I've recently taken over a Hyper-V failover cluster, but I'm struggling without a build document. There are several VMs already running, but I need to learn how to create a new VM properly within this cluster. Here's what I know step by step: 1. Storage teams create the volumes and link them to the two physical nodes. 2. The disks show up as offline, so I need to bring them online and create partitions without assigning any letters. 3. I add them to the failover cluster's available disks. However, I get stuck here. 4. I'm not sure if I should add them to the Cluster Shared Volumes (CSV) or assign them directly to the VM. 5. Once I connect the disks to the VM via the SCSI controller, the VM fails to power on with a message saying there isn't enough disk space, even though there's plenty available. I believe each VM should only need one disk—sized for the VM files, checkpoints, and VHDX files. For example, I have a VM with 8GB memory, a C drive of 120GB, and a D drive of 600GB; hence I think a 1TB disk would be adequate. I can't figure out where I'm going wrong as the newly created VM doesn't show up in the C:ClusterStorage directory. I'm feeling really lost here, and since this is a production environment, I can't afford any mistakes. Any advice or even a build document would be immensely helpful!
2 Answers
For your scenario, the disks used for the Cluster Shared Volumes (CSV) need to be visible to all nodes before anything else. One node can bring the disks online and format them to create initial folders like "ISO" and "VMs". Once you have added the storage to the cluster, the CSVs will show up as drives under C:\ClusterStorage\VolumeX.
It's worth mentioning that in a typical cluster setup, VMs don’t get a dedicated disk each. Instead, CSVs act as a centralized datastore where you can have folders for your VMs and their files.
When creating a new VM, you have to specify where the configuration and VHDX files are stored. This might be where you're getting stuck. Make a new folder on your volume for the VM, create your new VM on one of the nodes, and point it to that folder when you set it up. Then, you can add the VM Role in the cluster to make it officially part of the cluster and test out a live migration to the second node after that.
Have you run a cluster validation on your setup yet? It's a crucial step to ensure everything is configured properly. Just be careful when doing a full storage validation; it might temporarily suspend your VMs. You may want to coordinate a downtime to run this if you haven't already. Getting a clean validation report could really help clarify if you're missing anything.
Just a heads-up on that—last time I ran a full storage validation, everything checked out clean, but it's best to communicate this to your team. Better to have a scheduled downtime than to take risks.