Hyper-V VMs crash when any SAN volume is expanded

I’d start by verifying the MPIO setup on both cluster nodes. A LUN expansion can trigger a target rescan, and an incorrect or incomplete MPIO configuration may leave paths unavailable or cause a long I/O interruption. Check that the correct HPE DSM or supported path policy is installed, all expected paths are visible, and both nodes have matching settings and firmware.

Confirm whether the VM disks are on Cluster Shared Volumes and determine whether they are formatted as NTFS or ReFS. CSV redirected mode can significantly increase dependence on the cluster network and the CSV owner. Verify that direct I/O is working where expected, the nodes have healthy low-latency networking, and storage ownership is behaving normally. Matching VM and storage ownership may be hiding the problem rather than fixing it.

This sounds like the CSV or underlying storage temporarily stopping I/O rather than the VM itself being the root cause. When the VM and CSV ownership are split, traffic may use redirected I/O through the node that owns the CSV. If the SAN pauses during the expansion, the guest can hit its disk timeout and crash.

Check both cluster nodes for CSV pause or connectivity events, especially Event ID 5120, at the exact time of the volume expansion. Also review the MSA controller firmware, host connectivity, iSCSI sessions, and any vendor advisories related to expanding volumes.

The fact that the failure disappears when roles are placed on the same node points toward a pathing, CSV redirection, or storage-failover issue rather than a specific VM disk problem. I’d reproduce it during a maintenance window while collecting cluster, iSCSI, MPIO, and SAN controller logs. Compare the timestamps with the expansion operation and check whether any paths briefly disappear or the CSV enters a paused state.