I have a very high load on zenhub which results in many heartbeat failures as
the different services are not able to connect to zenhub instantly but have to
wait for "a free slot" (don't get the "pbnone" string back when connecting to
zenhub)
unfortunately i currently run zenoss on a two core machine, thus can't increase
the workers...
in my case the high load is due to the ping process which builds the pingTree
to try do create dependencies among the devices... (i assume many people with
not sooo strong hardware have the same issue)
Code:
grep "Time in" zenhub.log | awk '{print $8, $7}' | sort -n
...
...
319.43 getPingTree:
328.78 getPingTree:
333.84 getPingTree:
337.13 getPingTree:
338.35 getPingTree:
338.73 getPingTree:
339.31 getPingTree:
even worse, it's not possible to turn off creating this dependency tree even if
it's not needed. I saw that there is already an open ticket for this issue.
note: zenhub has to be run with logseverity 10 in the config file to see where
the time is spent!
-------------------- m2f --------------------
Read this topic online here:
http://forums.zenoss.com/viewtopic.php?p=35558#35558
-------------------- m2f --------------------
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
