canvon IRC bot for monitoring Nagios monitoring

Send Nagios log messages to IRC, and let Nagios status be queried actively from IRC

This is an IRC (Internet Relay Chat) bot that parses the Nagios log file on the central IT infrastructure monitoring host and passes "interesting" things on to IRC. At the moment, that means it will send host/service alerts as notices, and notifications as normal messages addressed to the Nagios CONTACTNAME (which presumably is the same as a nick on IRC, otherwise just set a highlight in your IRC client).

The bot also can be talked to, which gives direct access to the Nagios status file, currently via Icinga icli invocations (which is compatible to Nagios). Addressing the bot and saying "problems" to it will dump the current state of services with problems, either to the channel or query where it was addressed. There are further commands, address the bot saying "help" to get a short overview.

This README contains: (Table of Contents)

• Setup instructions

  • Basic setup instructions

  • Bot commands and how to make them work

  • Log rotation and how to work-around it

• Thanks

• Contact information

You can run the bot right from the git repository, as ./cvnircbot.pl; but you have to copy the example-cvnircbotrc to ~/.cvnircbot/cvnircbotrc first, and edit the file to set the IRC server to use, bot nickname, channel and other things.

Be sure to set up your Nagios log file properly (undef will do, and default to where it's stored on Debian -- otherwise, you probably want to set an absolute path, but it can be any Perl expression). The default currently is /var/log/nagios3/nagios.log.

Also be sure to set the nagios_channels to some of your bot channels so you will be receiving the Nagios messages as intended.

If you like rather having output of what the software is doing than silence, leave the debug definition in; otherwise, comment it out.

To make the active queries from IRC work, you'll also have to set up an icli that can be invoked by the bot to get data. In my case, this is a wrapper script in ~/bin/icli (it has to be on your $PATH) which uses sudo to aquire read permissions on the Nagios status file. Often the group the web server runs in will have such read permission. So I use (on Debian 8):

#!/bin/bash
exec sudo -g www-data icli --config /etc/nagios3/nagios.cfg --status-file /var/cache/nagios3/status.dat "$@"

Save the two-liner to ~/bin/icli and chmod +x ~/bin/icli. Giving password-less sudo permissions for group www-data (or what the webserver your Nagios web-frontend is running on uses) to the user the bot runs as is left as an exercise to the reader.

The bot currently does not cope with Nagios' automatic log rotation.

As a work-around, I recommend running the bot in a loop, like this:

nagios-cvnircbot$ while true ; do ./cvnircbot.pl ; sleep 600 ; done  # restart every 10 minutes

Then make it exit regularly from cron. In a per-user crontab, using crontab -e to edit and crontab -l to list:

# m h  dom mon dow   command
2 0  1 * *   pkill cvnircbot.pl

This would make the bot exit on the 1st of each month, two minutes past mid-night. (This is just after my Nagios instance rotates its logs.) It will come back automatically after the timeout you set on the command line above expires.

Thanks go to ArneB http://www.arneb.de/, for explaining how non-blocking I/O might be used to poll a log file from a Perl IRC bot.

The nagios-cvnircbot, canvon IRC bot for monitoring Nagios monitoring, has been written by:

• Fabian Pietsch [email protected] (primary author from 2015-2016)

The project is currently (as of 2016-08-29) hosted on GitHub:

• https://github.com/canvon/nagios-cvnircbot