NAME
SquidAnalyzer - Squid access log report generation tool
DESCRIPTION
SquidAnalyzer parse native access log format of the Squid proxy and
generate general statistics about hits, bytes, users, networks, top url
and top second level domain.
Statistic reports are oriented to user and bandwidth control, this is
not a pure cache statistics generator. SquidAnalyzer use flat files to
store data and don't need any SQL, SQL Lite or Berkeley databases.
This analyzer is incremental so it should be run in a daily cron. Take
care if you have rotate log enable to run it before rotation is done.
REQUIREMENT
Nothing is required than a modern perl version 5.8 or higher. Graphics
are based on the Flotr2 Javascript library so they are drawn at your
browser side without extra installation required.
INSTALLATION
Generic install
If you want the package to be intalled into the Perl distribution just
do the following:
perl Makefile.PL
make
make install
Follow the instruction given at the end of install. With this default
install everything configurable will be installed under
/etc/squidanalyzer. The Perl library SquidAnalyzer.pm will be installed
under your site_perl directory and the squid-analyzer Perl script will
be copied under /usr/local/bin.
The default output directory for html reports will be
/var/www/squidanalyzer/.
On FreeBSD, if make install is freezing and you have the following
messages:
FreeBSD: Registering installation in the package database
FreeBSD: Cannot determine short module description
FreeBSD: Cannot determine module description
please proceed as follow:
perl Makefile.PL INSTALLDIRS=site
make
make install
as the issue is related to an install into the default Perl vendor
installdirs it will then use Perl site installdirs.
Custom install
You can create your fully customized SquidAnalyzer installation by using
the Makefile.PL Perl script. Here is a sample:
perl Makefile.PL \
LOGFILE=/var/log/squid3/access.log \
BINDIR=/usr/bin \
CONFDIR=/etc \
HTMLDIR=/var/www/squidreport \
BASEURL=/squidreport \
MANDIR=/usr/man/man3 \
DOCDIR=/usr/share/doc/squidanalyzer
If you want to build a distro package, there are two other options that
you may use. The QUIET option is to tell to Makefile.PL to not show the
default post install README. The DESTDIR is to create and install all
files in a package build base directory. For example for Fedora RPM,
thing may look like that:
# Make Perl and SendmailAnalyzer distrib files
%{__perl} Makefile.PL \
INSTALLDIRS=vendor \
QUIET=1 \
LOGFILE=/var/log/squid/access.log \
BINDIR=%{_bindir} \
CONFDIR=%{_sysconfdir} \
BASEDIR=%{_localstatedir}/lib/%{uname} \
HTMLDIR=%{webdir} \
MANDIR=%{_mandir}/man3 \
DOCDIR=%{_docdir}/%{uname}-%{version} \
DESTDIR=%{buildroot} < /dev/null
See spec file in packaging/RPM for full RPM build script.
Local install
You can also have a custom installation. Just copy the SquidAnalyzer.pm
and the squid-analyzer perl script into a directory, copy and modify the
configuration file and run the script from here with the -c option.
Then copy files sorttable.js, squidanalyzer.css and
logo-squidanalyzer.png into the output directory.
Post installation
1. Modify your httpd.conf to allow access to HTML output like follow:
Alias /squidreport /var/www/squidanalyzer
<Directory /var/www/squidanalyzer>
Options -Indexes FollowSymLinks MultiViews
AllowOverride None
Order deny,allow
Deny from all
Allow from 127.0.0.1
</Directory>
2. If necessary, give additional host access to SquidAnalyzer in
httpd.conf. Restart and ensure that httpd is running.
3. Browse to http://my.host.dom/squidreport/ to ensure that things are
working properly.
4. Setup a cronjob to run squid-analyzer daily or more often:
# SquidAnalyzer log reporting daily
0 2 * * * /usr/local/bin/squid-analyzer > /dev/null 2>&1
or run it manually. For more information, see README file.
