Storage migration of virtual machines with Shared VMDKs
search cancel

Storage migration of virtual machines with Shared VMDKs

book

Article ID: 390585

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

How to storage migrate VMs with shared VMDK disks as in virtual disks with multi-writer flag, clustered VMDK datastores (e.g. Oracle RAC, WSFC(MSCS), RHEL High Availability Cluster).


Environment

VMware vSphere ESXi 8.x
VMware vSphere ESX 9.x

Resolution

Please note that an online Storage vMotion of the primary node will fail regardless of whether secondary cluster nodes are powered off or detached due to the active physical SCSI bus-sharing configuration on the controller.

As per best practices, migrating virtual machines configured with SCSI bus-sharing requires a cold migration. All associated WSFC nodes must be completely powered down during the migration process. Furthermore, following the storage migration, the shared disk mapping locations must be manually updated on all secondary cluster virtual machines to prevent power-on initialization failures.

 

Precheck:

  • Note down the current SCSI controller settings, SCSI number and SCSI ID to which every Disk is connected, as the disks will need to be attached to the same SCSI ID when they are attached back once the migration is complete

  • Record the datastore paths of disks

  • Ensure the destination datastore has Clustered VMDK Support enabled, otherwise the VMs will fail to power on once migrated.

Migration steps:

  • Shutdown/Stop the cluster application gracefully by following all the required vendor best practices

  • Shut down all VMs that are sharing the multi-writer VMDKs (1st shutdown all the secondary/passive VMs and then primary VM)

  • Remove shared disks from the secondary VMs (choose Remove from VM only, not delete from disk)

  • Storage vMotion all the VMs to new datastore

  • Re-attach the disks to secondary VMs (Please pay attention to do that reflecting same SCSI number and ID to which every Disk is connected as noted in the Precheck)

  • Power On the VMs starting with Primary/Active Node before powering on the secondary/passive

Additional Information

Storage vMotion failing in scenario where VM has SCSI bus-sharing set to physical .Refer to this article for more information: Error: "Device is a SCSI controller engaged in bus-sharing" during VM migration

For migrating Virtual Machines with RDMs please refer to the below KB article: Migrating virtual machines with Raw Device Mappings (RDMs)