SQL Failover Cluster - "Failover" process monitoring...

[srai]

We have couple of SQL 2008 R2 Failover Clusters running in production that we have been monitoring via Nagios.
Is there a way to monitor Failover Cluster service itself, esp. is there a way Nagios can detect the actual Failover of the cluster and notify us?

Thanks,
Sharad

[abrist]

Stackoverflow has good answer:
From: http://serverfault.com/questions/24440/ ... -in-nagios

When it comes to SQL Server on a cluster, it must be installed as a virtual node, whether it's a default instance or a named instance. So a TCP check where the hostname is the virtual name with the correct port should work. For instance, if I have a two node cluster with physical nodes:

Physical1
Physical2

Don't monitor either of those for SQL Server. When SQL Server is installed, let's say I chose the virtual node name of Virtual1, then it's Virtual1 you'll want to monitor. And if SQL Server is up, then the cluster is up. If you just want to check to see if the cluster is up, look for the Network Name of the cluster group itself.

[srai]

This is what we have been doing and it does work this way... monitoring the virtual name instead of physical nodes.
However, we would like to know in real-time when a fail-over occurs on the cluster, helpful when it happens in off-hours, weekends or offdays

Thanks,

[abrist]

If the change of master is logged in a failover scenario, you could check the windows event log for such information through nscient/wmi.

[srai]

Yes, I think that's the only best route to go on this issue. I just implemented the script in production after testing it few times... works!

[sreinhardt]

Awesome, I'll lock this up. Adding just another point for more immediate alerting, you might look into sending snmptraps from the windows server to alert for this issue. I imagine M$ has something to do this.