How do I monitor for a failover?

[jhowell75]

I inherited a 2 node cluster, and am forced to use CA's UniCenter product for monitoring. They want me to give them a specific eventID to scrape for, in order to set up their software to monitor our cluster. I need to know what shows up in what log when an actual failover occurs. The goal would be to get as few false notifications as possible, and only get paged when an actual failover needs to be checked.

Any help would be appreciated.

 

[mrdenny]

An actual failover will be some sort of hardware failure on the active node.

That said:
Event ID 1200 is the cluster trying to come online.
Event ID 1201 is the cluster comming online.
Event ID 1203 is the cluster trying to go offline.
Event ID 1204 is the cluster comming offline.

On a true fail over you'll probably see 1200 and 1201 without a matching 1203 and 1204.

Denny
MCSA (2003) / MCDBA (SQL 2000) / MCTS (SQL 2005)

--Anything is possible. All it takes is a little research. (Me)

http://www.mrdenny.com

 

[AndyPeck]

Assuming you running something like sql or exchange on the cluster the application service will only be running on 1 node at a time in Active / passive mode.

If you monitor this service, you'll know when it goes off-line..

Some monitoring software will also allow you to alert when the service is running, therefore create a rule to monitor the passive node and alert when the app services starts.

you can also use snmp

http://www.snmp-informant.com/

hope this helps

Andy