses the TOS value 0x00
2238 and normal_service_net uses 0x20
2240 acl normal_service_net src 10.0.0.0/255.255.255.0
2241 acl good_service_net src 10.0.1.0/255.255.255.0
2242 tcp_outgoing_tos 0x00 normal_service_net 0x00
2243 tcp_outgoing_tos 0x20 good_service_net
2245 TOS/DSCP values really only have local significance - so you should
2246 know what you're specifying. For more, see RFC 2474
2248 The TOS/DSCP byte must be exactly that - a byte, value 0 - 255, or
2249 "default" to use whatever default your host has.
2251 Processing proceeds in the order specified, and stops at first fully
2255 NAME: tcp_outgoing_address
2258 LOC: Config.accessList.outgoing_address
2260 Allows you to map requests to different outgoing IP addresses
2261 based on the username or sourceaddress of the user making
2264 tcp_outgoing_address ipaddr [[!]aclname] ...
2266 Example where requests from 10.0.0.0/24 will be forwareded
2267 with source address 10.1.0.1, 10.0.2.0/24 forwarded with
2268 source address 10.1.0.2 and the rest will be forwarded with
2269 source address 10.1.0.3.
2271 acl normal_service_net src 10.0.0.0/255.255.255.0
2272 acl good_service_net src 10.0.1.0/255.255.255.0
2273 tcp_outgoing_address 10.0.0.1 normal_service_net
2274 tcp_outgoing_address 10.0.0.2 good_service_net
2275 tcp_outgoing_address 10.0.0.3
2277 Processing proceeds in the order specified, and stops at first fully
2281 NAME: reply_body_max_size
2282 COMMENT: bytes allow|deny acl acl...
2285 DEFAULT_IF_NONE: 0 allow all
2286 LOC: Config.ReplyBodySize
2288 This option specifies the maximum size of a reply body. It
2289 can be used to prevent users from downloading very large files,
2290 such as MP3's and movies. When the reply headers are recieved,
2291 the reply_body_max_size lines are processed, and the first line with
2292 a result of "allow" is used as the maximum body size for this reply.
2293 This size is then checked twice. First when we get the reply headers,
2294 we check the content-length value. If the content length value exists
2295 and is larger than the allowed size, the request is denied and the
2296 user receives an error message that says "the request or reply
2297 is too large." If there is no content-length, and the reply
2298 size exceeds this limit, the client's connection is just closed
2299 and they will receive a partial reply.
2301 WARNING: downstream caches probably can not detect a partial reply
2302 if there is no content-length header, so they will cache
2303 partial responses and give them out as hits. You should NOT
2304 use this option if you have downstream caches.
2306 WARNING: A maximum size larger than the size of squid's error messages
2307 will cause an infinite loop and crash squid. Ensure that the smallest
2308 non-zero value you use is greater that the maximum header size plus
2309 the size of your largest error page.
2311 If you set this parameter to zero (the default), there will be
2316 ADMINISTRATIVE PARAMETERS
2317 -----------------------------------------------------------------------------
2323 LOC: Config.adminEmail
2325 Email-address of local cache manager who will receive
2326 mail if the cache dies. The default is "webmaster."
2330 NAME: cache_effective_user
2333 LOC: Config.effectiveUser
2336 NAME: cache_effective_group
2339 LOC: Config.effectiveGroup
2342 If the cache is run as root, it will change its effective/real
2343 UID/GID to the UID/GID specified below. The default is to
2344 change to UID to nobody and GID to the default group of nobody.
2346 If Squid is not started as root, the default is to keep the
2347 current UID/GID, and only the GID can be changed to any of
2348 the groups the user starting Squid is member of. Note that if
2349 Squid is not started as root then you cannot set http_port to
2350 a value lower than 1024.
2354 NAME: visible_hostname
2356 LOC: Config.visibleHostname
2359 If you want to present a special hostname in error messages, etc,
2360 then define this. Otherwise, the return value of gethostname()
2361 will be used. If you have multiple caches in a cluster and
2362 get errors about IP-forwarding you must set them to have individual
2363 names with this setting.
2367 NAME: unique_hostname
2369 LOC: Config.uniqueHostname
2372 If you want to have multiple machines with the same
2373 'visible_hostname' then you must give each machine a different
2374 'unique_hostname' so that forwarding loops can be detected.
2378 NAME: hostname_aliases
2380 LOC: Config.hostnameAliases
2383 A list of other DNS names that your cache has.
2387 OPTIONS FOR THE CACHE REGISTRATION SERVICE
2388 -----------------------------------------------------------------------------
2390 This section contains parameters for the (optional) cache
2391 announcement service. This service is provided to help
2392 cache administrators locate one another in order to join or
2393 create cache hierarchies.
2395 An 'announcement' message is sent (via UDP) to the registration
2396 service by Squid. By default, the announcement message is NOT
2397 SENT unless you enable it with 'announce_period' below.
2399 The announcement message includes your hostname, plus the
2400 following information from this configuration file:
2406 All current information is processed regularly and made
2407 available on the Web at http://www.ircache.net/Cache/Tracker/.
2410 NAME: announce_period
2412 LOC: Config.Announce.period
2415 This is how frequently to send cache announcements. The
2416 default is `0' which disables sending the announcement
2419 To enable announcing your cache, just uncomment the line
2423 #To enable announcing your cache, just uncomment the line below.
