I'm in a bit of a back-and-forth with out virtualization folks on how to model individual ESX host in Spectrum. Now were aren't using virtual host manager or anything and are able to model a single host in the cluster, but not any other. Turns out all the hosts in the cluster are using the same SNMP EngineID. Everywhere I read says this should be unique to the host, but when then create a unique ID, I am unable to negotiate SNMP to it. Is this simply the limitation of not using Virtual Host Manager or is there a way to get these hosts modeled without it?
If you turn on the SNMP agent for the ESX host then it should be discoverable and modelled in Spectrum.
The engine_id is used when you have SNMPv3 enabled. It must be unique (and Spectrum is a stickler for this). If you are only using SNMPv1/2 then it doesn't matter as it isn't used at all. If you have already discovered the device then delete it and re-discover it after you sort out the engine_id issue.
If you want to "do things properly" then I would suggest getting:
So once I get them to create the unique IDs, I shouldn't have any issues regarding the hosts being clustered?
As long as the engine_ids are unique (presuming that you are using SNMPv3 enabled agents on the ESX hosts) then give the discovery a go and see as to what happens.
Still failing to discover using any snmp version.