# $Id: README,v 0.2.2.1 2002/06/12 11:59:42 doug Exp $
httpdstats is a perl program that can be used to generate simple statistical summaries of Apache access log files. The program is intended to be used on small, infrequently visited, web servers, where a daily summary of the access log is all that is needed.
Install
The out of the tarball version of httpdstats assumes that perl is in /usr/bin/perl. You may need to modify the first line of the program if perl is elsewhere. httpdstats assumes that the GetOpt::Long module is available.
To make httpdstats, type 'make'.
To install httpdstats, type 'make install'.
The cron job is placed in the /etc/cron.daily directory; ie. anacron is assumed. Instead, you may wish to edit root's crontab.
Using httpdstats with logrotate
The httpdstats cron job assumes that you are using logrotate to rotate the httpd access_log logs every night. This is not the default behavior of logrotate -- which normally rotates weekly or monthly. To fix this, add the line
daily
to the /var/log/httpd/access_log entry in /etc/logrotate.d/apache.
Changes
0.2.2
URLs for both requests and referrers can be presented
without arguments and restricted to a maximum size.
The agent can be restricted to a simple name.
0.2.1
Can now read the standard "combined" log format, as well as the
"common" log format
Two new partitions: referrer and agent
0.2.0
Reports are now done through a roll-your-own description format,
rather than the restricted set allowed before.
Time partitioning of requests added.
Protocol partitioning of requests added.
Host name lookup allowed.
Host names sort in a natural domain order.
Escape sequences in URLs are de-escaped.
Ignore domains can handle IP addresses, as well as host names.
Thanks to Jon Masters <jonathan@jonmasters.org> for pointing out
a bug to do with spaces in URLs.
0.1.1
Added requests-total and bytes-total summaries.
Added help and subject command line switches.
Added ignore-domains option for more accurate reporting.
Config reading command line conflict bug-fix.
All output options default to 'no', rather than the original odd mix.
Thanks to Tim Gurney <tjg@greencathedral.com> for suggestions.