In our environment we have a lot of Microsoft Clusters ranging from 2003 and up. Ideally we're trying to get rid of them, but for the ones we keep for now we want to be sure to have accurate information. Our current setup for monitoring a 2 node cluster is:
1. Monitor the two nodes of the cluster like regular servers (CPU, memory, NIC's, drives), Exclude all cluster exclusive resources and application resources (Cluster NIC, cluster drives, etc)
2. Monitor the Cluster IP (Captures cluster NIC, Quorum, exclude CPU, Memory, server NIC's, server drives (C &D))
3. Monitor the Application IP (Excludes the aforementioned items, includes application specific items like drives)
Obviously keeping the IP's all included somehow is important for which piece may be in a down state, but we recently got asked if we could include things like CPU, Memory, NIC, etc in the cluster service name for ease of viewing by say the DB Admins. This way they don't have to go investigate the individual node that was active at the time of an issue.
The question is: is there any downside to adding CPU, Memory, Drives, NIC, etc to 2 and 3? Does it poll the items a 2nd and 3rd time or does it just use the information it may have already polled from 1?
Not looking to reinvent the wheel with any SAM templates or anything here, just want to make sure we have good data to support why or why not we do it the way it's been.
Did that for several years, no problems at all. We just got rid of all of the windows clusters so I'm on cloud 9. There is also a nice cluster resource report that will list the active node, just search for "RH" in online templates.