2424 #announce_period 1 day
2431 DEFAULT: tracker.ircache.net
2432 LOC: Config.Announce.host
2438 LOC: Config.Announce.file
2444 LOC: Config.Announce.port
2446 announce_host and announce_port set the hostname and port
2447 number where the registration message will be sent.
2449 Hostname will default to 'tracker.ircache.net' and port will
2450 default default to 3131. If the 'filename' argument is given,
2451 the contents of that file will be included in the announce
2456 HTTPD-ACCELERATOR OPTIONS
2457 -----------------------------------------------------------------------------
2460 NAME: httpd_accel_host
2462 LOC: Config.Accel.host
2466 NAME: httpd_accel_port
2468 LOC: Config.Accel.port
2471 If you want to run Squid as an httpd accelerator, define the
2472 host name and port number where the real HTTP server is.
2474 If you want virtual host support then specify the hostname
2477 If you want virtual port support then specify the port as "0".
2479 NOTE: enabling httpd_accel_host disables proxy-caching and
2480 ICP. If you want these features enabled also, then set
2481 the 'httpd_accel_with_proxy' option.
2484 NAME: httpd_accel_single_host
2487 LOC: Config.Accel.single_host
2490 If you are running Squid as an accelerator and have a single backend
2491 server then set this to on. This causes Squid to forward the request
2492 to this server irregardles of what any redirectors or Host headers
2495 Leave this at off if you have multiple backend servers, and use a
2496 redirector (or host table or private DNS) to map the requests to the
2497 appropriate backend servers. Note that the mapping needs to be a
2498 1-1 mapping between requested and backend (from redirector) domain
2499 names or caching will fail, as cacing is performed using the
2500 URL returned from the redirector.
2502 See also redirect_rewrites_host_header.
2505 NAME: httpd_accel_with_proxy
2509 LOC: Config.onoff.accel_with_proxy
2511 If you want to use Squid as both a local httpd accelerator
2512 and as a proxy, change this to 'on'. Note however that your
2513 proxy users may have trouble to reach the accelerated domains
2514 unless their browsers are configured not to use this proxy for
2515 those domains (for example via the no_proxy browser configuration
2519 NAME: httpd_accel_uses_host_header
2523 LOC: opt_accel_uses_host
2525 HTTP/1.1 requests include a Host: header which is basically the
2526 hostname from the URL. Squid can be an accelerator for
2527 different HTTP servers by looking at this header. However,
2528 Squid does NOT check the value of the Host header, so it opens
2529 a big security hole. We recommend that this option remain
2530 disabled unless you are sure of what you are doing.
2532 However, you will need to enable this option if you run Squid
2533 as a transparent proxy. Otherwise, virtual servers which
2534 require the Host: header will not be properly cached.
2539 -----------------------------------------------------------------------------
2544 LOC: Config.dns_testname_list
2546 DEFAULT_IF_NONE: netscape.com internic.net nlanr.net microsoft.com
2548 The DNS tests exit as soon as the first site is successfully looked up
2550 This test can be disabled with the -D command line option.
2554 NAME: logfile_rotate
2557 LOC: Config.Log.rotateNumber
2559 Specifies the number of logfile rotations to make when you
2560 type 'squid -k rotate'. The default is 10, which will rotate
2561 with extensions 0 through 9. Setting logfile_rotate to 0 will
2562 disable the rotation, but the logfiles are still closed and
2563 re-opened. This will enable you to rename the logfiles
2564 yourself just before sending the rotate signal.
2566 Note, the 'squid -k rotate' command normally sends a USR1
2567 signal to the running squid process. In certain situations
2568 (e.g. on Linux with Async I/O), USR1 is used for other
2569 purposes, so -k rotate uses another signal. It is best to get
2570 in the habit of using 'squid -k rotate' instead of 'kill -USR1
2577 LOC: Config.appendDomain
2580 Appends local domain name to hostnames without any dots in
2581 them. append_domain must begin with a period.
2583 Be warned that there today is Internet names with no dots in
2584 them using only top-domain names, so setting this may
2585 cause some Internet sites to become unavailable.
2588 append_domain .yourdomain.com
2592 NAME: tcp_recv_bufsize
2596 LOC: Config.tcpRcvBufsz
2598 Size of receive buffer to set for TCP sockets. Probably just
2599 as easy to change your kernel's default. Set to zero to use
2600 the default buffer size.
2605 LOC: Config.errHtmlText
2608 HTML text to include in error messages. Make this a "mailto"
2609 URL to your admin address, or maybe just a link to your
2610 organizations Web page.
2612 To include this in your error messages, you must rewrite
2613 the error template files (found in the "errors" directory).
2614 Wherever you want the 'err_html_text' line to appear,
2615 insert a %L tag in the error template file.
2621 LOC: Config.denyInfoList
2624 Usage: deny_info err_page_name acl
2625 or deny_info http://... acl
2626 Example: deny_info ERR_CUSTOM_ACCESS_DENIED bad_guys
2628 This can be used to return a ERR_ page for requests which
2629 do not pass the 'http_access' rules. A single ACL will cause
2630 the http_access check to fail. If a 'deny_info' line exists
2631 for that ACL then Squid returns a corresponding error page.
