OpsMgr: MP Update: New Base OS MP 6.0.6989.0 Adds support for monitoring Windows Server 2012 OS and fixes some previous issues.


 

A new Base OS MP Version 6.0.6989.0 has shipped.  This management pack includes updated MP’s for Windows 2000 through Windows 2008R2 operating systems, and additionally adds new support for Windows Server 2012 OS.  This new MP will import into OpsMgr 2007 or 2012 management groups.

 

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

 

Lets talk for a minute about Windows Server 2012.  This new server OS has hit RTM already.  However – initially this OS version was only going to be supported as an agent monitored system for SCOM 2012.  There was to be no support for monitoring WS2012 with SCOM 2007R2.  However – that position has changed somewhat – as we will now support monitoring of Windows Server 2012, with a SCOM 2007R2 CU5+ management group, as long as you deploy a SCOM2012 *agent* to any monitored Windows Server 2012 instances.  Using a SCOM 2012 agent *requires* that the customer is licensed for SCOM 2012, but they aren’t ready to upgrade just yet, or are in the process of upgrading.

Here is a sample table that is correct to the best of my understanding – based on the details about licensing above:

OpsMgr Windows Server 2012 agent Support Grid:

Management Group Version

Management Server or Gateway Version

OpsMgr Agent Version

Agent OS Version

Supported Configuration?

2007 R2

2007 R2

2007 R2

Windows Server 2012

No

2007 R2

2007 R2

2012 RTM

Windows Server 2012

Yes

2007 R2

2007 R2

2012 SP1

Windows Server 2012

Yes

2012 RTM

2012 RTM

2012 RTM

Windows Server 2012

Yes

2012 SP1

2012 SP1

2012 SP1

Windows Server 2012

Yes

 

 

 

 

For using Windows Server 2012 as a server role in your SCOM infrastructure (management server, gateway, web console, etc…) we will require SCOM 2012 SP1 at a minimum.

image

 

Ok – so what’s new in this MP?

 

From the guide:

The September 2012 update (version 6.0.6989.0) of the Windows Server Operating System Management Pack included the following changes:

  • Introduced support for Windows Server 2012
  • Added monitoring for NTFS events to check if they are corrupted or unavailable
  • Disabled monitoring of Event 55 in Windows 2012 as this event can be raised even if there is no corruption. Event 55 has been replaced with Event 98.
  • Update rules and monitors for Counter Processor Information in the 2008 Management Pack to fix the issue of renaming “Processor” to “Processor Information”.
  • Updated Cluster disk monitoring to fix the warning 999 and the Script error generated when a cluster disk is removed.

 

 

First off – Windows Server 2012.

An oddity I noticed right away – was that this MP places the WS2012 classes in the Microsoft.Windows.Server.Library.mp.  That’s a divergence from previous OS management packs and will make removing WS2012 MP’s from the environment more difficult down the road.  But once this is done, it is done, and cannot be reversed as many other MP’s will build dependencies on this.  Probably not a big deal, because by the time WS2012 is end-of-life, we will be moved on to some completely new way of monitoring… or everything will be cloud based by then.  I’m personally hoping we will be riding surfboard hovercrafts by then.  🙂

 

image

 

There are 14 discoveries in the Microsoft.Windows.Server.2012.Discovery.mp which aligns exactly to the Windows Server 2008 discovery MP.  This means we will be discovering essentially the same classes and objects on this new version.

 

image

 

Note – see above – just as in previous OS MP’s – the mount point discovery is disabled by default.  The mount point discovery discovers any disks that are not assigned a drive letter.  If you have these, then you would need to enable this.  This will also discover all those tiny 100mb (server 2008R2) and 350mb (WS2012) volumes that are installed by default.

 

There are 41 Unit monitors in the Microsoft.Windows.Server.2012.Monitoring.mp.  These look at items like disk free space, disk latency, memory utilization and exhaustion, network adapter bandwidth, critical OS service availability, CPU utilization and queue length, etc.

