MP Update: Windows Server Operating System 6.0.7316.0 released


 

The base operating system MP’s have been updated:

https://www.microsoft.com/en-us/download/details.aspx?id=9296

 

Previously, there were a couple interim releases that were pulled due to issues, mostly affecting older operating system versions.  This was due to the MP focusing on changes for Windows Server 2016.  This MP update addresses those issues caused by the interim MP changes.  The previous stable MP version was 6.0.7297.0 so I will focus on changes since that MP:

 

Changes in version 6.0.7303.0

  • MP used to discover physical CPU, which performance monitor instance name property was not correlated with Windows PerfMon object (expecting instance name in (socket, core) format). That affected related rules and monitors. With this release, MP discovers logical processors, rather than physical, and populates performance monitor instance name in proper format
  • Microsoft.Windows.Server.ClusterSharedVolumeMonitoring.mp and Microsoft.Windows.Server.Library.mp scripts code migration to PowerShell in scope of Windows Server 2016 Nano support (relevantly introduced in Windows Server 2016 MP version 10.0.1.0).
  • Updated Microsoft.Windows.Server.ClusterSharedVolumeMonitoring.ClusterSharedVolume.Monitoring.State monitor alert properties and description. The fix resolved property replacement failure warning been generated on monitor alert firing.

 

Changes in version 6.0.7310.0

  • Several bugs located in Cluster Shared Volumes MP were fixed (see below); error handling migrated to common recommended scenario. Enabled Quorum monitoring via changing the monitoring logic. The monitoring logic is splitting for Nano Server (with usage of PowerShell) and all other operation systems.
    • Fixed bug: disk free space monitoring issue on Quorum disks in failover clusters; the monitor was displayed as healthy, but actually it did not work and no performance data was collected.
    • Fixed bug: logical disk discovery did not discover logical disk on non-clustered server with Failover Cluster Feature enabled.
    • Fixed bug: Clustered Shared Volumes were being discovered twice – as a Clustered Shared Volume and as a logical disk; now they are discovered as Clustered Shared Volumes only.
    • Fixed bug (partially): mount points were being discovered twice for cluster disks mounted to a folder – as a cluster disk and as a logical disk. See Troubleshooting and Known Issues section for details.
    • Fixed bug: Cluster Shared Volume objects were being discovered incorrectly when they had more than one partition (applied to discovery and monitoring): only one partition was discovered, while the monitoring data was discovered for all partitions available. The key field is changed, and now partitions are discovered correctly; see Troubleshooting and Known Issues section for details.
  • Error handling was corrected. Logical disk correct discoveries on non-cluster server with Failover Clustered Server Feature is installed.
  • Created new overrides for Cluster Shared Volume MP, as long as the old ones did not work.
  • Cluster disk monitors alert messages: alert title might be disorienting and was corrected.

 

Changes in version 6.0.7316.0

  • Due to incompatibility issues in monitoring logic, several Cluster Shared Volumes MP bugs remained in version 6.0.7310.0. These are now fixed in the current version (see the complete list of bugs below). To provide compatibility with the previous MP versions, all monitoring logic (structure of classes’ discovery) was reverted to the one present in version 6.0.7297.0.
    • Fixed bug: disk free space monitoring issue on Quorum disks in failover clusters; the monitor was displayed as healthy, but actually it did not work and no performance data was collected.
    • Fixed bug: logical disk discovery did not discover logical disk on non-clustered server with Failover Cluster Feature enabled.
    • Fixed bug: Clustered Shared Volumes were being discovered twice – as a Clustered Shared Volume and as a logical disk; now they are discovered as Clustered Shared Volumes only.
    • Fixed bug (partially): mount points were being discovered twice for cluster disks mounted to a folder – as a cluster disk and as a logical disk. See Troubleshooting and Known Issues section for details.
    • Fixed bug: Cluster Shared Volume objects were being discovered incorrectly when they had more than one partition (applied to discovery and monitoring): only one partition was discovered, while the monitoring data was discovered for all partitions available. The key field is changed, and now partitions are discovered correctly; see Troubleshooting and Known Issues section for details.
    • Fixed bug: physical CPUs are now discovered on Windows Server 2008 R2 platforms; logical CPUs are no longer discovered, see Troubleshooting and Known Issues section for details.
    • Fixed bug: Windows Server 2008 Max Concurrent API Monitor did not work on Windows Server 2008 platform. Now, it is supported on Windows Server platforms starting from Windows Server 2008 R2.
    • Fixed bug: when network resource name contained more than 15 symbols, the last symbols of the name were cut off, which was resulting in cluster disks and Cluster Shared Volume discovery issues.
  • Cluster disk monitors alert messages: alert title might be disorienting and was corrected.

 

I have been running this version for a few weeks now, and I haven’t seen any major issues.  However, like ALL MP’s, I recommend careful testing and evaluation in your lab and test environments before moving to production.


Comments (7)

  1. Balazs says:

    I’ve start testing the Microsoft.Windows.Cluster.Management.Library :: 10.0.1.0 MP along with these modifications. Unfortunately the new PS based discovery @ Microsoft.Windows.Cluster.Monitoring.Discovery.ScriptProbe.WithResourceFilter truncate the names of the cluster’s network name if it is longer than ~40 character.
    I don’t know why is this happens. When the script call the $oNetwork.PartComponent.ToString() method it returns a truncated (limited to 70 characters) string with ‘…’ ending. Unfortunately the next WMI query will failed because the name will not match with the truncated one.
    This error may appear in similar MPs where the vbs discovery script was transformed to powershell. This is why I post it here.

    Thank you for your blog by the way, I have learnt a lot from these articles!

  2. LTHPETER says:

    Hi!

    Can anyone explain why “Windows Server Operating System Library” version is 6.0.7316.0 in Download Center, while 10.0.3.0 in the integrated catalog? One don´t want to mess up the core OS monitoring.

    Regards
    Peter

  3. Thanks Kelvin, Excellent document.

  4. Michael says:

    Hi Kevin,

    Did they also pull 6.0.7316.0? I’m trying to find what the latest Server 2012R2 base OS MP is and keep getting the page is no longer here message.

    I’m currently on 60.7297.0. I’m having a weird issue where the CPU counter is not resolving. Event 10102 “In PerfDataSource, could not resolve counter instance % Processor”. I found some details that the MP changes for physical to logical discovery. The pain is all but one of 6 server all built at the same time and patch level have detected the % Processor Total monitor. When I open Perfmon the counter is available and already added by default.

    I have recreated the Health Service Cache and still get the same event. These are production servers so I want to tread carefully. Any thoughts on the MP side or how to see how the Perf counter is names to see how it is different?

    Thanks,

    Mike

    1. Kevin Holman says:

      Mike – I don’t know why this isnt available. Usually – when this happens, it means there is an update coming in the next couple days. I’d check back in a couple and see what happens.

      As to CPU monitoring – I’d need more data on what the issue is.

      1. Yousef says:

        why the download link is not working?

      2. Michael says:

        Thank Kevin,

        Sorted the CPU issue. apparently I needed to restart service/flush health service cache another 4 or 5 time before it sorted itself out. Very odd but glad i didn’t need to hack at the OS to get the perf counter so the SCOM agent could resolve the counter.

        Thanks again and as always appreciate all your knowledge

        Mike

Skip to main content