2633 You may use ERR_ pages that come with Squid or create your own pages
2634 and put them into the configured errors/ directory.
2636 Alternatively you can specify an error URL. The browsers will then
2637 get redirected (302) to the specified URL. %s in the redirection
2638 URL will be replaced by the requested URL.
2640 Alternatively you can tell Squid to reset the TCP connection
2641 by specifying TCP_RESET.
2648 LOC: Config.onoff.mem_pools
2650 If set, Squid will keep pools of allocated (but unused) memory
2651 available for future use. If memory is a premium on your
2652 system and you believe your malloc library outperforms Squid
2653 routines, disable this.
2656 NAME: memory_pools_limit
2660 LOC: Config.MemPools.limit
2662 Used only with memory_pools on:
2663 memory_pools_limit 50 MB
2665 If set to a non-zero value, Squid will keep at most the specified
2666 limit of allocated (but unused) memory in memory pools. All free()
2667 requests that exceed this limit will be handled by your malloc
2668 library. Squid does not pre-allocate any memory, just safe-keeps
2669 objects that otherwise would be free()d. Thus, it is safe to set
2670 memory_pools_limit to a reasonably high value even if your
2671 configuration will use less memory.
2673 If not set (default) or set to zero, Squid will keep all memory it
2674 can. That is, there will be no limit on the total amount of memory
2675 used for safe-keeping.
2677 To disable memory allocation optimization, do not set
2678 memory_pools_limit to 0. Set memory_pools to "off" instead.
2680 An overhead for maintaining memory pools is not taken into account
2681 when the limit is checked. This overhead is close to four bytes per
2682 object kept. However, pools may actually _save_ memory because of
2683 reduced memory thrashing in your malloc library.
2690 LOC: opt_forwarded_for
2692 If set, Squid will include your system's IP address or name
2693 in the HTTP requests it forwards. By default it looks like
2696 X-Forwarded-For: 192.1.2.3
2698 If you disable this, it will appear as
2700 X-Forwarded-For: unknown
2703 NAME: log_icp_queries
2707 LOC: Config.onoff.log_udp
2709 If set, ICP queries are logged to access.log. You may wish
2710 do disable this if your ICP load is VERY high to speed things
2711 up or to simplify log analysis.
2718 LOC: Config.onoff.icp_hit_stale
2720 If you want to return ICP_HIT for stale cache objects, set this
2721 option to 'on'. If you have sibling relationships with caches
2722 in other administrative domains, this should be 'off'. If you only
2723 have sibling relationships with caches under your control, then
2724 it is probably okay to set this to 'on'.
2725 If set to 'on', then your siblings should use the option "allow-miss"
2726 on their cache_peer lines for connecting to you.
2730 NAME: minimum_direct_hops
2733 LOC: Config.minDirectHops
2735 If using the ICMP pinging stuff, do direct fetches for sites
2736 which are no more than this many hops away.
2739 NAME: minimum_direct_rtt
2742 LOC: Config.minDirectRtt
2744 If using the ICMP pinging stuff, do direct fetches for sites
2745 which are no more than this many rtt milliseconds away.
2748 NAME: cachemgr_passwd
2749 TYPE: cachemgrpasswd
2751 LOC: Config.passwd_list
2753 Specify passwords for cachemgr operations.
2755 Usage: cachemgr_passwd password action action ...
2757 Some valid actions are (see cache manager menu for a full list):
2795 * Indicates actions which will not be performed without a
2796 valid password, others can be performed if not listed here.
2798 To disable an action, set the password to "disable".
2799 To allow performing an action without a password, set the
2802 Use the keyword "all" to set the same password for all actions.
2805 cachemgr_passwd secret shutdown
2806 cachemgr_passwd lesssssssecret info stats/objects
2807 cachemgr_passwd disable all
2810 NAME: store_avg_object_size
2814 LOC: Config.Store.avgObjectSize
2816 Average object size, used to estimate number of objects your
2817 cache can hold. See doc/Release-Notes-1.1.txt. The default is
2821 NAME: store_objects_per_bucket
2824 LOC: Config.Store.objectsPerBucket
2826 Target number of objects per bucket in the store hash table.
2827 Lowering this value increases the total number of buckets and
2828 also the storage maintenance rate. The default is 50.
2835 LOC: Config.onoff.client_db
2837 If you want to disable collecting per-client statistics, then
2838 turn off client_db here.
2845 LOC: Config.Netdb.low
2851 LOC: Config.Netdb.high
2853 The low and high water marks for the ICMP measurement
2854 database. These are counts, not percents. The defaults are
2855 900 and 1000. When the high water mark is reached, database
2856 entries will be deleted until the low mark is reached.
2860 NAME: netdb_ping_period
2862 LOC: Config.Netdb.period
2865 The minimum period for measuring a site. There will be at
2866 least this much delay between successive pings to the same
2867 network. The default is five minutes.
2875 LOC: Config.onoff.query_icmp
2877 If you want to ask your peers to include ICMP data in their ICP
2878 replies, enable this option.
2880 If your peer has configured Squid (during compilation) with
2881 '--enable-icmp' then that peer will send ICMP pings to origin server
2882 sites of the URLs it receives. If you enable this option then the
2883 ICP replies from that peer will include the ICMP data (if available).
2884 Then, when choosing a parent cache, Squid will choose the parent with
2885 the minimal RTT to the origin server. When this happens, the
2886 hierarchy field of the access.log will be
2887 "CLOSEST_PARENT_MISS". This option is off by default.
2890 NAME: test_reachability
2894 LOC: Config.onoff.test_reachability
2896 When this is 'on', ICP MISS replies will be ICP_MISS_NOFETCH
2897 instead of ICP_MISS if the target host is NOT in the ICMP
2898 database, or has a zero RTT.
2905 LOC: Config.onoff.buffered_logs
2907 cache.log log file is written with stdio functions, and as such
2908 it can be buffered or unbuffered. By default it will be unbuffered.
2909 Buffering it can speed up the writing slightly (though you are
2910 unlikely to need to worry unless you run with tons of debugging
2911 enabled in which case performance will suffer badly anyway..).
2914 NAME: reload_into_ims
2915 IFDEF: HTTP_VIOLATIONS
2919 LOC: Config.onoff.reload_into_ims
2921 When you enable this option, client no-cache or ``reload''
2922 requests will be changed to If-Modified-Since requests.
2923 Doing this VIOLATES the HTTP standard. Enabling this
2924 feature could make you liable for problems which it
2927 see also refresh_pattern for a more selective approach.
2932 LOC: Config.accessList.AlwaysDirect
2935 Usage: always_direct allow|deny [!]aclname ...
2937 Here you can use ACL elements to specify requests which should
2938 ALWAYS be forwarded directly to origin servers. For example,
2939 to always directly forward requests for local servers use
2942 acl local-servers dstdomain my.domain.net
2943 always_direct allow local-servers
2945 To always forward FTP requests directly, use
2948 always_direct allow FTP
2950 NOTE: There is a similar, but opposite option named
2951 'never_direct'. You need to be aware that "always_direct deny
2952 foo" is NOT the same thing as "never_direct allow foo". You
2953 may need to use a deny rule to exclude a more-specific case of
2954 some other rule. Example:
2956 acl local-external dstdomain external.foo.net
2957 acl local-servers dstdomain .foo.net
2958 always_direct deny local-external
2959 always_direct allow local-servers
2961 This option replaces some v1.1 options such as local_domain
2967 LOC: Config.accessList.NeverDirect
2970 Usage: never_direct allow|deny [!]aclname ...
2972 never_direct is the opposite of always_direct. Please read
2973 the description for always_direct if you have not already.
2975 With 'never_direct' you can use ACL elements to specify
2976 requests which should NEVER be forwarded directly to origin
2977 servers. For example, to force the use of a proxy for all
2978 requests, except those in your local domain use something like:
2980 acl local-servers dstdomain .foo.net
2981 acl all src 0.0.0.0/0.0.0.0
2982 never_direct deny local-servers
2983 never_direct allow all
2985 or if squid is inside a firewall and there is local intranet
2986 servers inside the firewall then use something like:
2988 acl local-intranet dstdomain .foo.net
2989 acl local-external dstdomain external.foo.net
2990 always_direct deny local-external
2991 always_direct allow local-intranet
2992 never_direct allow all
2994 This option replaces some v1.1 options such as inside_firewall
2999 TYPE: http_header_access[]
3000 LOC: Config.header_access
3003 Usage: header_access header_name allow|deny [!]aclname ...
3005 This option replaces the old 'anonymize_headers' and the
3006 older 'http_anonymizer' option with something that is much
3007 more configurable. This new method creates a list of ACLs
3008 for each header, allowing you very fine-tuned header
3011 You can only specify known headers for the header name.
3012 Other headers are reclassified as 'Other'. You can also
3013 refer to all the headers with 'All'.
3015 For example, to achieve the same behaviour as the old
3016 'http_anonymizer standard' option, you should use:
3018 header_access From deny all
3019 header_access Referer deny all
3020 header_access Server deny all
3021 header_access User-Agent deny all
3022 header_access WWW-Authenticate deny all
3023 header_access Link deny all
3025 Or, to reproduce the old 'http_anonymizer paranoid' feature
3028 header_access Allow allow all
3029 header_access Authorization allow all
3030 header_access Cache-Control allow all
3031 header_access Content-Encoding allow all
3032 header_access Content-Length allow all
3033 header_access Content-Type allow all
3034 header_access Date allow all
3035 header_access Expires allow all
3036 header_access Host allow all
3037 header_access If-Modified-Since allow all
3038 header_access Last-Modified allow all
3039 header_access Location allow all
3040 header_access Pragma allow all
3041 header_access Accept allow all
3042 header_access Charset allow all
3043 header_access Accept-Encoding allow all
3044 header_access Accept-Language allow all
3045 header_access Content-Language allow all
3046 header_access Mime-Version allow all
3047 header_access Retry-After allow all
3048 header_access Title allow all
3049 header_access Connection allow all
3050 header_access Proxy-Connection allow all
3051 header_access All deny all
3053 By default, all headers are allowed (no anonymizing is
3057 NAME: header_replace
3058 TYPE: http_header_replace[]
3059 LOC: Config.header_access
3062 Usage: header_replace header_name message
3063 Example: header_replace User-Agent Nutscrape/1.0 (CP/M; 8-bit)
3065 This option allows you to change the contents of headers
3066 denied with header_access above, by replacing them with
3067 some fixed string. This replaces the old fake_user_agent
3070 By default, headers are removed if denied.
3073 NAME: icon_directory
3075 LOC: Config.icons.directory
3076 DEFAULT: @DEFAULT_ICON_DIR@
3078 Where the icons are stored. These are normally kept in
3082 NAME: error_directory
3084 LOC: Config.errorDirectory
3085 DEFAULT: @DEFAULT_ERROR_DIR@
3087 If you wish to create your own versions of the default
3088 (English) error files, either to customize them to suit your
3089 language or company copy the template English files to another
3090 directory and point this tag at them.
3093 NAME: minimum_retry_timeout
3096 LOC: Config.retry.timeout
3099 This specifies the minimum connect timeout, for when the
3100 connect timeout is reduced to compensate for the availability
3101 of multiple IP addresses.
3103 When a connection to a host is initiated, and that host has
3104 several IP addresses, the default connection timeout is reduced
3105 by dividing it by the number of addresses. So, a site with 15
3106 addresses would then have a timeout of 8 seconds for each
3107 address attempted. To avoid having the timeout reduced to the
3108 point where even a working host would not have a chance to
3109 respond, this setting is provided. The default, and the
3110 minimum value, is five seconds, and the maximum value is sixty
3111 seconds, or half of connect_timeout, whichever is greater and
3112 less than connect_timeout.
3115 NAME: maximum_single_addr_tries
3117 LOC: Config.retry.maxtries
3120 This sets the maximum number of connection attempts for a
3121 host that only has one address (for multiple-address hosts,
3122 each address is tried once).
3124 The default value is three tries, the (not recommended)
3125 maximum is 255 tries. A warning message will be generated
3126 if it is set to a value greater than ten.
3131 LOC: Config.Port.snmp
3135 Squid can now serve statistics and status information via SNMP.
3136 By default it listens to port 3401 on the machine. If you don't
3137 wish to use SNMP, set this to "0".
3142 LOC: Config.accessList.snmp
3144 DEFAULT_IF_NONE: deny all
3147 Allowing or denying access to the SNMP port.
3149 All access to the agent is denied by default.
3152 snmp_access allow|deny [!]aclname ...
3155 snmp_access allow snmppublic localhost
3156 snmp_access deny all
3159 NAME: snmp_incoming_address
3161 LOC: Config.Addrs.snmp_incoming
3165 NAME: snmp_outgoing_address
3167 LOC: Config.Addrs.snmp_outgoing
3168 DEFAULT: 255.255.255.255
3171 Just like 'udp_incoming_address' above, but for the SNMP port.
3173 snmp_incoming_address is used for the SNMP socket receiving
3174 messages from SNMP agents.
3175 snmp_outgoing_address is used for SNMP packets returned to SNMP
3178 The default snmp_incoming_address (0.0.0.0) is to listen on all
3179 available network interfaces.
3181 If snmp_outgoing_address is set to 255.255.255.255 (the default)
3182 then it will use the same socket as snmp_incoming_address. Only
3183 change this if you want to have SNMP replies sent using another
3184 address than where this Squid listens for SNMP queries.
3186 NOTE, snmp_incoming_address and snmp_outgoing_address can not have
3187 the same value since they both use port 3401.
3190 NAME: as_whois_server
3192 LOC: Config.as_whois_server
3193 DEFAULT: whois.ra.net
3194 DEFAULT_IF_NONE: whois.ra.net
3196 WHOIS server to query for AS numbers. NOTE: AS numbers are
3197 queried only when Squid starts up, not for every request.
3202 LOC: Config.Wccp.router
3206 Use this option to define your WCCP ``home'' router for
3207 Squid. Setting the 'wccp_router' to 0.0.0.0 (the default)
3213 LOC: Config.Wccp.version
3217 According to some users, Cisco IOS 11.2 only supports WCCP
3218 version 3. If you're using that version of IOS, change
3222 NAME: wccp_incoming_address
3224 LOC: Config.Wccp.incoming
3228 NAME: wccp_outgoing_address
3230 LOC: Config.Wccp.outgoing
3231 DEFAULT: 255.255.255.255
3234 wccp_incoming_address Use this option if you require WCCP
3235 messages to be received on only one
3236 interface. Do NOT use this option if
3237 you're unsure how many interfaces you
3238 have, or if you know you have only one
3241 wccp_outgoing_address Use this option if you require WCCP
3242 messages to be sent out on only one
3243 interface. Do NOT use this option if
3244 you're unsure how many interfaces you
3245 have, or if you know you have only one
3248 The default behavior is to not bind to any specific address.
3250 NOTE, wccp_incoming_address and wccp_outgoing_address can not have
3251 the same value since they both use port 2048.
3256 DELAY POOL PARAMETERS (all require DELAY_POOLS compilation option)
3257 -----------------------------------------------------------------------------
3261 TYPE: delay_pool_count
3266 This represents the number of delay pools to be used. For example,
3267 if you have one class 2 delay pool and one class 3 delays pool, you
3268 have a total of 2 delay pools.
3272 TYPE: delay_pool_class
3277 This defines the class of each delay pool. There must be exactly one
3278 delay_class line for each delay pool. For example, to define two
3279 delay pools, one of class 2 and one of class 3, the settings above
3283 delay_pools 2 # 2 delay pools
3284 delay_class 1 2 # pool 1 is a class 2 pool
3285 delay_class 2 3 # pool 2 is a class 3 pool
3287 The delay pool classes are:
3289 class 1 Everything is limited by a single aggregate
3292 class 2 Everything is limited by a single aggregate
3293 bucket as well as an "individual" bucket chosen
3294 from bits 25 through 32 of the IP address.
3296 class 3 Everything is limited by a single aggregate
3297 bucket as well as a "network" bucket chosen
3298 from bits 17 through 24 of the IP address and a
3299 "individual" bucket chosen from bits 17 through
3300 32 of the IP address.
3302 NOTE: If an IP address is a.b.c.d
3303 -> bits 25 through 32 are "d"
3304 -> bits 17 through 24 are "c"
3305 -> bits 17 through 32 are "c * 256 + d"
3309 TYPE: delay_pool_access
3314 This is used to determine which delay pool a request falls into.
3315 The first matched delay pool is always used, i.e., if a request falls
3316 into delay pool number one, no more delay are checked, otherwise the
3317 rest are checked in order of their delay pool number until they have
3318 all been checked. For example, if you want some_big_clients in delay
3319 pool 1 and lotsa_little_clients in delay pool 2:
3322 delay_access 1 allow some_big_clients
3323 delay_access 1 deny all
3324 delay_access 2 allow lotsa_little_clients
3325 delay_access 2 deny all
3328 NAME: delay_parameters
3329 TYPE: delay_pool_rates
3334 This defines the parameters for a delay pool. Each delay pool has
3335 a number of "buckets" associated with it, as explained in the
3336 description of delay_class. For a class 1 delay pool, the syntax is:
3338 delay_parameters pool aggregate
3340 For a class 2 delay pool:
3342 delay_parameters pool aggregate individual
3344 For a class 3 delay pool:
3346 delay_parameters pool aggregate network individual
3348 The variables here are:
3350 pool a pool number - ie, a number between 1 and the
3351 number specified in delay_pools as used in
3354 aggregate the "delay parameters" for the aggregate bucket
3357 individual the "delay parameters" for the individual
3358 buckets (class 2, 3).
3360 network the "delay parameters" for the network buckets
3363 A pair of delay parameters is written restore/maximum, where restore is
3364 the number of bytes (not bits - modem and network speeds are usually
3365 quoted in bits) per second placed into the bucket, and maximum is the
3366 maximum number of bytes which can be in the bucket at any time.
3368 For example, if delay pool number 1 is a class 2 delay pool as in the
3369 above example, and is being used to strictly limit each host to 64kbps
3370 (plus overheads), with no overall limit, the line is:
3372 delay_parameters 1 -1/-1 8000/8000
3374 Note that the figure -1 is used to represent "unlimited".
3376 And, if delay pool number 2 is a class 3 delay pool as in the above
3377 example, and you want to limit it to a total of 256kbps (strict limit)
3378 with each 8-bit network permitted 64kbps (strict limit) and each
3379 individual host permitted 4800bps with a bucket maximum size of 64kb
3380 to permit a decent web page to be downloaded at a decent speed
3381 (if the network is not being limited due to overuse) but slow down
3382 large downloads more significantly:
3384 delay_parameters 2 32000/32000 8000/8000 600/64000
3386 There must be one delay_parameters line for each delay pool.
3389 NAME: delay_initial_bucket_level
3390 COMMENT: (percent, 0-100)
3394 LOC: Config.Delay.initial
3396 The initial bucket percentage is used to determine how much is put
3397 in each bucket when squid starts, is reconfigured, or first notices
3398 a host accessing it (in class 2 and class 3, individual hosts and
3399 networks only have buckets associated with them once they have been
3403 NAME: incoming_icp_average
3406 LOC: Config.comm_incoming.icp_average
3409 NAME: incoming_http_average
3412 LOC: Config.comm_incoming.http_average
3415 NAME: incoming_dns_average
3418 LOC: Config.comm_incoming.dns_average
3421 NAME: min_icp_poll_cnt
3424 LOC: Config.comm_incoming.icp_min_poll
3427 NAME: min_dns_poll_cnt
3430 LOC: Config.comm_incoming.dns_min_poll
3433 NAME: min_http_poll_cnt
3436 LOC: Config.comm_incoming.http_min_poll
3438 Heavy voodoo here. I can't even believe you are reading this.
3439 Are you crazy? Don't even think about adjusting these unless
3440 you understand the algorithms in comm_select.c first!
3443 NAME: max_open_disk_fds
3445 LOC: Config.max_open_disk_fds
3448 To avoid having disk as the I/O bottleneck Squid can optionally
3449 bypass the on-disk cache if more than this amount of disk file
3450 descriptors are open.
3452 A value of 0 indicates no limit.
3457 LOC: Config.onoff.offline
3460 Enable this option and Squid will never try to validate cached
3464 NAME: uri_whitespace
3465 TYPE: uri_whitespace
3466 LOC: Config.uri_whitespace
3469 What to do with requests that have whitespace characters in the
3472 strip: The whitespace characters are stripped out of the URL.
3473 This is the behavior recommended by RFC2616.
3474 deny: The request is denied. The user receives an "Invalid
3476 allow: The request is allowed and the URI is not changed. The
3477 whitespace characters remain in the URI. Note the
3478 whitespace is passed to redirector processes if they
3480 encode: The request is allowed and the whitespace characters are
3481 encoded according to RFC1738. This could be considered
3482 a violation of the HTTP/1.1
3483 RFC because proxies are not allowed to rewrite URI's.
3484 chop: The request is allowed and the URI is chopped at the
3485 first whitespace. This might also be considered a
3492 LOC: Config.accessList.brokenPosts
3494 A list of ACL elements which, if matched, causes Squid to send
3495 an extra CRLF pair after the body of a PUT/POST request.
3497 Some HTTP servers has broken implementations of PUT/POST,
3498 and rely on an extra CRLF pair sent by some WWW clients.
3500 Quote from RFC 2068 section 4.1 on this matter:
3502 Note: certain buggy HTTP/1.0 client implementations generate an
3503 extra CRLF's after a POST request. To restate what is explicitly
3504 forbidden by the BNF, an HTTP/1.1 client must not preface or follow
3505 a request with an extra CRLF.
3508 acl buggy_server url_regex ^http://....
3509 broken_posts allow buggy_server
3512 NAME: mcast_miss_addr
3513 IFDEF: MULTICAST_MISS_STREAM
3515 LOC: Config.mcast_miss.addr
3516 DEFAULT: 255.255.255.255
3518 If you enable this option, every "cache miss" URL will
3519 be sent out on the specified multicast address.
3521 Do not enable this option unless you are are absolutely
3522 certain you understand what you are doing.
3525 NAME: mcast_miss_ttl
3526 IFDEF: MULTICAST_MISS_TTL
3528 LOC: Config.mcast_miss.ttl
3531 This is the time-to-live value for packets multicasted
3532 when multicasting off cache miss URLs is enabled. By
3533 default this is set to 'site scope', i.e. 16.
3536 NAME: mcast_miss_port
3537 IFDEF: MULTICAST_MISS_STREAM
3539 LOC: Config.mcast_miss.port
3542 This is the port number to be used in conjunction with
3546 NAME: mcast_miss_encode_key
3547 IFDEF: MULTICAST_MISS_STREAM
3549 LOC: Config.mcast_miss.encode_key
3550 DEFAULT: XXXXXXXXXXXXXXXX
3552 The URLs that are sent in the multicast miss stream are
3553 encrypted. This is the encryption key.
3556 NAME: nonhierarchical_direct
3558 LOC: Config.onoff.nonhierarchical_direct
3561 By default, Squid will send any non-hierarchical requests
3562 (matching hierarchy_stoplist or not cachable request type) direct
3565 If you set this to off, then Squid will prefer to send these
3566 requests to parents.
3568 Note that in most configurations, by turning this off you will only
3569 add latency to these request without any improvement in global hit
3572 If you are inside an firewall then see never_direct instead of
3578 LOC: Config.onoff.prefer_direct
3581 Normally Squid tries to use parents for most requests. If you by some
3582 reason like it to first try going direct and only use a parent if
3583 going direct fails then set this to off.
3585 By combining nonhierarchical_direct off and prefer_direct on you
3586 can set up Squid to use a parent as a backup path if going direct
3590 NAME: strip_query_terms
3592 LOC: Config.onoff.strip_query_terms
3595 By default, Squid strips query terms from requested URLs before
3596 logging. This protects your user's privacy.
3601 LOC: Config.coredump_dir
3603 DEFAULT_IF_NONE: none
3605 By default Squid leaves core files in the directory from where
3606 it was started. If you set 'coredump_dir' to a directory
3607 that exists, Squid will chdir() to that directory at startup
3608 and coredump files will be left there.
3611 # Leave coredumps in the first cache dir
3612 coredump_dir @DEFAULT_SWAP_DIR@
3616 NAME: redirector_bypass
3618 LOC: Config.onoff.redirector_bypass
3621 When this is 'on', a request will not go through the
3622 redirector if all redirectors are busy. If this is 'off'
3623 and the redirector queue grows too large, Squid will exit
3624 with a FATAL error and ask you to increase the number of
3625 redirectors. You should only enable this if the redirectors
3626 are not critical to your caching system. If you use
3627 redirectors for access control, and you enable this option,
3628 then users may have access to pages that they should not
3629 be allowed to request.
3632 NAME: ignore_unknown_nameservers
3634 LOC: Config.onoff.ignore_unknown_nameservers
3637 By default Squid checks that DNS responses are received
3638 from the same IP addresses that they are sent to. If they
3639 don't match, Squid ignores the response and writes a warning
3640 message to cache.log. You can allow responses from unknown
3641 nameservers by setting this option to 'off'.
3644 NAME: digest_generation
3645 IFDEF: USE_CACHE_DIGESTS
3647 LOC: Config.onoff.digest_generation
3650 This controls whether the server will generate a Cache Digest
3651 of its contents. By default, Cache Digest generation is
3652 enabled if Squid is compiled with USE_CACHE_DIGESTS defined.