There are 97 rules.  The majority of these are performance collection.  However some are event based rules, which generate alerts based on specific events from the system or application event log.

As you can see – this is a very simple and straightforward MP for the Operating System.

 

They all show up nicely in my OpsMgr 2012 RTM management group.  The agents deploy without issue.  Didn’t notice any strange alerts or noise.  As always – before moving to production you should test and validate that your key monitoring scenarios are working as expected.

image

 

 

Ok – what about the fixes and changes?

The best way I know to REALLY find out what changed in a MP – is to either trust the guide lists all the changes made, or run a DIFF from a diff tool and see for yourself.  Here are some changes I see:

 

Server 2008 Monitoring MP:

  • Several scripts were modified – adding in a new function for getting registry info.  Nothing major here.
  • Several monitoring workflows were change from Processor, to “Processor Information” perf object.  This change was made because a new perf counter/object (Processor Information) was added to the OS to support more than 64 logical processors.  The old perf counter object (Processor) was limited to 64 CPU’s.  As physical hardware is starting to ship 6+ core systems, with HT, and multiple sockets, this was a problem for measuring utilization for VERY large boxes.  See this for more info:  http://blogs.msdn.com/b/saponsqlserver/archive/2010/09/24/windows-2008-r2-performance-monitor-processor-information-counters-have-replaced-processor-counters.aspx  NOTE:  This might BREAK your existing reports and dashboard views that are expecting “Processor” object, as we no longer collect that.…. so be prepared to make some changes there.  Additionally, this Perf Object was added in Server 2008R2, so it does not apply to Server 2008 (nonR2)
  • The interval for the Microsoft.Windows.Server.2008.OperatingSystem.MemoryAvailableMBytes Monitor was changed from 120 seconds to 900 seconds, and the timeout was changed from 100 seconds to 300 seconds.  This is not documented in the “changes” section in the guide.
  • The monitor for Microsoft.Windows.Server.2008.OperatingSystem.TotalCPUUtilization was modified.  Interval was changed from 300 to 900 seconds.  Timeout was changed from 100 to 300 seconds.  This means the monitor will not change state in a high CPU condition until 45 minutes has passed – as opposed to 15 minutes in the previous configuration.  Additionally – this monitor looks at the “Processor Information” perf object instead of “Processor”.

Server 2008 Discovery MP:

  • Server discovery script was updated – looks like some bug fixes, and filtering to ensure we don’t discover Windows Server 2012 in this discovery.
  • Several scripts were modified – adding in a new function for getting registry info.  Nothing major here.

Windows Server Cluster Disks MP:

  • Updates to the Cluster Name discovery script.
  • Updates to the Cluster disk discovery script.

 

These all look very straightforward.  In my previous article on this MP – I talked about some known issues in the previous version:

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

If you want a deeper understanding of how the Base OS MP works with cluster disks and free space monitoring – read that article and this one:

http://blogs.technet.com/b/kevinholman/archive/2011/09/30/opsmgr-new-base-os-mp-6-0-6956-0-adds-cluster-shared-volume-monitoring-bpa-and-many-changes.aspx?PageIndex=4#comments

 

 

 

Known Issues in this MP:

 

The previous issues in the previous Base OS MP:

1.  The knowledge for the Total CPU Utilization Percentage is incorrect – the monitor was updated to a default value of 3 samples but the knowledge still reflects 5 samples.  This is still an issue (no biggee)

2.  The CPU monitor and collection rule for Server 2008 was changed from “Processor” to “Processor Information” for the Performance Object.  HOWEVER – this new object only exists in Server 2008 R2 and later, it does not exist in Server 2008 (non R2).  Therefore, there is no performance monitoring or collection for your CPU on that OS.  You will need to create a rule and a monitor to address this for “Processor” as a workaround for Server 2008 machines.

