Reading the wrong documentation

Created: 25 Apr 2013

Given that everyone likes to learn from the mistakes of others indulge in a little schadenfreude, here’s something amusing that I did a few weeks ago.

I wanted to increase the frequency with which puppet agent ran, so looking at the first official documentation page returned by a google search I read this:

runinterval

How often puppet agent applies the client configuration; in seconds. Note that a runinterval of 0 means “run continuously” rather than “never run.” If you want puppet agent to never run, you should start it with the –no-client option. Can be specified as a duration.

Default: 30m

I shipped a configuration change with runinterval = 15m and worked on something else, while it rolled out.

A short time later I noticed signficant load on the puppet masters and also that every single node seemed to be repeatedly asking for a puppet catalog. Anyone might think that someone had accidentally configured puppet agents to check in every 15 seconds.

Here is the documentation I should have read, for the version of puppet that was installed:

runinterval

How often puppet agent applies the client configuration; in seconds. Note that a runinterval of 0 means “run continuously” rather than “never run.” If you want puppet agent to never run, you should start it with the –no-client option.

Default: 1800

The circumstances that collided here were:

• it’s easier to find the documentation for the most recent version of puppet than the specific version we are using
• the puppet 2.7 configuration parser adopts Postel’s principle of robustness
• I routinely make an efficiency-thoroughness tradeoff which saves me a few seconds every time I consult the puppet documenation until it fails spectacularly.

It was also very funny - after I’d managed to fix it.