Customizing Logwatch
There are two basic customizations we often perform:
- increase the header-information with machine-specific info (hardware, admininstrator(s), usage, etc):
- adjust the order of the information - we care very much about disk-usage, and prefer it first
Start by ensuring logwatch is actually installed! A mail-transport agent (MTA) like Postfix or nullmailer must be installed and working (to receive the daily logwatch summary).
Logwatch Headers
Use this template, to create /usr/share/logwatch/custom_header (ASCII, plain-text file; previous Gentoo and other distributions used /etc/log.d/custom_header).
Machine info: Pentium4 2.4GHz, 1GB, GigE, 40GB+120GB, RivaTNT Running RedHat 9 Linux Located in Podium shared server-room, SFU Surrey campus Used by John Doe <john_doe@sfu.ca> Into service xxx. Re-purposed Dec. 15, 2004 Administered by Gordon Pritchard <gordonp+corsair@sfu.ca>
Another example:
Machine info: Pentium4 dual-core 3GHz, 2GB, GigE, 80GB, ATI-R300-mobility Running Gentoo Linux Located in office 525 / The Axis (Beta Space, Surrey Campus) Used by Gordon Pritchard <gordonp+zero@sfu.ca) Into service September 6, 2005. Administered by Gordon Pritchard <gordonp+zero@sfu.ca>
To include this custom header, we now have to modify the main logwatch Perl script logwatch.pl to add the bold / red line:
- on a Gentoo machine, this file is /usr/sbin/logwatch.pl:
output( $index_par, "\n ################### Logwatch $Version ($VDate) #################### \n", "line");
}
output( $index_par, " Processing Initiated: " . localtime(time) . "\n", "line");
output( $index_par, " Date Range Processed: $Config{'range'}\n", "line");
output( $index_par, " $print_range\n", "line") if ($Config{'range'} ne 'all');
output( $index_par, " Period is " . GetPeriod() . ".\n", "line")
if ($Config{'range'} ne 'all');
output( $index_par, " Detail Level of Output: $Config{'detail'}\n", "line");
output( $index_par, " Type of Output: $Config{'output'}\n", "line");
output( $index_par, "\n" . `/bin/cat $BaseDir/custom_header` . "\n", "line");
output( $index_par, " `/bin/cat $BaseDir/custom_header` . "\n", "line");
if ( $outtype_html ) {
output( $index_par, "\n", "stop");
} else {
output( $index_par, "################################################################## \n", "line");
- on a RedHat 9 machine, this file is /etc/log.d/scripts/logwatch.pl:
$printing = 'y';
print OUTFILE "\n ################### LogWatch $Version ($VDate) #################### \n";
print OUTFILE " Processing Initiated: " . localtime(time) . "\n";
print OUTFILE " Date Range Processed: $Config{'range'}\n";
print OUTFILE " Detail Level of Output: $Config{'detail'}\n";
print OUTFILE " Logfiles for Host: $Config{'hostname'}\n";
print OUTFILE "\n" . `/bin/cat $BaseDir/custom_header` . "\n";
print OUTFILE "################################################################ \n\n";
}
Lately, it's been more-difficult to remember which machines have up-to-date kernels, so we want logwatch to assist us, by printing the kernel-version in the header. Using a Gentoo example, it's a one-line addition:
output( $index_par, "\n ################### Logwatch $Version ($VDate) #################### \n", "line");
}
output( $index_par, " Processing Initiated: " . localtime(time) . "\n", "line");
output( $index_par, " Date Range Processed: $Config{'range'}\n", "line");
output( $index_par, " $print_range\n", "line") if ($Config{'range'} ne 'all');
output( $index_par, " Period is " . GetPeriod() . ".\n", "line")
if ($Config{'range'} ne 'all');
output( $index_par, " Detail Level of Output: $Config{'detail'}\n", "line");
output( $index_par, " Type of Output: $Config{'output'}\n", "line");
output( $index_par, " Logfiles for Host: $Config{'hostname'}\n", "line");
output( $index_par, " Kernel: " . `/bin/uname -r` . "\n", "line");
output( $index_par, "\n" . `/bin/cat $BaseDir/custom_header` . "\n", "line");
Logwatch Information Order
The scripts located in /etc/log.d/scripts/services/ are run according to their file-name ordering. Simply renaming a file changes it's relationship to other information. We should move zz-disk_space to aa-disk_space to have our disk-usage appear right after the custom header; as root:
zero services # mv /usr/share/logwatch/scripts/services/zz-disk_space /usr/share/logwatch/scripts/services/aa-disk_space
The following is no longer necessary for current Gentoo systems; they *always* print the disk-usage
But wait - there's more!!
This disk-space script evolves through the various releases, often suppressing output if nothing prior is printed. And, (now) as the first entry, nothing prior will ever get printed :-O So, we'll strip out all the conditionals, and produce a pared-down script /etc/log.d/scripts/services/aa-disk_space which forever more will always give us our disk-usage as the first thing:
Gentoo:
#!/bin/bash
##########################################################################
# $Id: aa-disk_space,v 1.8 2005/02/24 17:08:05 kirk Exp $
##########################################################################
echo
df -hP | head -n1; df -hP | grep '^[/ ]'
RedHat 9:
#!/bin/bash ########################################################################## # $Id: disk_space,v 1.1 2003/01/13 04:00:59 kirk Exp $ ########################################################################## echo echo echo "------------------ Disk Space --------------------" echo df -h echo
One more thing to do - there is a corresponding configuration file which we simply move (rename) to align with the above:
- Gentoo shown; earlier Gentoo and RedHat 9 had these files at /etc/log.d/conf/services/zz-disk_space.conf and /etc/log.d/scripts/services/zz-disk_space
zero # mv /usr/share/logwatch/default.conf/services/zz-disk_space.conf /usr/share/logwatch/default.conf/services/aa-disk_space.conf
To test it all, invoke the daily logwatch cron-job manually (Gentoo shown; others similar):
zero # sh /etc/cron.daily/00-logwatch