DragonFly On-Line Manual Pages
SQUIDCLAMAV(1) User Contributed Perl Documentation SQUIDCLAMAV(1)
NAME
SquidClamav v6 - HTTP Antivirus for Squid based on ClamAv and the ICAP
protocol
DESCRIPTION
SquidClamav v6 is an antivirus for the Squid proxy based on the ICAP
protocol and the awards-winning ClamAv anti-virus toolkit. Using it
will help you securing your home or enterprise network web traffic.
SquidClamav is the most efficient antivirus tool for HTTP traffic
available for free, it is written in C as a c-icap service and can
handle several thousands of connections at once.
SquidClamav v6 only scan the HTTP stream sent by Squid through the ICAP
server. It doesn't make HTTP requests itself so this is a gain of
performance and ensures that the data scanned is the same as the user
has requested.
Why use c-icap server? This is the only open source icap server written
in C, it is very fast and stable.
Why writing another clamav c-icap module? Well, to be honest, outside
the survival of SquidClamav, I think that using clamd instead of
libclamav to scan files is speediest and more simple than the
srv_clamav module provided with the c-icap server.
SquidClamav v6 is faster than any other HTTP antivirus and can handle
several thousands of simultaneous users at once, this is what we need.
The other unique feature of SquidClamav is that you can have Clamd
failover by setting up up to 4 clamd server IP addresses. When a clamd
server is not reachable in one second, SquidClamav switches to the next
IP address.
If you are using ClamAV above 0.95, SquidClamav will have support for
Google Safe Browsing database. All signatures provided by Google Safe
Browsing Database will be prefixed with the Safebrowsing tag. If ClamAV
reports:
Safebrowsing.<something> FOUND
This will be redirected by squidclamav just like if a virus was found.
USAGE
Generic Program Information
SquidClamav v6 has been completely rewritten to be used through the
Squid v3.x ICAP feature allowing "on stream" scanning. It is now built
as a c-icap server service but keeps all features from v5 and is fully
compatible with the old SquidClamav configuration file. The squidclamav
configuration file is unchanged minus some obsolete directives.
This also means that SquidClamav can no more be run into an interactive
console for testing your URL. All debug information will now go to the
c-icap logfile.
Installing Squid
Setting SquidClamav as Squid Icap service
I want SquidClamav to be installed as a c-icap service, to be
configured as easy as possible and to be compatible with the old
configuration file. This means that I voluntary omit some capabilities
of c-icap server to preserve a full compatibility with the old
squidclamav.conf file.
Squid v3.x installation and configuration
To have full and stable icap support with Squid you must use the 3.x
branch and configure squid with the following option:
--enable-icap-client
I don't know what other options you are using but you have to add this
one to your configure command. If you prefer to use distribution
packaging you may already have it configured like this if you can
install the c-icap package too.
If you don't know, run the following command an search for the
configuration directive: --enable-icap-client
/usr/local/squid/sbin/squid -v | grep "enable-icap-client"
If it is not enable you must reinstall Squid with this configuration
option or install the additional packages.
Once you have it enabled, to integrate c-icap and SquidClamav to your
squid cache just edit squid.conf and set the following directives.
Squid 3.4.x configuration
There are some configuration differences between 3.1.x and 3.4.x
Squid version. Here are the directives I use for Squid 3.4.x:
icap_enable on
icap_send_client_ip on
icap_send_client_username on
icap_client_username_encode off
icap_client_username_header X-Authenticated-User
icap_preview_enable on
icap_preview_size 1024
icap_service service_avi_req reqmod_precache
icap://localhost:1344/squidclamav bypass=off
adaptation_access service_avi_req allow all
icap_service service_avi_resp respmod_precache
icap://localhost:1344/squidclamav bypass=on
adaptation_access service_avi_resp allow all
If you don't know where to put them in squid.conf, just search for
'icap_.*' and add those configuration lines at the end of the icap
section.
