Eric,
Thanks very much. This worked very well! It was actually really easy
to do as well. This should maybe be part of the FAQ or the admin guide.
I now have 2 separate zenperfsnmp daemons running on my Zenoss box and
most of my devices are being polled on a 5-minute interval and only
about 10 of the devices are being polled at a 30 second interval. Saved
ALOT of load average on the Zenoss box too. :)
Thanks again,
-Daniel
Daniel McKinney wrote:
Eric,
Yup. The 64-bit counters have helped quite a bit. It looks like the
1-minute SNMP Cycle Interval with the 64-bit counters will be reliable
enough to create a threshold and alarm off of this data, even with my
high traffic over these links.
Thanks for the guidance for the ability to make a second monitor for
this purpose. If i can get that to work, it will be perfect for my
means.
-Daniel
P.S. Zenoss folks. I never cease to be amazed at what Zenoss can do.
Such a great product. Keep up the good work folks!
Eric Newton wrote:
Hi Daniel,
Hopefully the 64-bit counter discussion was helpful to you as well.
There is one way to do this with the current Zenoss. You can adjust
the collection rate by modifying collector's configuration (Monitors
-> Performance -> localhost). You can create a new configuration
(Monitors -> Performance -> faster) and run a new monitor that will
use that configuration. Then move the devices that need a faster
collect to that configuration.
Last week I posted directions on how to do this for zenping:
http://community.zenoss.com/forums/viewtopic.php?p=4811
You will want to create new instance of zenperfsnmp which collects
most of the SNMP performance data.
-Eric
Daniel McKinney wrote:
Zenoss Devs,
A fix to a big problem i was having with my GigabitEthernet graphs
was to shorten the duration for the Monitor/Performance
Monitors/localhost/Snmp Cycle Interval from 5 minutes, to 1 minute.
Is there a way to change this metric on a device-level basis, or
even a data source-level basis?
I need to be able to poll those GigabitEthernet links really at a
30-second interval to get reliable results, but i definitely don't
need to poll the memory, CPU %, etc... of ALL the rest of my servers
on a 30-second interval.
Is there a way to do this now without having to spin up an entirely
new Zenoss with a shortened cycle time for the critical devices?
If not, is there plans to add this as a feature in future releases
of Zenoss?
Thanks ahead of time,
-Daniel
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