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 still an issue and will impact your ability to run reports on free disk space for any disk in a cluster.  This is a significant issue, and you should plan to create performance collection rules for free space and target these cluster disks.  Since this can be complicated to get it right – I wrote an article on how to do this correctly, and offer a sample MP for download:  http://blogs.technet.com/b/kevinholman/archive/2012/09/27/opsmgr-authoring-performance-collection-rules-for-cluster-disks-the-right-way.aspx

4.  It is being reported that this MP is un-discovering logical disks that are part of a cluster on Server 2008 OS.  This is bad, because we only discover cluster disk instances on Server 2008 R2.  This would leave any disks that are part of a Server 2008 OS cluster unmonitored.  This is under investigation.

5.  Previous versions of this MP:  Perf collection and disk monitoring for cluster disks and CSV’s only works when the resource group hosting the disks, are on the same node that is hosting the cluster name (quorum) resource.  If the disk’s resource group is running on a different node than the cluster name itself, perf collection and monitoring will cease.  This was a HUGE issue for monitoring clusters.  I am happy to announce this is fixed now.

image

As you can see in the graphic above – the disks hosted in a cluster resource group are now correctly associated with the resource group name and virtual server, and not the quorum/cluster name as it was in the previous version.  What this means – is that once we discover a cluster disk, we associate it to the resource group, which is associated to a network name object in the resource group, which is associated with a Windows Computer object.  This is the correct way to discover a clustered resource, and now when this disk moves from node to node – it is NOT discovered as a totally new entity, it retains the same entity GUID and is the same object from a reporting perspective.  Additionally, it also provides continuous performance collection after a failover.

image

In the graphic above – you can see the red arrow drawn – which represents the failover – where the database disks in a resource group were failed over to a different node – and perf collection and monitoring picked right back up once the object initialized on the new node.

 

 

Summary:

 

With the additional of Windows Server 2012 monitoring, and the critical fixes to the cluster disk behavior – this looks like a pretty solid update and MP.  If you are running 6.0.6972.0 then I’d say upgrade to this version as soon as you can, as it addresses some critical issues.  If you are running 6.0.6958.0 or older, then you might consider holding off on this version, as it has some radical changes and some known issues.  Make sure you test and are prepared for the changes to how cluster disks are discovered and monitored, and plan for the changes to the processor utilization performance collection for your reports, especially the rules and monitors you need to create concerning Windows Server 2008 (non-R2)


Comments (35)

  1. Anonymous says:

    Logical Disk Free Space Monitor Stopped Working with Upgrade, i have two servers with Window Server 2008 R2 Standards PT-BR but dont working Monitor Logical Disk Free Space after this upgrade.

  2. Anonymous says:

    Hi Kevin,

    Hope all is well. In summary, you mentioned not to upgrade the MP if we are running 6.0.6958.0 . I am aware of the changes done to previous version 6.0.6972.0 of MP. What other concerns you got regarding this? Could you describe a bit please.?

    Thanks again.

    Fahim

  3. Anonymous says:

    Anyone else having trouble with the download link? I can download the file, but it's still the 6.0.6972.0 version of the mp's.

  4. Kevin Greene says:

    Excellent posting and synopsis as usual Kevin

    – Thanks!

  5. Kevin Holman says:

    @Vishi –

    I don't recommend upgrading from 6.0.6958.0 to 6.0.6989, as I stated in the summary of this article.  Since this version breaks cluster monitoring of 2008 cluster disks… I'd recommend waiting for the next release where this will be fixed.

  6. Kevin Holman says:

    @Sylvain and Jarrad – yes – I see this as well.  The Windows Server 2012 OS discovery needs to run on ANY Windows Computer – that is normal and by design…  this script error is just a sign that the script is checking something that doesn't exist or fails on a Non-WS2012 server.  Ignore it.  Yes – the script could potentially be optimized to not fail on a non-windows server 2012 computer… but this isn't a sign of any issue.  I'll report it up the chain or feel free to open a bug-case on this.

