CONTAP-829947: Surprise removal of a Hyper-V passthrough disk panics an ONTAP Select node and leaves its data disks unknown
Issue
On an ONTAP Select software-RAID node running on Hyper-V, removing a passthrough data disk while the node is using it can panic the node. After the node reboots, Deploy shows that node's data disks as unknown and ONTAP no longer lists them.
When a passthrough data disk is removed from a running ONTAP Select virtual machine on Hyper-V:
This follows a removal of the disk at the first location (location 0) of the virtual SCSI controller that holds the data disks. The other disks on that controller are still attached to the virtual machine, but ONTAP does not discover them on the next boot.
When a passthrough data disk is removed from a running ONTAP Select virtual machine on Hyper-V:
- The node can panic, write a core file, and reboot.
- After the reboot, storage disk show does not list that node's local data disks.
- In Deploy, node storage disk show reports ONTAP Name as unknown for those disks.
- Putting the disk back and refreshing the cluster does not restore the ONTAP names.
- On a two-node software-RAID cluster, each mirrored aggregate loses one plex and stays online on the partner node's disks. storage failover show reports that takeover is not possible.
This follows a removal of the disk at the first location (location 0) of the virtual SCSI controller that holds the data disks. The other disks on that controller are still attached to the virtual machine, but ONTAP does not discover them on the next boot.