3655 NAME: digest_bits_per_entry
3656 IFDEF: USE_CACHE_DIGESTS
3658 LOC: Config.digest.bits_per_entry
3661 This is the number of bits of the server's Cache Digest which
3662 will be associated with the Digest entry for a given HTTP
3663 Method and URL (public key) combination. The default is 5.
3666 NAME: digest_rebuild_period
3667 IFDEF: USE_CACHE_DIGESTS
3670 LOC: Config.digest.rebuild_period
3673 This is the number of seconds between Cache Digest rebuilds.
3676 NAME: digest_rewrite_period
3678 IFDEF: USE_CACHE_DIGESTS
3680 LOC: Config.digest.rewrite_period
3683 This is the number of seconds between Cache Digest writes to
3687 NAME: digest_swapout_chunk_size
3690 IFDEF: USE_CACHE_DIGESTS
3691 LOC: Config.digest.swapout_chunk_size
3694 This is the number of bytes of the Cache Digest to write to
3695 disk at a time. It defaults to 4096 bytes (4KB), the Squid
3699 NAME: digest_rebuild_chunk_percentage
3700 COMMENT: (percent, 0-100)
3701 IFDEF: USE_CACHE_DIGESTS
3703 LOC: Config.digest.rebuild_chunk_percentage
3706 This is the percentage of the Cache Digest to be scanned at a
3707 time. By default it is set to 10% of the Cache Digest.
3712 LOC: Config.chroot_dir
3715 Use this to have Squid do a chroot() while initializing. This
3716 also causes Squid to fully drop root privileges after
3717 initializing. This means, for example, that if you use a HTTP
3718 port less than 1024 and try to reconfigure, you will get an
3722 NAME: client_persistent_connections
3724 LOC: Config.onoff.client_pconns
3728 NAME: server_persistent_connections
3730 LOC: Config.onoff.server_pconns
3733 Persistent connection support for clients and servers. By
3734 default, Squid uses persistent connections (when allowed)
3735 with its clients and servers. You can use these options to
3736 disable persistent connections with clients and/or servers.
3739 NAME: pipeline_prefetch
3741 LOC: Config.onoff.pipeline_prefetch
3744 To boost the performance of pipelined requests to closer
3745 match that of a non-proxied environment Squid tries to fetch
3746 up to two requests in parallell from a pipeline.
3749 NAME: extension_methods
3751 LOC: Config.ext_methods
3754 Squid only knows about standardized HTTP request methods.
3755 You can add up to 20 additional "extension" methods here.
3758 NAME: high_response_time_warning
3761 LOC: Config.warnings.high_rptm
3764 If the one-minute median response time exceeds this value,
3765 Squid prints a WARNING with debug level 0 to get the
3766 administrators attention. The value is in milliseconds.
3769 NAME: high_page_fault_warning
3771 LOC: Config.warnings.high_pf
3774 If the one-minute average page fault rate exceeds this
3775 value, Squid prints a WARNING with debug level 0 to get
3776 the administrators attention. The value is in page faults
3780 NAME: high_memory_warning
3782 LOC: Config.warnings.high_memory
3785 If the memory usage (as determined by mallinfo) exceeds
3786 value, Squid prints a WARNING with debug level 0 to get
3787 the administrators attention.
3790 NAME: store_dir_select_algorithm
3792 LOC: Config.store_dir_select_algorithm
3795 Set this to 'round-robin' as an alternative.
3802 LOC: Config.Log.forward
3804 Logs the server-side requests.
3806 This is currently work in progress.
3812 LOC: Config.onoff.ie_refresh
3815 Microsoft Internet Explorer up until version 5.5 Service
3816 Pack 1 has an issue with transparent proxies, wherein it
3817 is impossible to force a refresh. Turning this on provides
3818 a partial fix to the problem, by causing all IMS-REFRESH
3819 requests from older IE versions to check the origin server
3820 for fresh content. This reduces hit ratio by some amount
3821 (~10% in my experience), but allows users to actually get
3822 fresh content when they want it. Note that because Squid
3823 cannot tell if the user is using 5.5 or 5.5SP1, the behavior
3824 of 5.5 is unchanged from old versions of Squid (i.e. a
3825 forced refresh is impossible). Newer versions of IE will,
3826 hopefully, continue to have the new behavior and will be
3827 handled based on that assumption. This option defaults to
3828 the old Squid behavior, which is better for hit ratios but
3829 worse for clients using IE, if they need to be able to
3830 force fresh content.
3833 NAME: vary_ignore_expire
3836 LOC: Config.onoff.vary_ignore_expire
3839 Many HTTP servers supporting Vary gives such objects
3840 immediate expiry time with no cache-control header
3841 when requested by a HTTP/1.0 client. This option
3842 enables Squid to ignore such expiry times until
3843 HTTP/1.1 is fully implemented.
3844 WARNING: This may eventually cause some varying
3845 objects not intended for caching to get cached.
3848 NAME: sleep_after_fork
3849 COMMENT: (microseconds)
3851 LOC: Config.sleep_after_fork
3854 When this is set to a non-zero value, the main Squid process
3855 sleeps the specified number of microseconds after a fork()
3856 system call. This sleep may help the situation where your
3857 system reports fork() failures due to lack of (virtual)
3858 memory. Note, however, that if you have a lot of child
3859 processes, then these sleep delays will add up and your
3860 Squid will not service requests for some amount of time
3861 until all the child processes have been started.