  7. Anonymous says:

    This person MaketPlace site has the same problem, systemcenter.pinpoint.microsoft.com/…/reviews

    "Logical disk free space monitor not triggering alerts after core mp upgrade"

    Thanks.

  8. Kevin Holman says:

    @John – I don't have any insight into the ETA.  Best way to get that is open a support case on the bug and request a hotfix.

  9. Kevin Holman says:

    There should be an updated MP out soon which will address those rules/monitors.

  10. Kevin Holman says:

    @Fahim –

    The list of known issues that were created by changing over to cluster disks (missing perf data, missing 2008 clusters) and the "Processor Information" change are the reasons I don't recommend upgrading to this version of the Base OS MP.  When those issues are fixed in a future update, I'll change the stance.  6.0.6958.0 is a pretty solid MP with a few smaller issues which I have long since blogged about workarounds.  I like where the MP is heading – changing to cluster disks is the *right* direction, to have these hosted by cluster resource groups, but with this change, came a few issues that will be resolved soon.

  11. Anonymous says:

    Hi Kevin

    Do you know if there's any update planned regarding the new Processor counter instance especially for reports?

    Actually processor performance is something essential for capacity planning and monitoring, so I don't really understand why the reports aren't reflecting this essential change.

    You've written "Therefore, there is no performance monitoring or collection for your CPU on that OS.  You will need to create a rule and a monitor to address this for “Processor” as a workaround for Server 2008 machines." => I hope that this will be fixed in the next release? For me this is base monitoring and therefore a must.

    Thanks

    BR

    Klaus

  12. Sven Wells says:

    Hello,

    We upgraded to 6.0.6989.0 just before Christmas 2012.  Previously we were running 6.0.6959.0.  We've discovered what Leandro above discovered.  Even though we've disabled the "Logicak Disk Free Space" monitor and enabled the "Windows Server 2003 Logical Disk Free Space (MB) Low" monitor, these alerts are not being produced.  We've 'played' with the samples, increased the thresholds to unconventional/extraordinary amounts (ie. 100GB for a 50GB drive) etc.  We see that the monitor is enabled and appears to be working, however, no Alerts.  Performance collection rules are working fine.

    Any ideas?

    Thanks,

    Sven

  13. Kevin Holman says:

    That is covered in my known issues #4 above.  I have reported this and filed a bug.  I recommend opening a case with Microsoft if you are impacted by this to provide feedback for getting this fixed.

  14. Anonymous says:

    We are running 6.0.6958.0 and have several high profile 2008 Clusters so we've held off…Thank goodness I read your blog!  We've been waiting patiently for the next release.  But, since several of the newer MPs require the Core OS update.  We have several other issues unresolved by fixes in these newer MPs.  For us, the MPs affected are:  

    -Microsoft Hyper-V (6.2.6640.0) – huge deal for us

    -Win2003 / Win 2008 Print Server (6.0.7004.0) – big deal for us

    -Active Directory (6.0.8070.0) MPs.

    Not to mention, the new Core OS MP itself will address several logical disk monitor capabilities we desperately need.  I've heard from the rumor mill that a private fix is out.  So, can you shed any light as to when we can expect the next release?

    Thanks

    –Steve

  15. Anonymous says:

    Hi,

    It would be great if we can get any Addendum management pack to address the “Processor” as a workaround for Server 2008 machines.

    Regards,

    Sri

  16. Anonymous says:

    Kevin,

    Thanks for your prompt response. Any idea exactly when the new MP version will be released?

    Regards,

    Fahim

  17. robg4 says:

    Hi Kevin,
    Not sure that you still do this but…
    You refer to SCOM 2007 R2 support for Server 2012 requiring the SCOM 2012 agent for support. Is it fair to assume that other management packs will work as expected on 2012 servers? I am looking at importing the following in to a 2007 R2 environment.
    Windows Server File and ISCSI Services 2012
    Windows Server File and ISCSI Services 2012 R2
    Cheers,
    Rob

  18. Mark Derouen says:

    Sneak peak at the 2012 MP Viewer. When can we download it?