Squid 3.1.x configuration
There are some configuration differences between 3.1.x and 3.0.x
Squid version. Here are the directives I use for Squid 3.1.x:
icap_enable on
icap_send_client_ip on
icap_send_client_username on
icap_client_username_encode off
icap_client_username_header X-Authenticated-User
icap_preview_enable on
icap_preview_size 1024
icap_service service_req reqmod_precache bypass=1 icap://127.0.0.1:1344/squidclamav
adaptation_access service_req allow all
icap_service service_resp respmod_precache bypass=1 icap://127.0.0.1:1344/squidclamav
adaptation_access service_resp allow all
If you don't know where to put them in squid.conf, just search for
'icap_.*' and add those configuration lines at the end of the icap
section.
Here the bypass is set to 1, that means that in case of squidclamav
problems squid will simply ignore the error and continue. This is
the equivalent of the bridge mode in version 5.x of suidclamav.
Squid 3.0.x configuration
For squid 3.0.x you must replace 'bypass=1' by '1' or 'bypass=0' by
'0' and the access to the service is defined at a class level. Only
the last four configuration lines change from version 3.1.x.
icap_enable on
icap_send_client_ip on
icap_send_client_username on
icap_client_username_encode off
icap_client_username_header X-Authenticated-User
icap_preview_enable on
icap_preview_size 1024
icap_service service_req reqmod_precache 1 icap://127.0.0.1:1344/squidclamav
icap_service service_resp respmod_precache 1 icap://127.0.0.1:1344/squidclamav
icap_class class_avreq service_req
icap_class class_avresp service_resp
icap_access class_avreq allow all
icap_access class_avresp allow all
If you don't know where to put them in squid.conf, just search for
'icap_.*' and add those configuration lines at the end of the icap
section.
Here the bypass is set to 1, that means that in case of squidclamav
problems squid will simply ignore the error and continue. This is
the equivalent of the bridge mode in version 5.x of suidclamav.
What do that configuration directives do? They enable Squid's ICAP
client and tell Squid to send the logged username and client's IP
address to the ICAP server. They also enable preview for faster
SquidClamav work. The last four lines define how to call the ICAP
server. Here we call the squidclamav service on localhost and port 1344
(host and port can be changed). The bypass parameter set to 1 means
that Squid will continue without bothering about ICAP server or
SquidClamav failure. This is just like the old bridge mode in previous
releases of SquidClamAV. I don't want users to be bored by a
continuously error message if SquidClamav or c-icap produce errors or
if there's an error in the configuration file. Users don't have to know
about that, they want to surf and don't care about your problems :-) If
you don't think like me, just set the bypass argument to 0 and Squid
will return an error message in case of a failure.
C-icap server installation/configuration
If you don't have package solutions or encounter problems when
installing SquidClamav I recommand you to install the c-icap server
from source as following. You can download it from SourceForge at
http://c-icap.sourceforge.net/. Choose version c-icap-0.3.2 or later
versions, then run:
./configure --prefix=/usr/local/c-icap --enable-large-files
make
make install
Then, edit the file /usr/local/c-icap/etc/c-icap.conf. It contains a
set of documented values that configure the c-icap server. To enable
the support of SquidClamav just add the following line to the end of
the file:
Service squidclamav squidclamav.so
Don't care about the srv_clamav.* configuration directives, this will
not break anything. SquidClamav doesn't use them but reads its own
directives from the file /usr/local/etc/c-icap/squidclamav.conf.
You can disable the c-icap embedded modules by commenting out these
lines:
#Service url_check_module srv_url_check.so
#Service antivirus_module srv_clamav.so
This will preserve some resources.
Following your installation you may need to create the /var/run/c-icap/
where c-icap server is writing pid and socket file.
You may also want to change the user/group owning c-icap's processes.
By default the owner is the user/group who runs the program. I
recommand you to change them to the same user/group running your Squid
cache. For example:
User proxy
Group proxy
Of course you will need to change the owner of directory
/var/run/c-icap/ and the directory of your server log. See the
ServerLog directive to get the path. For me, I use the following
commands to set the good rights on my installation:
mkdir /var/run/c-icap/
chown -R proxy:proxy /var/run/c-icap/
chown -R proxy:proxy /usr/local/c-icap/
After that you can run the c-icap server as explained below.
SquidClamav installation/configuration
Installing SquidClamav requires that you already have installed the
c-icap as explained above. You must provide the installation path of
c-icap to the configure command as following:
./configure
make
make install
This will install the squidclamav.so library into the c-icap
modules/services repository.
