The new OS management pack (see http://blogs.technet.com/b/kevinholman/archive/2012/06/14/opsmgr-mpupdate-new-base-os-mp-6-0-6972-0-adds-new-cluster-disks-changes-free-space-monitoring-other-fixes.aspx) has tons of improvements, among them, there’s a new way to manage cluster disks and cluster shared volumes. Now these disks are associated with the cluster name, as it should have always been, and not to the node name. This new paradigm has a drawback: the MP author didn’t port all the performance rules collection and monitors from the plain old OS logical disk to the new cluster disk and cluster shared volume classes. As Kevin states “3. There is no free space perf collection rules for “Cluster Disks”. We have multiple performance collection rules for Logical Disks, and for Cluster Shared Volumes, however there are none for the new Cluster Disks class. If you want performance reports on free space, disk latency, idle time, etc, you will need to create these.“
This is true for both classes: “Cluster Disk” and “Cluster Shared Volume”. Since we need to report on disk performances and be alerted if something is really going slow, I decided to add the few key elements that we need. Initially I thought this was going to be an easy job, it should have taken about 1 hour of work. In reality, things went a little harder.
The tough part I did not find a way to solve, was to associate the CSV or cluster disk without any drive letter to the performance monitor counter instance:
The cluster disks management pack uses the cluster WMI interface to get what it needs, but those classes just return the disk size and space used counters. What in WMI is a volume GUID in Perfmon is something like HarddiskVolume<N> and I did not find a reliable correlation key between the two, after a couple of hours of researches I gave up. (If you know such a way please let me know).
If we consider CSV are just used for VMs we can live using the performance reported by the single VMs, not accurate and suboptimal, hopefully Windows Server 2012 will address this with a meaningful perfom instance name (?).
For traditional cluster disks, as far as they have a letter associated, it’s pretty simple to add the monitors and performance collection rules needed using the PartitionName property as in the following example:
<Rule ID="QND.ClusterDisk.FreeSpace.Collection" Enabled="true" Target="ClusterDisk!Microsoft.Windows.Server.ClusterDisksMonitoring.ClusterDisk">
<DataSource ID="PerformanceDS" TypeID="SystemPerf!System.Performance.OptimizedDataProvider">
<CounterName>% Free Space</CounterName>
<WriteAction ID="WriteToDB" TypeID="SC!Microsoft.SystemCenter.CollectPerformanceData" />
<WriteAction ID="WriteToDW" TypeID="SCDW!Microsoft.SystemCenter.DataWarehouse.PublishPerformanceData" />
If you’re lazy J here is a sample MP that adds:
· Average logical disk sec/transfer monitor
· Average logical disk sec/transfer collection rule
· Logical disk reads/sec collection rule
· Logical disk writes/sec collection rule
· Logical disk free megabytes collection rule
· Logical disk % free space collection rule
‘til next time
This posting is provided "AS IS" with no warranties, and confers no rights.