If your squid logfiles are rotated then cron isn't going to give the
expected result as there exists a time between when the cron is run and
the logfiles are rotated. It would be better to call squid-analyzer from
logrotate, eg:
prerotate
/usr/local/bin/squid-analyzer
endscript
You can use network name instead of network ip addresses by using the
network-aliases file. Also if you don't have authentication enable and
want to replace client ip addresses by some know user or computer you
can use the user-aliases file to do so.
See the file squidanalyzer.conf to customized your output statistics and
match your network and file system configuration.
USAGE
SquidAnalyzer can be run manually or by cron job using the
squid-analyzer Perl script. Here are authorized usage:
Usage: squid-analyzer [ -c squidanalyzer.conf ] [-l logfile]
-c | --configfile filename : path to the SquidAnalyzer configuration file.
By default: /etc/squidanalyzer.conf
-b | --build_date date : set the day to be rebuilt, format: yyyy-mm-dd,
yyyy-mm or yyyy. Used with -r or --rebuild.
-d | --debug : show debug informations.
-h | --help : show this message and exit.
-l | --logfile filename : path to the Squid logfile to parse.
By default: /var/log/squid/access.log
-p | --preserve number : used to set the statistic obsolescence in
number of month. Older stats will be removed.
-r | --rebuild : use this option to rebuild all html and graphs
output from all data files.
-v | version : show version and exit.
--no-year-stat : disable years statistics, reports will
start from month level only.
There is special options like --rebuild that force SquidAnalyzer to
rebuild all HTML reports, useful after an new feature or a bug fix. If
you want to limit the rebuild to a single day, a single month or year,
you can use the --build_date option by specifying the date part to
rebuild, format: yyyy-mm-dd, yyyy-mm or yyyy.
The --preserve option should be used if you want to rotate your
statistics and data. The value is the number of months to keep, older
reports and data will be removed from the filesystem. Useful to preserve
space, for example:
squid-analyzer -p 6 -c /etc/squidanalyzer/squidanalyzer.conf
will only preserve six month of statistics from the last run of
squidanalyzer.
CONFIGURATION
Unless previous version customization of SquidAnalyzer is now done by a
single configuration file squidanalyzer.conf.
Here follow the configuration directives used by Squid Analyzer.
Output output_directory
Where SquidAnalyzer should dump all HTML, data and images files. You
should give a path that can be read by a Web browser.
WebUrl
The URL of the SquidAnalyzer javascript, HTML and images files.
Default: /squidreport
LogFile squid_access_log_file
Set the path to the Squid log file.
UseClientDNSName 0
If you want to use DNS name instead of client Ip address as username
enable this directive. When you don't have authentication, the
username is set to the client ip address, this allow you to use the
DNS name instead. Note that you must have a working DNS resolution
and that it can really slow down the generation of reports.
DNSLookupTimeout 0.0001
If you have enabled UseClientDNSName and have lot of ip addresses
that do not resolve you may want to increase the DNS lookup timeout.
By default SquidAnalyzer will stop to lookup a DNS name after 0.0001
second (100 ms).
NetworkAlias network-aliases_file
Set path to the file containing network alias name. Network are show
as Ip addresses so if you want to display name instead create a file
with this format:
LOCATION_NAME IP_NETWORK_ADDRESS
Separator must be a tabulation.
You can use regex to match and group some network addresses. See
network-aliases file for examples.
UserAlias user-aliases_file
Set path to the file containing user alias name. If you don't have
auth_proxy enable users are seen as ip addresses. So if you want to
show username or computer name instead, create a file with this
format:
FULL_USERNAME IP_ADDRESS
If you have auth_proxy enable but want to replace login name by full
user name for example, create a file with this format:
FULL_USERNAME LOGIN_NAME
Separator for both must be a tabulation.
You can use regex to match and group some user login or ip
addresses. See user-aliases file for examples.
You can also replace default ip address by his DNS name by enabling
directive 'UseClientDNSName'.
AnonymizeLogin 0
Set this to 1 if you want to anonymize all user login. The username
will be replaced by an unique id that change at each squid-analyzer
run. Default disable.
