Cold Migration of a Virtual Machine (While Powered Off) Might Lead to Data Loss if a Network Failure Occurs

• A network failure occurs at the same time a cold migration is moving a virtual machine between ESX Servers.
• The virtual machines disk files are stored on a VMFS SAN volume shared between the source/destination ESX Servers.
• The VMFS SAN volume being used to store the virtual machine disk files remains unchanged by the cold migration.

In other words, this might occur when the network fails while a cold-migration operation is being used to unregister a virtual machine on Host1, and register it on Host2, and the disk files remain on the same shared VMFS SAN volume. Under these failure conditions, VirtualCenter may incorrectly, and permanently, remove a virtual machines disk file from the VMFS volume.

VMware released a patch to VirtualCenter 1.3.1, called VirtualCenter 1.3.1 patch 1, and we strongly recommend the customers on all versions of VirtualCenter upgrade to prevent this issue from occurring under network failure conditions. Please review the updated VirtualCenter 1.3.1 release notes for installation & upgrade instructions.

NOTE: The VirtualCenter 1.3.1 patch 1 binaries available for download are a full installer that can be used to upgrade from any previous version of VirtualCenter.

Until the patch is applied to your VirtualCenter environments, the recommended workaround is to avoid the use of cold migration to migrate virtual machines when the underlying VMFS SAN volume remains unchanged.