A ZenPack to provide support for Varnish version 3.x metrics. It introduces a new command parser that is intended to be run over SSH. Metrics are parsed from the output of the varnishstat command on the Varnish 3.x server

The ZenPack has the following:

• A new command parser
  • ZenPacks.community.Varnish3.parsers.VarnishStat
• A new monitoring template
  • Varnish3 in /Server/Linux

• Tested on Zenoss Version(s): 3.2.1
• External Dependencies: SSH Access to Varnish Server
• ZenPack Dependencies: None

Download the appropriate package for your Zenoss version from the list below.

• Zenoss 3.0+ Latest Package for Python 2.6
• Zenoss 4.1+ Latest Package for Python 2.7

Copy the downloaded .egg to your Zenoss server and run the following commands as the zenoss user:

zenpack --install <package.egg>
zenoss restart

If you wish to further develop and possibly contribute back to the Varnish3 ZenPack you should clone the git repository, then install the ZenPack in developer mode using the following commands:

git clone git://github.com/ZCA/ZenPacks.community.Varnish3.git
zenpack --link --install ZenPacks.community.Varnish3
zenoss restart

• Ensure SSH access from your Zenoss server to your Varnish server is setup and working properly.
• Ensure the SSH user you are using has access to run the varnishstat command
• The provided monitoring template is not automatically bound to any specific device class. You will need to either create a new device class and bind the template to this device class, or bind the template directly to your Varnish 3.x servers in an existing device class.

Varnish 3.x outputs a large quantity of statistics. Different stats are going to be valuable to different people. The provided template is simply a starting point. I tried to pick stats I thought would have common interest.

The command parser will support any of the stats that varnishstat will output however you'll need to create a new datapoint that matches the name of the stat you are interested in if its not already configured. Running varnishstat -x on your Varnish 3.x server will produce XML output containing all the stats. The <name>...</name> attribute of a stat would be the value to use when creating the new datapoint.

When adding datapoints, be sure to add them to the existing datasource. There is no need to create a second datasource. Keeping a single datasource will keep the number of SSH calls needed to a minimum.

When dumping the varnishstat output via XML, some stats will actually have the same value in their <name> node. Since each datapoint needs to have a unique name the command parser will do some alerations to ensure uniqueness. The parser will prepend the value of the <ident> node followed by a "-". Lets look at a real world example:

<stat>
   <type>SMA</type>
   <ident>s0</ident>
   <name>g_bytes</name>
   <value>1973411487</value>
   <flag>i</flag>
   <description>Bytes outstanding</description>
</stat>
<stat>
   <type>SMA</type>
   <ident>Transient</ident>
   <name>g_bytes</name>
   <value>297633</value>
   <flag>i</flag>
   <description>Bytes outstanding</description>
</stat>

In this example the command parser will produce the name "s0-g_bytes" for the first stat and "Transient-g_bytes" for the second one. Its important to understand this if you intend to add any additional datapoints to the template.

• 1.0
  • Initial Release
• 1.1
  • Some new datapoints and graphs on monitoring template
  • Added logic into the parser to handle conditions when multiple stats share the same name value
  • Updated install method to work around Trac Ticket 7551 and enforce an rrdmin of 0 on all DERIVE datapoints in the provided template.
  • Added varnishd OSProcess
• 1.2
  • Added varnishncsa and varnishlog processes: #3
  • Updated storage graph to be a little less confusing.
  • Fixed Hit Graph so it better aligns with varnishstat output: #2