System Center Virtual Machine Manager (SCVMM) is a great tool for managing virtual machines once you have more than 2 or 3 hosts. It exposes a lot of tools and resources that are hidden or not available when managing machines at the base Hyper-V level. However, it may require you to rethink some of the ways you manage virtual machines and their hosts because it manages things at a datacenter level. One example is in the management of virtual networks within a virtual machine cluster.
In a standalone Hyper-v cluster, one of the main requirements to ensure you can move VM’s from one machine to another is that the virtual switches have the same name on each server. Virtual Machine manager adds another abstraction layer, such as logical networks and logical switches to make things easier to manage at scale. This can sometimes trip up the smaller deployment that may be adding VMM to an existing network.
Recently, I came across an issue where the VMs in the Hyper-V cluster were showing up as ‘unsupported cluster configuration’. The machines were still running and I could log on and manage them fine:
Looking at the properties of one of the machines helped to solve the puzzle by letting me know that the network adapter was not configured correctly (Error 13921). However I knew it was configured correctly because I could verify and validate everything when looking at the machine in Hyper-V and I was able to access the machine on the network.
Going back to my note above about virtual abstraction layers, VMM introduced the concept of Logical Switches to improve datacenter management. You can read more than you would ever care to know here – http://blogs.technet.com/b/scvmm/archive/2013/02/14/networking-in-vmm-2012-sp1-logical-networks-part-i.aspx. Looking at the virtual switch configuration of the cluster, I saw it was blank:
So now that I knew what the problem was, how did I fix it? The trick was to add Logical Networks and then assign them to the network adapters.
In this cluster, I had two major network groups: iSCSI and LAN. I added those as logical networks under the fabric settings. Again going back to how powerful VMM can be, there are a lot of settings for subnets and Vlans that can be configured. Since I was just representing a physical network, I only had to name it, choose ‘one connected network’ and click next through the rest.
Rinse and repeat for your networks and I ended up with something that looks like this:
Now I needed to go back and assign the logical networks to the physical network adapters on each server. To do that, I had to go to the properties of the hosts in your cluster and under the hardware tab, choose the proper logical network to align with my ports. Then I repeated for each network adapter attached to a virtual switch.
Once that was completed, a refresh of cluster and hosts in VMM allowed the Virtual Switches to show up within the cluster properties:
After some time (or if you don’t have patience you can refresh all the virtual machines), you will see the machines return to their stable state and the logical networks will show up on the VM’s virtual NIC configuration:
Virtual machine management is not something I do personally on a daily basis so it required some research to work through the concepts of the VM fabric and how they operate. I hope this quick troubleshooting walk-through can help prevent you from banging your head against the wall when troubleshooting your clusters in the future.