Monitoring multiple Varnish instances

Written by Claudio Kuenzler - 0 comments

Published on November 14th 2017 - Listed in Varnish Linux Monitoring Icinga Nagios

To monitor Varnish performance I've been using check_varnish.py for quite some time now. It uses the varnishadm command in the background to get all kinds of values; for example number of hits, number of misses, number of requests etc.

Varnish also allows to run in parallel of other Varnish processes, as long as:

- The work directory is different. This is handled by assigning a different work directory name (-n name) to the process. This setting is better known as "instance name", although work directory would be the technical correct naming.

- The listener ports are different. This applies to both the http listener (-a) and the management listener (-T).

The problem with check_varnish.py? It doesn't support the -n parameter to query a certain Varnish process. Or better said: It didn't support the -n parameter. I modified the plugin and created a pull request for the upstream/original plugin.

With the modifications, the plugin check_varnish.py, is now able to monitor multiple Varnish processes/instances. And it stays backward compatible to single Varnish processes launched without -n parameter:

# ./check_varnish.py -f MAIN.cache_miss -n varnish-test
VARNISH OK - MAIN.cache_miss is 683744
| 'MAIN.cache_miss'=683744