How can I turn an SNMP alert into a DFM Event ?

[rickymartin]

Colleagues,

     I'm trying to create a way of getting DFM to centrally report on the success and or failure of Snapmanager backups, I'll describe my proposed approach which I'm having problems with, but if anyone knows of a more elegant way of acheiving this, please let me know.

When there is a failure of a Snapmanager backup you get a log message similar to the following

error         app.log.err        Sun Apr 19 23:00:26 EST        SLAB-EXCH3: SME Version 3.2: (111) Backup: SnapManager for Exchange online backup failed.  (Exchange Version 6.5 (Build 7638.2: Service Pack 2)) Error code: 0xc00413bc, There is no storage group which can be backed up using VSS. All of the databases in the storage group may be dismounted.

OnTap will happily send out an SNMP trap to dfm which picks this up, but the error event isnt reflected in the status of the controller. This is mentioned in the DFM documentation which states

When the SNMP trap listener receives an SNMP trap, DataFabric Manager issues an Information event, but does not change the status of the host.
Instead, the corresponding monitor associated with the trap generates the proper event and continues to monitor the host to report status changes"

This makes me wonder

1. what is the corresponding monitor ?

2. can I change the monitor or configure it in such a way as to generate an event of my own choosing at the appropriate error level such that it changes the status of the host ?

3. Is there a better way to acheive what I'm after instead of relying on snmp events ?

Regards

John

[raja]

Can you tell me the name of the trap in netapp.mib ? Regarding the monitor, based on the type of trap, the monitor that generates the equivalent Operations Manager event is generated. volumeFull and volumeNearlyFull traps will trigger dfMon, cpuTooBusy trap will trigger cpuMon and so on.

Even though we do not change the status of a storage system when an SNMP trap is received, all the traps are saved to Operations Manager database as events. The severity of all such events is Information, but the event will have all the information you would need to understand the issue. You can see the traps and their details in Events > SNMP Traps reports.

A possible solution for the use case you mention:

You can configure an alarm for the information event (depending on the trap severity), to run a custom script. In that script, you can check if the trap is the one you are looking for and take additional action such as sending a new trap to another management server using 'snmptrap' utility on linux:

# dfm eventtype list | grep trap
alert-trap-received                           Information  alert-trap-received
critical-trap-received                        Information  critical-trap-received
emergency-trap-received                       Information  emergency-trap-received
error-trap-received                           Information  error-trap-received
information-trap-received                     Information  information-trap-received
notification-trap-received                    Information  notification-trap-received
warning-trap-received                         Information  warning-trap-received

Thanks,

Raj.

[rickymartin]

I'm not 100% sure, but I think the trap type in the netapp.mib is in the section entitled  "This section contains traps related to Client Side Applications", and is the error notification vis.

appError NOTIFICATION-TYPE

OBJECTS {productTrapData, productSerialNum}

STATUS current

DESCRIPTION

"The application encountered an error condition."

::= { netapp 0 154 }

There should be a better way of doing centralised success/failure reporting for the snapmanagers, but until we do something creative within protection manager, this may be the most elegant way of doing it.