False Management Agent Alarms on switchover to secondary SpectroServer.

[Brian Hunter]

We recently upgraded to Spectrum r10.2 from r10.1.2.  We did the r10.2.0 upgrade, then immediately applied the r10.2.1 update.  So the behavior that I describe could have been new with r10.2.0 but I can't say for sure.

We have 6 SpectroServers configured in fault tolerant configurations with secondary SpectroServers.  Since r10.2.1, when we stop a primary SpectroServer and a landscape switches over to the secondary, we get hundreds of false "Management Agent Lost" alarms.  It takes a matter of minutes for the false alarms to clear.  I can immediately SNMP poll the alarmed SNMP agents after switchover and they poll successfully.  Since the secondary was not active until switchover, I don't know how it's coming up with these alarms.

I have a support case open for this issue, but I'm wondering if others experience this issue, or if there's a solution available.

Thanks.

[Brian Hunter]

The other thing is that it doesn't matter the model type, router, switches of various models, and systemEdge agents.  The common thing is that we are polling them using SNMP.

[Rene Cantwell]

Brain,

 

I think you need two patches:

 

PTF_10.2.103

Symptom: False Contact lost alarms asserted on pingable models that are actually reachable.

Resolution: Faulty behavior with pingable models is corrected and false contact lost alarms are not generated.

 

Here's the knowledgebase doc that further explains it:  TEC1760694       

 

And

 

PTF_10.2.118

Symptom: SpectroSERVER's SNMP listen thread consumes 100% cpu on systems with a single interface.

 Resolution: SpectroSERVER's SNMP listen thread no longer consumes 100% cpu on systems with a single interface.

Here's the knowledgebase doc that further explains it: TEC1908617   

 

--

Rene'