[c-nsp] vs Cisco 6800 with sup2t and HP NodeManager

[Phil Mayers]

On 31/08/15 14:52, Arne Larsen  / Region Nordjylland wrote:
> Hi all
> This might be a bit off topic, but I'll try anyway.
> Has anyone experienced something like this.
> I'll installed a new HP NodeManager 10. and every time the NNM does a rediscover,
> then snmp demon off 6800 vith sup 2T stops answering  for up to 30 min .

Either it's a bug or just some very slow MIBs (some tables end up being 
really slow because the requirement to sort the table by OID when doing 
a getnext, which might potentially happen on every request)

If it's a bug, then TAC is the way to go.

If it's a subset of slow MIBs you'd need to look at what SNMP queries it 
was making to know which ones are slow - packet capture on the NNM 
easiest - but once you do you could define a mib-view and exclude those 
MIBs.

Likely candidates - anything to do with the flash/storage, 
sensors/hardwre, idprom/serial numbers, or any of the 
ipRouteEightHundredColumns MIBs...

If those MIBs are needed by NNM, then you're stuck. Make sure it's using 
SNMPv2/v3 and doing getbulk as opposed to getnext, this may help.

But in either case, slow or bug, I'd first determine which queries, to 
which MIBs, are slowing it down - I've done a full:

snmpwalk router .1

...on a 6880 and sup2t several times, and not seen this, so it is likely 
something non-obvious.

Cheers,
Phil