Dear Tom,

I trust your expertise and knowledge about this solution. However, I’m just stating a fact here.

If I disable “unwanted” modules from being polled, total polling time per device drops. If I re-enable it, it rizes again. This is something that is easily reproductible.

I’ll try to look for options to upping SNMP timeout but I’m not confident about the existence of such an option. Otherwise, I’ll just disable modules that I don’t need on a per-device basis.

Best regards.

De : observium <observium-bounces@observium.org> au nom de Tom Laermans <tom.laermans@powersource.cx>
Répondre à : Observium <observium@observium.org>
Date : dimanche 20 août 2017 à 12:07
À : "observium@observium.org" <observium@observium.org>
Objet : Re: [Observium] Important polling time increase since upgrade to r8697

Almost none of those will actually do anything on that device.

Brocade seems to have gone for "let's make one problem 2 problems" with their silly approach.

I can only recommend upping the SNMP timeout for those devices to ungodly high numbers to avoid net-snmp requerying things making the queue larger.

Tom

On 20/08/2017 11:40, Youssef BENGELLOUN - ZAHR wrote:

Hi Adam,

That very likely, it’s a known fact that NetIron SNMP stack quality is… debatable, let’s put it this way ;-)

May I suggest to disable by default polling of modules that have nothing to do with a router/switch role ? For instance, are those really necessary : hr-mib / ucd-mib / printersupplies / ucd-diskio / wifi / p2p-radios / loadbalancer ?