OrderNetwork bytes|hits|duration
OrderUser bytes|hits|duration
OrderUrl bytes|hits|duration
Used to set how SquidAnalyzer sort Network, User and User detailed
Urls reports screen. Value can be: bytes, hits or duration. Default
is bytes. Note that OrderUrl is limited to User detailed Urls
reports and does not apply to Top Url and Top domain report where
there is three reports each already ordered.
OrderMime bytes|hits
Used to set how SquidAnalyzer sort Mime types report screen Value
can be: bytes or hits. Default is bytes.
UrlReport 0|1
Should SquidAnalyzer display user details. This will show all URL
read by user. Take care to have enougth space disk for large user.
Default is 0, no url detail report.
QuietMode 0|1
Run in quiet mode for batch processing or print debug information.
Default is 0, verbose mode.
CostPrice price/Mb
Used to set a cost of the bandwith per Mb. If you want to generate
invoice per Mb for bandwith traffic this can help you. Value 0 mean
no cost, this is the default value, the "Cost" column is not
displayed
Currency currency_abreviation
Used to set the currency of the bandwith cost. Preferably the html
special character. Default is €
TopNumber number
Used to set the number of top url and second level domain to show.
Default is top 100.
TopUrlUser Use this directive to show the top N users that look at an
URL or a domain. Set it to 0 to disable this feature. Default is top 10.
Exclude exclusion_file
Used to set client ip addresses, network addresses, auth login or
uri to exclude from report.
You can define one by line exclusion by specifying first the type of
the exclusion (USER, CLIENT or URI) and a space separated list of
valid regex.
You can also use the NETWORK type to define network address with
netmask using the CIDR notation: xxx.xxx.xxx.xxx/n
See example bellow:
NETWORK 192.168.1.0/24 10.10.0.0/16
CLIENT 192\.168\.1\.2
CLIENT 10\.169\.1\.\d+ 192\.168\.10\..*
USER myloginstr
USER guestlogin\d+ guestdemo
URI http:\/\/myinternetdomain.dom.*
URI .*\.webmail\.com\/.*\/login\.php.*
you can have multiple line of the same exclusion type.
Include inclusion_file
Used to set client ip addresses, network addresses or auth login to
include into the report. All others will not be included. It works
as the opposite of the Include parameter.
You can define one by line inclusion by specifying first the type of
the inclusion (USER or CLIENT) and a space separated list of valid
regex.
You can also use the NETWORK type to define network address with
netmask using the CIDR notation: xxx.xxx.xxx.xxx/n
See example bellow:
NETWORK 192.168.1.0/24 10.10.0.0/16
CLIENT 192\.168\.1\.2
CLIENT 10\.169\.1\.\d+ 192\.168\.10\..*
USER myloginstr
USER guestlogin\d+ guestdemo
URI http:\/\/myinternetdomain.dom.*
URI .*\.webmail\.com\/.*\/login\.php.*
you can have multiple line of the same inclusion type.
Lang language_file
Used to set the translation file to be used. Value must be set to a
file containing all string translated. See the lang directory for
translation files. Default is defined internally in English.
DateFormat
Date format used to display date (year = %y, month = %m and day =
%d) You can also use %M to replace month by its 3 letters
abbreviation. Default: %y-%m-%d
SiblingHit
Adds peer cache hit (CD_SIBLING_HIT) to be taken has local cache
hit. Enabled by default, you must disabled it if you don't want to
report peer cache hit onto your stats.
TransfertUnit
Allow to change the default unit used to display transfert size.
Default is BYTES, other possible values are KB, MB and GB.
MinPie
Minimum percentage of data in pie's graphs to not be placed in the
others item. Lower values will be summarized into the others item.
Locale
Set this to your locale to display generated date in your language.
Default is to use the current locale of the system. If you want date
in German for example, set it to de_DE.
Rapport genere le mardi 11 decembre 2012, 15:13:09 (UTC+0100).
with a Locale set to fr_FR.
AUTHOR
Gilles DAROLD <gilles@darold.net>
COPYRIGHT
Copyright (c) 2001-2013 Gilles DAROLD
This package is free software and published under the GPL v3 or above
license.