Note that if the c-icap installation does not save the c-icap-config
program in a directory that can be found in your default path you will
need to give the path to this program to squidclamav at configure time:
./configure --with-c-icap=/usr/local/c-icap/
make && make install
Running c-icap server
Finally, you can run the c-icap server as root user:
/usr/local/c-icap/bin/c-icap
or any other path to the binary. If you want to display debugging
information on the terminal, the previous command should be executed
with the following arguments:
/usr/local/c-icap/bin/c-icap -N -D -d 10
The first argument -N prevents the c-icap server from forking into the
background, the second argument -D enables the printing of messages to
standard output, and the third argument -d 10 enables the printing of
full debugging information.
Reloading configuration without restarting the c-icap server
To force SquidClamav to reread its configuration file after changes you
can send the following command to the c-icap server
echo -n "squidclamav:cfgreload" > /var/run/c-icap/c-icap.ctl
It will reread all its configuration directives and restart pipes to
squidGuard. So if you make changes to squidGuard you must execute this
command to activate them in SquidClamav.
Or to be sure that everything is really initialized or that you have
made change to the c-icap configuration file you can run the following
command:
echo -n "reconfigure" > /var/run/c-icap/c-icap.ctl
The service will reread the config file without the need for stopping
and restarting the c-icap server. The service will just be
reinitialized.
CONFIGURATION
By default, the configuration file must be
/usr/local/etc/c-icap/squidclamav.conf, you may not use an other path
unless you change it in the source code (see src/squidclamav.h).
SquidClamav installation will create a default file with the maximum
security level. If you have low resources on your server there's some
predefined pattern optimized for speed. Feel free to modify it to match
your desired security level.
The format of the configuration file consists in always lower case
configuration directive names followed by a value. The name and the
value must be separated by a single space character. Comments are lines
starting with a '#' character.
Global configuration
Log file and debug
In version 6.x the directives 'logfile', 'debug' and 'stat' are
obsolete as logging and debug are now handled by the c-icap server. You
can control them using the following c-icap.conf directives:
ServerLog /usr/local/c-icap/var/log/server.log
DebugLevel 0
Debug information is disable by default, do not enable it on production
systems as it costs a lot of performance. The debug level can be set
from 1 up to 3 for SquidClamav but can be up to 10 for c-icap.
Clamd daemon
SquidClamav needs to know where to contact clamd, the ClamAV daemon,
for on stream virus scanning.
clamd_local /tmp/clamd
#clamd_ip 192.168.1.5
#clamd_port 3310
By default SquidClamav will contact clamd locally on the /tmp/clamd
unix socket (clamd_local). If your clamd daemon uses INET socket or
stays in a remote server, you have to set the IP address and the port
with clamd_ip and clamd_port.
If you use INET socket the 'clamd_local' directive must be commented,
or SquidClamav will always use the clamd_local directive.
Clamd failover
If you have multiple ClamAv servers, SquidClamav is able to do failover
between them. You just have to set 'clamd_ip' to a list of IP adresses
separated by a comma. Do not insert space characters in this list or it
will break all. For example:
clamd_ip 192.168.1.5,192.168.1.13,192.168.1.9
clamd_port 3310
timeout 1
You can set up to 5 clamd servers. The clamd port must be the same for
all these servers as 'clamd_port' only accepts one single value.
SquidClamav will always connect to the first IP address available. If
this fails it will try the next defined IP address after 1 second. When
a connect can be established SquidClamav will reuse this last "working"
IP address first to not slow down process the next time.
If you think 1 second is a low value, you can change the connect
timeout by editing file squidclamav.conf and set the 'timeout'
directive to a higher value. For example :
timeout 2
Value must be set in seconds. Do not set it too high (< 5) or you can
slow down everything.
Redirection
URL redirect
When a virus is detected SquidClamav needs to redirect the client to a
warning page. The SquidClamav distribution contains a set of Perl CGI
scripts with different languages that you can use. To specify this
redirection you have to have to specify a redirect URL to the
'redirect' directive as follow:
redirect http://proxy.samse.fr/cgi-bin/clwarn.cgi
Take a look in the cgi-bin directory to see all translations of this
cgi script.
