2009-08-26 02:10 PM
My customer is having some doubts about the best way to monitor his 7 NetApp machines.
On the one hand he has the NetApp SNMP traps (as listed in TR-3608) that will provide basic alerts on machine-level failures.
On the other hand he has the Operations Manager who sees all these events too, and he's wanted to know if he should have the NetApp send traps directly to his UniCenter for all standard issues, and have o Manager only send some user defined events (like SnapMirror failures Etc.) or is there a way to have Ops. Mgr receive all the traps and relay only events above a certain level to the UniCenter ?
I've seen TR-3688 which dicusses a similar situation in an HP OV environemtn but didn't really answer my question..
Is there a best practice /official NetApp recommendation on this ?
2009-08-26 09:01 PM
You could do all the things you suggest: use OM, send directly o Unicenter, or combine. Where I am now I use OM to manage the filesystem alerts and I then send SNMP traps directly to Unicenter that I deem important, like LUN offline, vol offline, failed hardware like PDUs, fans etc.
The reason we made this choice is that if you use OM for everything it becomes a SPOF (single point of failure). Also if a controller was to drop that would be picked up by OM. Anyways, you ve got the embarrassement of choice.
2009-08-26 10:21 PM
Any event in OM can be set to generate an SNMP trap so the list of traps is technically the list of all events that OM can handle. From memory I believe the list of event types and their default severities is in an appendix at the back of the OM Admin guide.
Hope this helps.
2009-08-27 04:04 AM
Yes, Richard is correct. All the events in OM can be set to generate equivalent SNMP traps. The definition of all these OpsManager specific SNMP traps is available at dfm.mib.
You can find this MIB (dfm.mib) in the DFM installation directory itself i.e under /misc folder.