The Squid Team are pleased to announce the release of Squid-3.0.STABLE21.
This new release is available for download from http://www.squid-cache.org/Versions/v3/3.0/ or the mirrors.
A large number of the show-stopper bugs have been fixed along with general improvements to the ICAP support and additional Languages.
We welcome feedback and bug reports. If you find a bug, please see http://wiki.squid-cache.org/SquidFaq/TroubleShooting#head-7067fc0034ce967e67911becaabb8c95a34d576d for how to submit a report with a stack trace.
Although this release is deemed good enough for use in many setups, please note the existence of open bugs against Squid-3.0.
The 3.0 change history can be viewed here.
Squid 3.0 represents a major rewrite of Squid and has a number of new features.
The most important of these are:
Most user-facing changes are reflected in squid.conf (see below).
Squid 3.0 supports ICAP/1.0. To enable ICAP support, use the --enable-icap-client ./configure option and icap_enable squid.conf option. You will also need to configure ICAP services in your squid.conf using icap_service, icap_class, and icap_access options. The following example instructs Squid to talk to two ICAP services, one for request and one for response adaptation:
icap_enable on icap_service service_req reqmod_precache 1 icap://127.0.0.1:1344/request icap_service service_resp respmod_precache 0 icap://127.0.0.1:1344/response icap_class class_req service_req icap_class class_resp service_resp icap_access class_req allow all icap_access class_resp allow all
Please see squid.conf.default for more details about these and many other icap_* options.
Squid supports pre-cache request and pre-cache response vectoring points. The following ICAP features are supported: message preview, 204 responses outside of preview, request satisfaction, X-Transfer-* negotiation, persistent ICAP connections, client IP/credentials sharing, and optional bypass of certain service failures.
No more than one ICAP service can be applied to an HTTP message. In other words, chaining or load balancing multiple services is not yet supported.
Proxy-directed data trickling and patience pages are not supported yet.
Following ICAP requirements, Squid never performs HTTP message adaptation without a successful and fresh ICAP OPTIONS response on file. A REQMOD or RESPMOD request will not be sent to a configured ICAP service until Squid receives a valid OPTIONS response from that service. If a service malfunctions or goes down, Squid may stop talking to the service for a while. Several squid.conf options can be used to tune the failure bypass algorithm (e.g., icap_service_failure_limit and icap_service_revival_delay).
The bypass parameter of the icap_service squid.conf option determines whether Squid will try to bypass service failures. Most connectivity and preview-stage failures can be bypassed.
More information about ICAP can be found from the ICAP-forum website http://www.icap-forum.org
ESI is an open specification of an markup language enabling reverse proxies to perform some simple XML based processing, offloading the final page assembly from the webserver and similar tasks.
More information about ESI can be found from the ESI website http://www.esi.org
Some of the features found in Squid-2.6 is not available in Squid-3. Some have been dropped as they are not needed. Some have not yet been forward-ported to Squid-3 and may appear in a later release.
The TCP_REFRESH_HIT and TCP_REFRESH_MISS log types have been replaced because they were misleading (all refreshes need to query the origin server, so they could never be hits). The following log types have been introduced to replace them:
The requested object was cached but STALE. The IMS query for the object resulted in "304 not modified".
The requested object was cached but STALE. The IMS query returned the new content.
See http://www.squid-cache.org/Doc/FAQ/FAQ-6.html#ss6.7 for a definition of all log types.
This Squid version can run on Windows as a system service using the Cygwin emulation environment,
or can be compiled in Windows native mode using the MinGW + MSYS development environment. Windows NT 4 SP4 and later are supported.
On Windows 2000 and later the service is configured to use the Windows Service Recovery option
restarting automatically after 60 seconds.
Some new command line options were added for the Windows service support:
The service installation is made with -i command line switch, it's possible to use -f switch at the same time for specify a different config-file settings for the Squid Service that will be stored on the Windows Registry.
A new -n switch specify the Windows Service Name, so multiple Squid instance are allowed. "Squid" is the default when the switch is not used.
So, to install the service, the syntax is:
squid -i [-f file] [-n name]
Service uninstallation is made with -r command line switch with the appropriate -n switch.
The -k switch family must be used with the appropriate -f and -n switches, so the syntax is:
squid -k command [-f file] -n service-namewhere service-name is the name specified with -n options at service install time.
