Gaps in Performance graphs

[abrist]

Yeah, 15 or more is also my suggestion. Don't forget to restart npcd. You may or may not have retained the perfdata for those services. Change the timeout, restart npcd, and we will see if they get processed. Judging by the fact that you have perf data after the missing chunk, you may not be able to recover the lost perf data.

[benhank]

Ben here. I said fugg it, and bumped it to 25. that's right 25 that's how I roll up in this mofo!

How long will it take to know if the change was successful?

[abrist]

You should stop seeing *new* timeout errors immediately.

[slansing]

And between 5-15 minutes later you should have enough graphed data to verify you have no random crevasses, cliffs into bottomless pits, or abysses.

[benhank]

I changed the Timeout for perfdata in the Nagios.cfg file and restarted NPCD. We were still getting timeouts after 5 secs so we restarted Nagios.
We are still seeing timeouts after 5 secs. Is there somewhere else we need to modify it?

Thanks,

[benhank]

What is the difference between the 'perfdata timeout' setting found in the Nagios.cfg file and the setting in the process_perfdata.cfg

Thanks,

[abrist]

From: http://nagios.sourceforge.net/docs/3_0/configmain.html

Format: perfdata_timeout=<seconds>
Example: perfdata_timeout=5

This is the maximum number of seconds that Nagios will allow a host performance data processor command or service performance data processor command to be run. If a command exceeds this time limit it will be killed and a warning will be logged.

But this is a misnomer, as we configure core/xi to use a custom script, so this value is ignored.

[benhank]

My boss says:

He changed the timeout in the perfdata services config file. and the issue went poof. you can lock it up.

[slansing]

Yes that would be the place to change it! Thanks for letting us know, locking as resolved.