Possible Workaround to Migrate Clustered VMDK VMs to New Storage Via Quasi-Online Way

[Aleem Syed]

Hello Community,

We have two node WSFC running on vSphere 7.0.3 via Clustered VMDK Datastore. We need to storage vmotion to a new datastore on vSphere 8 ESXi hosts. I know that hot migration is not permitted for Clustered VMDK as outline below 

https://knowledge.broadcom.com/external/article/338634/error-device-is-a-scsi-controller-engage.html

https://knowledge.broadcom.com/external/article/390585/storage-migration-of-virtual-machines-wi.html

We cannot take VMs offline to do cold migration. WSFC is running SQL Server.

What are options we have to perform migration while minimizing downtime?

I have following approach to put into action;

1. Put passive node into maintenance mode inside Failover Cluster Manager.Shutdown passive node and change SCSI Bus Sharing to None on passive node.
2.  Storage vMotion passive node with just OS drive to new storage. 
3. On Active Node, change SCSI BUS sharing to None after shutting it down. Turn on active node and ensure that drive appears healthy inside Failover Cluster Manager and SQL Server FCI starts successfully. Disk should and will retain same UUID and GUID.
4. Since SCSI Bus Sharing is set to None in step#3 , I can perform Hot Migration to new storage.
5. Once migration to new storage completes. Enable Clustered VMDK support inside new Storage LUN. 
6. Change SCSI Bus Sharing to Physical on active node. Turn it on. Disk should retain same UUID and GUID . SQL Server FCI will now treat VMDKs as having SCSI-3 persistent reservation or WEAR etc.
7. Configure passive node to enable SCSI Bus Sharing to Physical and turn it on. Failover Cluster Manager will reflect it online