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.
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
Windows Server 2012
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.
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:
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. :-)
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.
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.
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:
Server 2008 Discovery MP:
Windows Server Cluster Disks MP:
These all look very straightforward. In my previous article on this MP – I talked about some known issues in the previous version:
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:
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.
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.
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.
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)
Excellent posting and synopsis as usual Kevin
Sneak peak at the 2012 MP Viewer. When can we download it?
Mark - this has been out for a while:
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.
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:\Windows\system32\cscript.exe" /nologo "Microsoft.Windows.Server.2012.Discovery.Probe.vbs"
and Event ID 21406
C:\Program Files\System Center Operations Manager\Agent\Health Service State\Monitoring Host Temporary Files 34\1893\HotFixValidation.vbs(117, 10) Microsoft VBScript runtime error: Subscript out of range: 'count'
on non Windows Server 2012 agents.
It would be great if we can get any Addendum management pack to address the “Processor” as a workaround for Server 2008 machines.
I'm adding my voice to Jarrad. I have the same issue in my environment.
@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.
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....
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.
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.
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"
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.?