block drop in log allrule which drops and logs all un-handled traffic. I’ve been rotating the logs around, but not doing anything more than troubleshooting with the logs. I often watch the live pflog scroll by, investigating the occasional IP of interest.
I run an OpenBSD system as a packet filter in front of my various virtual machines at my colo. I’ve got a default
block drop in log all rule which drops and logs all un-handled traffic. I’ve been rotating the logs around, but not doing anything more than troubleshooting with the logs. I often watch the live pflog scroll by, investigating the occasional IP of interest.
This often involves doing a whois on the IP space, as well as a GeoIP lookup. This can be repetitive, and sometimes after a few hours worth of copy/paste/lookup/etc I feel like a robot. So I set out to gather some stats on my firewall logs. I googled for “openbsd pflog stats”, which gave me the OpenBSD PF FAQ on logging, and then an article about Monitoring PF which suggested using fwanalog. I tried fwanalog out, and was unhappy with the complexity, and the lack of output results. Things just seemed broken after sitting abandoned for so long.
I went back to my Google search, and on the 5th hit, found a nice Perl script, Pantz PFlog Stats, which had simple, but informative output. This script had also sat for a while, and hadn’t been updated in years, however it ran without error, and immediately produced a nice summary of blocked packets. The code was well documented, and the script was easily extensible. I’ve hacked on many Perl scripts before, I actually started code hacking on Perl.
I set out with a few goals in mind:
I have accomplished both goals, and have put the result of my hacking up on Github, at github.com/tbaschak/Pantz-PFlog-Stats.
You can see some sample output generated at the time this blog was published here.