NetXMS Support Forum

Hi,

we have a problem with busy snmp agents on some switches.

We have some nodes that are changing the status in 1 minute from normal, to down, to normal.
For example:
"20:33:47" "warning" "switch1" "SYS_SNMP_UNREACHABLE" "SNMP agent is not responding"
"20:33:47" "critical" "switch1" "SYS_NODE_DOWN" "Node down"
"20:33:47" "critical" "switch1" "SYS_NODE_CRITICAL" "Node status changed to CRITICAL"
"20:34:50" "normal" "switch1" "SYS_SNMP_OK" "Connectivity with SNMP agent restored"
"20:34:50" "normal" "switch1" "SYS_NODE_UP" "Node up"

I think the reason is that our older switches are sometimes heavy loaded at the management interface because we use Kiwi Cat Tools to get the actual mac table and port infos every 30 minutes.

Is there a way to change the retry intervals for snmp connectivity? It would be no problem to receive a error message after 2 minutes instead of 1 minute. Its better than a system that produces 5 or more false error messages a day...

Greetings
Tom

Hi!

You can only try to increase SNMP request timeout by changing server's configuration variable SNMPRequestTimeout (it's value is in milliseconds). However, I doubt that it will help in your case. I will improve that in the next release.

Best regards,
Victor

Hi,

I thought about increasing the snmp timeout value, but I think this will not change anything, because the initial request can not be made. But I will test it.

Thanks you for your plans to improve this

Greetings
Tom

Hi,

I increased the value of of "SNMPREQUESTTIMEOUT" to 5000 (ms) and restarted the server.
But this night there are still node down/node up events.

So still another request for the next release

Greetings Tom

Just found the parameter "PollCountForStatusChange"  and I think I will test this one.

Unfortunately this parameter does not affect SNMP agent status detection. It only affects interface status checks.

Best regards,
Victor

hm, ok... it's a pity cause it sounded so good