SNMP Performance with Brocade DCX Chassis

[Ben_Sova]

Greetings!!

I’m looking for some guidance on how to handle performance and checking for SNMP on devices contain A LOT of information.

In my example I have several Brocade DCX Chassis with FC Blades in them. I am able to add them into CheckMK and the checks do work however because I have so many Ports/Blades per DCX I am running into issues with either “Stale Checks”, “Service Connection Timeouts”, or the Checks are “unknown/vanished” since they no longer appear on the latest walk. Roughly each DCX has anywhere from 400 - 1200 checks depending on what’s plugged in.

This isn’t a constant issue but happens enough to make monitoring the Brocades though CheckMK not viable. Looking for some guidance on the forums to see what can be done to help increase performance and/or reduce the issues with Stale Checks, Service Connection Timeouts, and Unknown/Vanished checks.

Thanks!
Ben

[andreas-doehler (Andreas)]

I don’t know a general solution for this problem. The biggest problem are the interface checks or?
For a test, if you switch of the interface checks it is working without problem i think.

[Ben_Sova]

I guess you could say it’s the entire SNMP walk and gathering info for the checks in a timely fashion. Once the SNMP data is there the checks are fine and report what they should. I’ve noticed when I do a “snmpbulkwalk” or “snmpwalk” against the device, it does take a while to generate all the info.

I could test out a Brocade DCX and just disabling all the interface checks and see if that makes a difference.