Inventory Polling and Collection causes alarms in HPOV

mwynen · ‎09-04-2008

Hy experts.

Evertime LMS 2.6 is running the Inventory Collection or Inventory Polling jobs, our HPOV shows lots of devices that doesn't seem to answer correctly to smnp requests.

This is a bit scary, because everytime these jobs are running it is looking like an extensive outage for the guys at the support desk.

Is this a normal behavior? Does the collection or polling challenge the device so much?

How can I fix this?

best regards

Marc

Joe Clarke · ‎09-04-2008

Inventory collection uses SNMP read-only to build the hardware inventory of a device. The objects queried depend on the device type and OS version. Typically, inventory collection is not an intensive process, but some versions of code on some devices do not respond quickly enough to certain branches (e.g. the CISCO-FLASH-MIB branch) and that may trigger CPUHOG messages.

I don't know what messages you're seeing in HPOV, but some might be cleared by applying SNMP views for certain branches (although this may break inventory collection). Some may also be cleared by upgrading the software on your devices.

g.meerkoetter · ‎09-04-2008

I've seen similar effects on devices with slow CPUs (e.g. cat5500, cat2980). These devices had

a cpu usage of about 20% even when not being polled by LMS or HPOV.

Reducing the number of VLANs configured on the trunks helped, since the CPUs spent a lot of cycles processing multicast traffic in VLANs we did not need on the switch.