Squidclamav will pass the following parameters to this CGI:
url=ORIGNAL_HTTP_REQUEST
virus=NAME_OF_THE_VIRUS
source=DOWNLOADER_IP_ADDRESS
user=DOWNLOADER_IDENT
If this directive is disabled squidclamav will use c-icap error
templates
to report issues. See below.
Using c-icap template instead of redirect scripts
If the redirect directive is not set, SquidClamav will attempt to load
a template up from disk and send this back to the user. By default this
template is found at the following path:
/usr/share/c_icap/templates/squidclamav/en/MALWARE_FOUND
Available format tokens are all of those available to the LogFormat
directive of c-icap, plus an additional token:
%mn - formatted name of the malware, as given by ClamAV.
Notice redirection into log file
To log every redirection enable the 'logredir' configuration directive:
logredir 1
By default it is disabled as you can also log this information with the
cgi-script or send an email.
Chained Url Checker
The squidguard directive is preserved for backward compatibility but
you must remove it from your configuration file as it could result in
many squidclamav crashes.
Please use the 'url_rewrite_program' squid.conf directive instead to
call squidGuard.
url_rewrite_program /usr/bin/squidGuard
url_rewrite_children 15
url_rewrite_access allow all
If you still want to use it, SquidClamav allows you to chain the
SquidGuard program to check the URL requested against blocklists using
the 'squidguard' directive. You just have to give the path to the
program.
squidguard /usr/local/squidGuard/bin/squidGuard
The chained program is called before the virus scan and any other
SquidClamav operations. The call to this program can be disabled with
the 'whitelist', 'trustuser' and 'trustclient' directives. See
SquidClamav Patterns for more information.
To log every chained program redirection enable the 'logredir'
configuration directive as following:
logredir 1
By default it is disabled as you can also log this information with
squidguard.
Maxsize
This directive allows to disable virus scan completely for files bigger
than the value in bytes. Default is 0, no size limit as you may want to
control download size into squid.conf or clamd.
maxsize 2000000
If you want to abort virus scan after a certain amount of data you must
take a look at the clamd configuration directive 'StreamMaxLength' that
will close a stream when the given size is reached.
Controlling SquidClamav behaviour
As in SquidClamav v5.x, v6.0 will scan all downloaded files by default.
You have five directives to control the way things must work.
All these directives used extended regex pattern matching and are case
insensitive.
Control both chained program and virus scan
There are 3 configuration directives that allow you to disable virus
scan and call to chained redirector like SquidGuard. Those pattern
matchings are searched as soon as a Squid entry is received.
whitelist
The 'whitelist' configuration directive allows you to disable
chained program and virus scan at URL level. When the given pattern
matches the URL, SquidClamav falls back to Squid instantly.
For example:
whitelist \.clamav\.net
will deliver any files from hosts on clamav.net domain directly.
You can alse use a file containing all regex that might be
whitelisted and provide the file name to the whitelist dorective.
Suppose that you have a file named
/usr/local/c-icap/etc/sc_whitelist with the following content:
\.clamav\.net
\.darold\.net
then you just have to set whitelist in squidclamav.conf as follow:
whitelist /usr/local/c-icap/etc/sc_whitelist
the file must contain only one regex per line and no extra
character.
trustuser
The 'trustuser' directive allows you to disable chained program and
virus scan when an ident matches the search pattern. On regex found
SquidClamav falls back to Squid instantly. Of course you must have
Squid authentication helper enabled.
For example:
trustuser administrator
will let user logged as administrator to not be bored by chained
program and virus scan.
trustclient
The 'trustclient' directive allows you to disable chained program
and virus scan if the client source IP address or DNS name match
the search pattern. The source IP address can be a single IP
address or an address range following the given regex pattern.
For example:
trustclient ^192\.168\.1\.1$
trustclient ^192\.168\.1\..*$
trustclient ^mypc\.domain\.dom$
The first and the last entry will disable chained program and virus
scan for a single computer and the second will do for en entire
class C network.
dnslookup
Enable / disable DNS lookup of client IP address. Default is
enabled '1' to preserve backward compatibility but you must
deactivate this feature if you don't use trustclient with hostname
in the regexp or if you don't have a DNS on your network. Disabling
it will also speed up squidclamav.
