Page 2 of 2
Re: Could not stat() check result file
Posted: Mon Aug 05, 2013 2:49 pm
by abrist
As long as the defunct clear out, it is WAI. Are you using a broker? (like mklivestatus)
Re: Could not stat() check result file
Posted: Mon Aug 05, 2013 3:01 pm
by paulds
Hm, I don't think so. Our nagios.cfg contains "event_broker_options=-1", but we don't have any broker_module specified.
Re: Could not stat() check result file
Posted: Mon Aug 05, 2013 4:20 pm
by abrist
What is the memory usage on this server? Is it swapping? You may want to see if the stat() errors are concurrent with nagios restarts. The init script / restart of the nagios process can (under the right circumstances) remove the lock file before all operations are complete, this may lead to these errors.
Re: Could not stat() check result file
Posted: Tue Aug 06, 2013 8:11 am
by paulds
The system is not swapping. I've looked back at our resource utilization logs, and at the time of all of the recent warnings, there was zero swap utilization, and roughly 3 - 3.5 GB of RAM (out of 8GB total) was in use. Also, FWIW, the system load has been pretty consistently around 1.00 (out of 8.00 on this 8-core system).
The warning messages do not coincide with nagios restarts.
All good theories, but alas. :)
Re: Could not stat() check result file
Posted: Tue Aug 06, 2013 12:32 pm
by abrist
Is there any other odd errors/warnings in the logs near the time of each of these errors? I am running out of ideas as to the cause - any other details you can provide?
Re: Could not stat() check result file
Posted: Tue Aug 06, 2013 1:32 pm
by paulds
Nothing else in either nagios.log or the system logs that shows up around the same times as these warnings.
I'm going to try using the audit subsystem to log all accesses of the checkresults directory, and then watch the system logs for another occurrence of this warning, and then see if I can correlate anything with that.
Re: Could not stat() check result file
Posted: Tue Aug 06, 2013 1:43 pm
by abrist
That sounds great. I will leave the thread open. Respond back when more info has been collected.