Script for getting webhits from rotated logfiles

[moojoo]

Hi there,

i have an apache webserver with a number of virtual hosts. The logfiles for each virtual host are rotated by cronolog daily. This is a problem when i want to see the webhits in a cacti graph because there are many files and the filenames are different from day to day.

The following script takes a logfile-pattern (the same as defined in cronolog) and returns the number of hits in all logfiles that match this pattern.

Code: Select all

#!/usr/bin/perl -w
#
# webhits_rot.pl - count apache webhits on cronolog-rotated logfiles
# Version 0.2 - by Achim Schrepfer <a.schrepfer@oncom.de>
#
# Changelog:
# 24.04.03 (as):        - removed the counterfile as it is not needed for use with rrdtool/cacti
#
# Note:
# The only disadvantage of this script is the fact, that the counter jumps to zero at midnight or
# whenever the logfiles are rotated. Rrdtool has an option to set a maximum value for the counter so
# it can correct the error when the counter jumps to zero (i.e. the delta is negative). The problem is, that
# there is *no* maximum value for the webhits counter.

use strict;

# check commandline-arguments
if (!$ARGV[0]) { exit 1; }
my $logfile_pattern = $ARGV[0];

# preparings
my ($day, $month, $year) = (localtime)[3,4,5];
$month ++;
$year += 1900;
$day = sprintf '%02d', $day;            # 2-digit zero-leading day (%d)
$month = sprintf '%02d', $month;        # 2-digit zero-leading month (%m)
$year = sprintf '%04d', $year;          # 4-digit year (%Y)

# replace placeholders with time-informations
$logfile_pattern =~ s/%d/$day/go;
$logfile_pattern =~ s/%m/$month/go;
$logfile_pattern =~ s/%Y/$year/go;

# count webhits for logfiles that match the expanded $logfile_pattern
my $todays_hits = `cat $logfile_pattern | wc -l`;
$todays_hits =~ s/^.*?(\d+).*$/$1/g;
chomp($todays_hits);

# printout todays_hits
print $todays_hits;

As you can see, the mechanism is very simple. There are not all placeholders implemented that cronolog understands but it shall be easy to put the missing ones in this script.

In cacti i call the script like this to gather the data:

Code: Select all

./webhits_rot.pl /var/log/apache/cronolog/%Y/%m/%Y%m%d-*-access_log

Hope it's useful for you...
moojoo[/code]