  19. Jarrad says:

    Hi Kevin,

    Since deploying this MP, I keep getting the following warnings

    Event ID 21405:

    Failed to create System.Discovery.Data, no errors detected in the output.  The process exited with 0 "C:Windowssystem32cscript.exe" /nologo "Microsoft.Windows.Server.2012.Discovery.Probe.vbs"

    and Event ID 21406

    C:Program FilesSystem Center Operations ManagerAgentHealth Service StateMonitoring Host Temporary Files 341893HotFixValidation.vbs(117, 10) Microsoft VBScript runtime error: Subscript out of range: 'count'

    on non Windows Server 2012 agents.

    Cheers,

    Jarrad

  20. Sylvain says:

    I'm adding my voice to Jarrad.  I have the same issue in my environment.

  21. Sylvain says:

    Kevin,

    I also don't see Windows 2008 cluster disk in my environment.  Only cluster disk coming from a Windows 2008 R2 cluster are showing up.  Do you know if this is going to be fixed soon ?  This is not nice….

  22. Leandro S. says:

    Logical Disk Free Space Monitor Stopped Working with Upgrade, i have two servers with Window Server 2008 R2 Standards PT-BR but dont working Monitor Logical Disk Free Space after this upgrade.

  23. John Yantorno says:

    Cluster disk discovery is definitely broken  any ETA on a fix?

  24. dom says:

    Hello Kevin,

    Happy Holidays

    I have installed the MP 6.0.6989.0 an hour ago and I do not get any performances back on any server… should I still wait…?

    Thanks,

    Dom

  25. Vishi says:

    Hi Kevin,

    We are running 6.0.6958.0 MP, we don't have the addidtional components installed(CSV monitoring, BPA and reports) as I have upgraded via MP catalog. I could not download the earlier versions from microsoft catalog. Would it be a good idea to upgrade to 6.0.6989? What would you recommend us to do?

  26. Vishi says:

    Thanks Kevin, will wait for the next release then

  27. Chris McIntyre says:

    We've imported this MP and I've worked around all the existing issues mentioned here.  But there is one thing that still annoys me.  If you go into the "Processor Performance" view, the legend shows two objects for each Server 2008 machine.  One for the "Processor Information" object and one for just the "Processor" object.  What puzzles me is that the same rule is indicated as collecting both counters, although it specifically only lists the "Processor Information" object in its configuration.  It's a pain to not know which one to click, and have hundreds of items in the legend that don't show any data.  Looks bad.  Any ideas on how to fix this?

  28. Eric Peterson says:

    Hi Kevin,

    Is there any more information or a possible ETA on a solution for the 2nd known issue in this MP relating to collection of CPU data on Windows 2008 R2 machines?

    Thanks,

    -Eric

  29. What about the Monitor "Total CPU Utilization Percentage"?  Doesn't it subscribe to the data published by "Processor % processor Time  _Total"?  My tests have shown the monitor to be broken by the CU5 update for Server 2008 Standard edition.  Is there a fix for this?

  30. My previous comment assumes we have built a new rule for "processor % processor time _total" and that it works.  I am lead to believe by my testing that the monitor "Total CPU Utilization Percentage"  subscribes to the original Rule and therefore for Server 2008 Standart Edition, it doesn't work. I can't trigger the monitor for a 2008 server but can easily trigger the monitor for a 2008 R2 server.  I do know how the monitor works and I've over ridden the thresholds for both queue and cpu utilization.

  31. NorthernTurstTim says:

    I incorrectly said CU5 update broke the monitor. It is the new Base OS MP that is responsible.  Sorry for any confusion.

  32. Is still Windows Server 2012 version 142.101.204.10 also required SCOM 2012 agent installed on the agent box or it will work with the SCOM 2007 R2 version agent?