To use the Squid original command line, the new -O switch must be used ONCE, the syntax is:
squid -O cmdline [-n service-name]If multiple service command line options must be specified, use quote. The -n switch is needed only when a non default service name is in use.
Don't use the "Start parameters" in the Windows 2000/XP/2003 Service applet: they are specific to Windows services functionality and Squid is not designed for understand they.
In the following example the command line of the "squidsvc" Squid service is set to "-D -u 3130":
squid -O "-D -u 3130" -n squidsvc
The process status helper functions make it easier for you to obtain information about processes and device drivers running on Microsoft® Windows NT®/Windows® 2000. These functions are available in PSAPI.DLL, which is distributed in the Microsoft® Platform Software Development Kit (SDK). The same information is generally available through the performance data in the registry, but it is more difficult to get to it. PSAPI.DLL is freely redistributable.
PSAPI.DLL is available only on Windows NT, 2000, XP and 2003. The implementation in Squid is aware of this, and try to use it only on the right platform.
On Windows NT PSAPI.DLL can be found as component of many applications, if you need it, you can find it on Windows NT Resource KIT. If you have problem, it can be downloaded from here: http://download.microsoft.com/download/platformsdk/Redist/4.0.1371.1/NT4/EN-US/psinst.EXE
On Windows 2000 and later it is available installing the Windows Support Tools, located on the Support\Tools folder of the installation Windows CD-ROM.
On Windows platforms, if no value is specified in the dns_nameservers option on squid.conf or in the /etc/resolv.conf file, the list of DNS name servers are taken from the Windows registry, both static and dynamic DHCP configurations are supported.
acl blocklist url_regex -i "c:/squid/etc/blocked1.txt"
redirect_program c:/perl/bin/perl.exe c:/squid/libexec/redir.pl redirect_program c:/winnt/system32/cmd.exe /C c:/squid/libexec/redir.cmd
A reasonably recent release of
Cygwin or
MinGW is needed.
The usage of the Cygwin environment is very similar to other Unix/Linux environments, and -devel version of libraries must be installed.
For the MinGW environment, the packages MSYS, MinGW and msysDTK must be installed. Some additional libraries and tools must be downloaded separately:
OpenSSL:
Shining Light Productions Win32 OpenSSL
libcrypt:
MinGW packages repository
db-1.85:
TinyCOBOL download area
uudecode:
Native Win32 ports of some GNU utilities
When running configure, --disable-wccp and --disable-wccpv2 options should always specified to avoid compile errors.
On Windows, cache manager (cachemgr.cgi) can be used with Microsoft IIS or Apache.
Some specific configuration could be needed:
ScriptAlias /squid/cgi-bin/ "c:/squid/libexec/"
<Location /squid/cgi-bin/cachemgr.cgi>
PassEnv TMP TEMP
Order allow,deny
Allow from workstation.example.com
</Location>
There have been many changes to Squid's configuration file since Squid-2.6.
This section gives a detailed account of those changes in three categories:
Default: 5
Normally the ICP query timeout is determined dynamically. But
sometimes it can lead to very small timeouts, even lower than
the normal latency variance on your link due to traffic.
Use this option to put an lower limit on the dynamic timeout
value. Do NOT use this option to always use a fixed (instead
of a dynamic) timeout value. To set a fixed timeout see the
'icp_query_timeout' directive.
Default: 10 seconds
Controls how often the ICP pings are sent to siblings that
have background-ping set.
Default: unset
Surrogates (http://www.esi.org/architecture_spec_1.0.html)
need an identification token to allow control targeting. Because
a farm of surrogates may all perform the same tasks, they may share
an identification token.
Default: off
Remote surrogates (such as those in a CDN) honour Surrogate-Control: no-store-remote.
Set this to on to have squid behave as a remote surrogate.
Default: custom
ESI markup is not strictly XML compatible. The custom ESI parser
will give higher performance, but cannot handle non ASCII character
encodings.
Default: on
If enabled, information about the occurred error will be
included in the mailto links of the ERR pages (if %W is set)
so that the email body contains the data.
Syntax is <A HREF="mailto:%w%W">%w</A>
Default: off
When you enable this option, squid will always check
the origin server for an update when a client sends an
If-Modified-Since request. Many browsers use IMS
requests when the user requests a reload, and this
ensures those clients receive the latest version.
