This is likely because of interactions between SNMP and the underlying software stack. What higher values of max-rep are doing is telling the SNMP stack it can return more values in a single go, so if the software has a block somewhere in the interaction between the SNMP stack and the metric collection, I can see how increasing that number could make the device take longer to respond, even if there’s nothing to respond to.

 

Some platforms even fail to respond at all if you increase the number, and it’s often sort of arbitrary at what point they stop responding.

 

Adam.

 

From: observium <observium-bounces@observium.org> On Behalf Of Kent Johannessen via observium
Sent: 08 April 2020 15:03
To: Observium <observium@observium.org>
Cc: Kent Johannessen <kent.johannessen@elkem.com>
Subject: [Observium] SNMP Max Repetitons enabled - some modules takes longer time to poll

 

Hello,

 

We recently enabled max-rep to improve our Observium installation. And it did reduce polling time for most modules.

But we see some module had significant increases. Specifically the poller modules for cisco eigrp, cef and cbqos and the bgp poller also seems to have increased somewhat.

 

Has anyone seen this behavior ? Where should be begin to look?

 

NOTICE: Please immediately e-mail back to sender if you are not the intended recipient. Thereafter delete the e-mail along with any attachments without making copies. The sender reserves all rights of privilege, confidentiality and copyright.