I realized recently that I had a bunch of newly-provisioned VMs that weren’t being monitored by MRTG (one of the tools we use to monitor network usage and other fun stats). Rather than manually run cfgmaker against all the new machines, I decided to script my way out of this.

Step 1: Build a list of hosts to monitor

For this I used an nmap ping test to generate a list of “up” hosts:

nmap -sP 10.0.0.0/24 -oG 10.0.0.0-255.255.255.0

This creates a list of IPs that are up in a format like this:


Host: 10.0.0.60 () Status: Up
Host: 10.0.0.61 () Status: Up
Host: 10.0.0.63 () Status: Up
Host: 10.0.0.64 () Status: Up
Host: 10.0.0.65 () Status: Up


Step 2: Parse out the IP address and run cfgmaker

Here’s the command I ran:

$ for i in `perl -ne 'chomp; next if $_ =~ /^#/; my @a = split(/ /); print "$a[1]\n";' 10.0.0.0-255.255.255.0`; do /usr/bin/cfgmaker --global 'LogFormat: rrdtool' --global 'WorkDir: /mrtg/data/mrtg' --global 'Options[_]: growright,bits' --global 'XSize[_]: 600' --global 'YSize[_]: 400' public@$i > /etc/mrtg/mrtg-$i ; done

This parses out the IP address from the input file and for each address, runs cfgmaker and puts the output in /etc/mrtg/mrtg-[ipaddress] .

Taking this a step further, I used gethostbyaddr to resolve the IPs to hostnames, so the MRTG files have the proper hostnames in them:

$ for i in `perl -MSocket -ne 'chomp; next if $_ =~ /^#/; my @a = split(/ /); my $ip = inet_aton($a[1]); my $host = gethostbyaddr($ip, AF_INET); print "$host\n";' 10.0.0.0-255.255.255.0`; do /usr/bin/cfgmaker --global 'LogFormat: rrdtool' --global 'WorkDir: /mrtg/data/mrtg' --global 'Options[_]: growright,bits' --global 'XSize[_]: 600' --global 'YSize[_]: 400' public@$i > /etc/mrtg/mrtg-$i ; done

Then all of the mrtg .cfg config files are in /etc/mrtg/, but who wants to put in all those cron entries to run each one? Not me. Simple fix?

cat /etc/mrtg/*.cfg > /etc/mrtg-all.cfg

Then you only have one cron entry.