By default (off), squid may return a Not Modified response
based on the age of the cached version.
Replaces the header_access directive of Squid-2.6 and earlier, but applies to requests only.
Replaces the header_access directive of Squid-2.6 and earlier, but applies to replies only.
Default: off
If you want to enable the ICAP module support, set this to on.
Default: off
Set this to 'on' if you want to enable the ICAP preview
feature in Squid.
Default: -1
The default size of preview data to be sent to the ICAP server.
-1 means no preview. This value might be overwritten on a per server
basis by OPTIONS requests.
Default: 60
The default TTL value for ICAP OPTIONS responses that don't have
an Options-TTL header.
Default: on
Whether or not Squid should use persistent connections to
an ICAP server.
Default: off
This adds the header "X-Client-IP" to ICAP requests.
Default: off
This adds the header "X-Client-Username" to ICAP requests
if proxy access is authentified.
Default: none
Defines a single ICAP service
icap_service servicename vectoring_point bypass service_url
vectoring_point = reqmod_precache|reqmod_postcache|respmod_precache|respmod_postcache
This specifies at which point of request processing the ICAP
service should be plugged in.
bypass = 1|0
If set to 1 and the ICAP server cannot be reached, the request will go
through without being processed by an ICAP server
service_url = icap://servername:port/service
Note: reqmod_postcache and respmod_postcache is not yet implemented
Example:
icap_service service_1 reqmod_precache 0 icap://icap1.mydomain.net:1344/reqmod
icap_service service_2 respmod_precache 0 icap://icap2.mydomain.net:1344/respmod
Default: none
Defines an ICAP service chain. If there are multiple services per
vectoring point, they are processed in the specified order.
icap_class classname servicename...
Example:
icap_class class_1 service_1 service_2
icap class class_2 service_1 service_3
Default: none
Redirects a request through an ICAP service class, depending
on given acls
icap_access classname allow|deny [!]aclname...
The icap_access statements are processed in the order they appear in
this configuration file. If an access list matches, the processing stops.
For an "allow" rule, the specified class is used for the request. A "deny"
rule simply stops processing without using the class. You can also use the
special classname "None".
For backward compatibility, it is also possible to use services
directly here.
Example:
icap_access class_1 allow all
The name of an accept(2) filter to install on Squid's
listen socket(s). This feature is perhaps specific to
FreeBSD and requires support in the kernel.
The 'httpready' filter delays delivering new connections
to Squid until a full HTTP request has been received.
See the accf_http(9) man page.
New option to import entire secondary configuration files into squid.conf.
Squid will follow the files immediately and insert all their content
as if it was at that position in squid.conf. As per squid.conf some
options are order-specific within the config as a whole.
A few layers of include are allowed, but too many are confusing and
squid will enforce an include depth of 16 files.
Syntax:
include /path/to/file1 /path/to/file2
New acl type myportname, matching the name of the http(s)_port where the request was accepted
acl aclname myportname 3128 ... # http(s)_port name
Ported from 2.6. Behaviour identical.
Minimum umask which should be enforced while the proxy
is running, in addition to the umask set at startup.
For a traditional octal representation of umasks, start
your value with 0.
New tag to fix handling of chunked requests.
A broken or confused HTTP/1.1 client may send a chunked HTTP
request to Squid. Squid does not have full support for that
feature yet. To cope with such requests, Squid buffers the
entire request and then dechunks request body to create a
plain HTTP/1.0 request with a known content length. The plain
request is then used by the rest of Squid code as usual.
The option value specifies the maximum size of the buffer used
to hold the request before the conversion. If the chunked
request size exceeds the specified limit, the conversion
fails, and the client receives an "unsupported request" error,
as if dechunking was disabled.
Dechunking is enabled by default. To disable conversion of
chunked requests, set the maximum to zero.
Request dechunking feature and this option in particular are a
temporary hack. When chunking requests and responses are fully
supported, there will be no need to buffer a chunked request.
New options:
disable-pmtu-discovery=
Control Path-MTU discovery usage:
off lets OS decide on what to do (default).
transparent disable PMTU discovery when transparent support is enabled.
always disable always PMTU discovery.
In many setups of transparently intercepting proxies Path-MTU
discovery can not work on traffic towards the clients. This is
the case when the intercepting device does not fully track
connections and fails to forward ICMP must fragment messages
to the cache server. If you have such setup and experience that
certain clients sporadically hang or never complete requests set
disable-pmtu-discovery option to 'transparent'.
