Friday, May 9, 2014

Re-associate your managed computers with VMM

Very often, I meet customer who has started with a single VMM management server.
That is all fine, because normally, these VMM servers (and the other SC components) are running as virtual machines on a Hyper-V Cluster, and ideally, it is a dedicated management cluster for those kind of workloads.
This was all good and great with VMM until the 2012 R2 release.

Since more and more business critical functions has been moved over to Virtual Machine Manager, the more important it becomes in your cloud fabric.

To highlight some few:

·         VMM act as a network controller when using network virtualization
·         VMM is the backend for your cloud, using Service Provider Foundation and Windows Azure Pack
·         VMM feeds SCOM with business critical information from a performance and monitoring perspective
·         VMM is responsible for optimization of your Hyper-V hosts, through Dynamic Optimization and Power Optimization

There are many more things to highlight in this context, but I guess you get the point. VMM is the key management layer in your datacenter.

In a nutshell, this is the process to shift from a single VMM management server, to a Highly Available VMM management server.

1)      Take a backup of the VMM database. First thing first.
2)      Uninstall the VMM server (please remember to retain the database during the process)
3)      You have already a Windows Failover Cluster installed on your nodes where you want to run VMM (yes, this can be a virtual guest cluster).
4)      Install VMM on your cluster and create a highly available VMM service – follow the instructions from TechNet, and use the existing SQL database containing your VMM data. This database can also be moved to a SQL cluster, or a different instance.
5)      You now have a HA VMM server

But what’s next?
You should re-associate your managed computers with the HA VMM server. Your HA VMM server has most likely a new name (this is a computer object too) and your will see some communication problems between VMM and its managed servers

The solution can easily be found in the shell:

$Credential = get-credential
Get-VMMManagedComputer | where {$_.State -eq "NotResponding"} | Register-SCVMMManagedComputer -Credential $Credential

You should now have a healthy fabric again.

No comments: