1/ check the logs
2/ if logs are not enabled, enable the logs
3/ run collectd with -f option to prevent forking. You can find some
problems like this
4/ check if there is a core file. If not, enable it to generate a core
(man ulimit, coreadm...)
5/ get information from the core file (gdb, dbx are debuggers. Get the
stacktrace to know where it cores. With gdb, "bt" is the command)
6/ run Collectd with fewer plugins in order to find out if (and which)
plugin is broken or misconfigured
Step 1 to 5 are generic to any program (step 3 is generit to any daemon)
Step 6 is specific to Collectd.
Regards
Yves
Le 2014-06-24 13:02, Stuart Cracraft a écrit :
I have crashing collectd's daily.
What is the best approach?
Currently I am just restarting them.
_______________________________________________
collectd mailing list
[email protected]
http://mailman.verplant.org/listinfo/collectd
_______________________________________________
collectd mailing list
[email protected]
http://mailman.verplant.org/listinfo/collectd