New options:
basetime=n
background-ping
weighted-round-robin
use 'basetime=n' to specify a base amount to
be subtracted from round trip times of parents.
It is subtracted before division by weight in calculating
which parent to fectch from. If the rtt is less than the
base time the rtt is set to a minimal value.
use 'background-ping' to only send ICP queries to this
neighbor infrequently. This is used to keep the neighbor
round trip time updated and is usually used in
conjunction with weighted-round-robin.
use 'weighted-round-robin' to define a set of parents
which should be used in a round-robin fashion with the
frequency of each parent being based on the round trip
time. Closer parents are used more often.
Usually used for background-ping parents.
Common options no-store, replaces the older read-only option
Removed Basic auth options:
blankpasswor, not yet ported to squid-3.
New format specifications:
%URI Requested URI
%PATH Requested URL path
New result keywords:
tag= Apply a tag to a request (for both ERR and OK results)
Only sets a tag, does not alter existing tags.
New options:
ignore-no-store
refresh-ims
ignore-no-store ignores any ``Cache-control: no-store''
headers received from a server. Doing this VIOLATES
the HTTP standard. Enabling this feature could make you
liable for problems which it causes.
refresh-ims causes squid to contact the origin server
when a client issues an If-Modified-Since request. This
ensures that the client will receive an updated version
if one is available.
The 'all' ACL is now provided as a built-in. Warnings will be displayed if any attempt is made to redefine it.
New types:
acl aclname http_status 200 301 500- 400-403 ... # status code in reply
New default:
Default: on
(Old default: off)
New delay classes:
class 4 Everything in a class 3 delay pool, with an
additional limit on a per user basis. This
only takes effect if the username is established
in advance - by forcing authentication in your
http_access rules.
class 5 Requests are grouped according their tag (see
external_acl's tag= reply).
New default to require the feature to be enabled in squid.conf:
Default: 0 (disabled)
(Old default: 4827)
New default to require the feature to be enabled in squid.conf:
Default: 0 (disabled)
(Old default: 3130)
New default to require the feature to be enabled in squid.conf:
Default: 0 (disabled)
(Old default: 3401)
New format tags:
rp Request URL-Path excluding hostname
et Tag returned by external acl
<sH Reply high offset sent
<sS Upstream object size
Syntax changed:
reply_body_max_size size [acl acl...]
allow/deny no longer used.
No urlgroup support in either requests or response
fake_auth helper for NTLM now accepts the '-S' parameter to strip NTLM domain off the username string. This is useful for class 4 Delay Pools in Squid 3.x
New default value of OFF
This has been replaced by request_header_access and reply_header_access
Replaced by disable-pmtu-discovery http_port option
equivalent to cache_peer + cache_peer_access.
There have been some changes to Squid's build configuration since Squid-2.6.
This section gives an account of those changes in three categories:
Build shared libraries. The default is to build without.
Build static libraries. The default is on.
Optimize for fast installation
default: yes
Avoid locking (might break parallel builds)
Don't compile Squid with compiler optimizations enabled. Optimization is good for production builds, but not good for debugging. During development, use --disable-optimizations to reduce compilation times and allow easier debugging. This option implicitly also enables --disable-inline
Don't compile trivial methods as inline. Squid is coded with much of the code able to be inlined. Inlining is good for production builds, but not good for development. During development, use --disable-inline to reduce compilation times and allow incremental builds to be quick. For production builds, or load tests, use --enable-inline to have squid make all trivial methods inlinable by the compiler.
Provide some debug information in cbdata
Build support for the list of disk I/O modules. The default is only to build the "Blocking" module. See src/DiskIO for a list of available modules, or Programmers Guide for details on how to build your custom disk module.
Enable ESI for accelerators. Requires libexpat. Enabling ESI will cause squid to follow the Edge Acceleration Specification (www.esi.org). This causes squid to IGNORE client Cache-Control headers.
DO NOT use this in a squid configured as a web proxy, ONLY use it in a squid configured for webserver acceleration.
Enable the ICAP client.
Disable SNMP monitoring support which is now built by default.
Disable HTCP protocol support which is now built by default.
Enable kqueue() support. Marked as experimental in 3.0.
Enable Transparent Proxy support for systems using FreeBSD IPFW style redirection.