Safebrowsing
ClamAV 0.95 introduced support for Google Safe Browsing database. The
database is packed inside a CVD file and distributed through our mirror
network as safebrowsing.cvd. This feature is disabled by default on all
clamav installations.
In order to enable this feature, you must first add aXXSafeBrowsing
YesaXX to freshclam.conf. There is no option in clamd.conf. If the
engine finds Google Safe Browsing files in the database directory,
ClamAV will enable safe browsing. To turn it off you need to update
freshclam.conf and remove the safebrowsing files from the database
directory before restarting clamd.
Then to enable this feature into SquidClamav you have to enable the
following configuration directive.
safebrowsing
Enable / Disable Clamav Safe Browsing feature. You mus have enabled
the corresponding behavior in clamd by enabling SafeBrowsing into
freshclam.conf Enabling it will first make a safe browsing request
to clamd and then the virus scan request.
Control virus scan
There are 3 configuration directives that allow you to disable virus
scan for downloaded files.
abort
The 'abort' directive will let you disable virus scanning at URL
level (not chained program). When the URL matches the regex
pattern, SquidClamav falls back to Squid immediately after the call
to the chained program, if one is defined there.
For example:
abort \.squid-cache\.org
abort .*\.(png|gif|jpg)$
The first regexp will exclude any file hosted on domain
squid-cache.org from virus scanning, the last one will exclude all
PNG, GIF and JPEG image from scanning.
abortcontent
The 'abortcontent' directive allows you to exclude any file from
virus scanning, whose Content-Type matches the regex pattern. This
directive costs more time because SquidClamav needs to download the
HTTP header for a file with a HEAD request. Note that some sites do
not answer to HEAD requests so the content type will not be able to
be retrieved so they will be scanned.
Example:
abortcontent ^image\/.*$
abortcontent ^video\/x-flv$
The first directive will complete the "abort .*\.(png|gif|jpg)$"
previous directive to match dynamic image or with parameters at
end. The second will allow your users to view streamed video
instantly.
maxsize
As said above, the 'maxsize' directive allows you not to scan a
file when the content-length of the file is bigger than the defined
value. By default there's no size limit.
Testing SquidClamav
As SquidClamav v6.0 is now a c-icap service, it can no more be run at
console in interactive mode. To check what is going wrong, you must
edit c-icap.conf file, set DebugLevel to 3 and enable ServerLog. Then
check for lines with squidclamav string in the log file which is
defined with ServerLog in squidclamav's config.
Performance
With SquidClamav v6.x the way to tune your service is to tune c-icap
server and clamd daemon. On heavy http access, putting the clamd daemon
on a dedicated server with multiple CPU will really help.
If you experience Squid "ICAP protocol error" (with bypass enabled)
please consider increasing the following c-icp parameters:
StartServers, MaxServers, MinSpareThreads, MaxSpareThreads,
ThreadsPerChild. Increasing MaxThreads parameter in clamd.conf may also
help.
BUGS
Please report any bugs, patches, discussion, etc. to <gilles AT darold
DOT net>.
FEATURE REQUESTS
If you need new features let me know at <gilles AT darold DOT net>.
This helps a lot to develop a better/useful tool.
HOW TO CONTRIBUTE ?
Any contribution to build a better tool is welcome, you just have to
send me your ideas, features requests or patches and they will be
applied.
AUTHOR
Gilles Darold <gilles AT darold DOT net>
ACKNOWLEDGEMENT
Thanks to Squid-cache.org, Clamav.net and c-icap.sf.net for their great
software.
Special thanks to Christos Tsantilas for his implementation of the
c-icap server. Lots of SquidClamav v6 source code has been learned or
simply cut and pasted from the source code of his clamav service.
I must also thank all the great contributors:
- Leonardo Humberto Liporati from www.ig.com.br
- Dale Laushman from The Uptime Group
- Rainer schoepf from Proteosys.com
- Yann Ormanns
and all others who help me to build a useful and reliable product.
LICENSE
Copyright (c) 2005-2015 Gilles Darold - All rights reserved.
Some code is Copyright (C) 2004-2008 Christos Tsantilas
This program is free software: you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation, either version 3 of the License, or any later
version.
This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program. If not, see < http://www.gnu.org/licenses/ >.
perl v5.14.2 2015-05-11 SQUIDCLAMAV(1)