Disable memPools. Note that this option now simply sets the default behaviour. Specific classes can override this at runtime, and only lib/MemPool.c needs to be altered to change the squid-wide default for all classes.
This option allows you to see which internal functions in Squid are consuming how much CPU. Compiles in probes that measure time spent in probed functions. Needs source modifications to add new probes. This is meant for developers to assist in performance optimisations of Squid internal functions.
If you are not developer and not interested in the stats you shouldn't enable this, as overhead added, although small, is still overhead. See lib/Profiler.c for more.
Assume the C compiler uses GNU ld. The default is to auto-detect.
Try to use only PIC/non-PIC objects. The default is to use both.
Include additional configurations. The default is automatic.
Sets the default System User account for squid permissions. The default is 'nobody' as in other releases of squid.
Path where the cppunit headers and libraries are found for unit testing. The default is automatic detection.
NOTE: Since 3.0-PRE6 and 2.6STABLE14 squid no longer comes bundled with CPPUnit. Compile-time validation will be disabled if it is not installed on your system.
CARP support is now built by default. --disable-carp can be used to build without it.
HTCP protocol support is now built by default. Use --disable-htcp to build without it.
SNMP monitoring is now build by default. Use --disable-snmp to build without it.
Please use --enable-removal-policies directive instead.
Replaced by --with-filedescriptors=N
Override maximum number of filedescriptors. Useful if you build as another user who is not privileged to use the number of filedescriptors you want the resulting binary to support
Deprecated. Automatic checks will enable best I/O loop method available.
Deprecated. Automatic checks will enable best I/O loop method available.
Deprecated. Automatic checks will enable best I/O loop method available.
kqueue support is marked Experimental in Squid 3.0. Known to have some issues under load.
The following configure options have been removed.
Most OS:es have good malloc implementations these days, and the version we used to ship with Squid was very very old..
Debug option, not needed and therefore removed.
Rarely used extra log file. Removed.
Rarely used feature, and multicast ICP acheives almost the same result. Removed.
Specific to the COSS implementation in Squid-2
Now enabled by default. Configure option was redundant and therefore removed.
Known to cause race conditions where cache objects may get corrupted, and this for at most a marginal performance improvement. Removed.
Some squid.conf and ./configure options which were available in Squid-2.7 are not yet available in Squid-3.0
If you need something to do then porting one of these from Squid-2 to Squid-3 is most welcome.
urllogin option not yet ported from 2.6
urlgroup option not yet ported from 2.6
Not yet ported from 2.6
concurrency option not yet ported from Squid-2
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.6
min-size option not yet ported from Squid-2
COSS storage type is lacking stability fixes from 2.6
COSS overwrite-percent= option not yet ported from 2.6
COSS max-stripe-waste= option not yet ported from 2.6
COSS membufs= option not yet ported from 2.6
COSS maxfullbufs= option not yet ported from 2.6
multicast-siblings not yet ported from 2.7
idle= not yet ported from 2.7
http11 not yet ported from 2.7
connection-auth= not yet ported from 2.6
monitorinterval= not yet ported from 2.6
monitorsize= not yet ported from 2.6
monitortimeout= not yet ported from 2.6
monitorurl= not yet ported from 2.6
Not yet ported from 2.6
Not yet ported from 2.6
Not yet ported from 2.6
Not yet ported from 2.6
%ACL format tag not yet ported from 2.6
%DATA format tag not yet ported from 2.6
Not yet ported from 2.7
Not yet ported from 2.6
Not yet ported from 2.6
act-as-origin not yet ported from 2.7
allow-direct not yet ported from 2.7
http11 not yet ported from 2.7
urlgroup= not yet ported from 2.6
no-connection-auth not yet ported from 2.6
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.6
Not yet ported from 2.6
Not yet ported from 2.6
Not yet ported from 2.6
Not yet ported from 2.6
Not yet ported from 2.7
%oa tag not yet ported from 2.7
%sn tag not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
stale-while-revalidate= not yet ported from 2.7
ignore-stale-while-revalidate= not yet ported from 2.7
max-stale= not yet ported from 2.7
negative-ttl= not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
Not yet ported from 2.7
Support for Solaris /dev/poll
Basic POSIX select() loop without any binary fd_set optimizations.
Support following the X-Forwarded-For HTTP header for determining the client IP address