Migration of a powered-on VM after growing a disk can result in data corruption.

Products

• VMware vSphere ESXi

Issue/Introduction

This article will make users aware of the potential data corruption if a VM with a hot-grown disk is migrated, and how to avoid it.

Symptoms:
Virtual disks that belong to a powered-on VM can be grown (extended) without first powering-off the VM. This operation is called "hot-grow". But, if a hot-grow is performed and then the VM is migrated to another location (via Storage vMotion or XvMotion), then this disk can potentially be corrupted. Specifically, data written to the growing area by the guest may not get copied.

Environment

VMware vSphere ESXi 8.0.2

Cause

Some optimizations were made for the hot-grow workflow, specifically to allow hot-grow of multiwriter VVol disks. These optimizations resulted in this bug as a side effect. (The bug can happen for disks of all datastore types, not just VVol.)

Resolution

This issue has been fixed in 8.0.2 P03.

Workaround:
Grow a disk only when the VM is powered-off ("cold-grow"). Or, if a disk is already hot-grown, perform a VM operation that will cause the disk to be internally re-opened. This includes power-off and power-on, any snapshot operation, suspending and resuming, or hot-removing and hot-adding that specific disk.

Additional Information

vSphere 8.0 U2 restored file corruption with resized VM disks

[VMC on AWS] VMDK Hot-extend operation and potential data loss

QueryChangedDiskAreas API returns incorrect sectors after extending virtual machine VMDK file with Changed Block Tracking (CBT) enabled in ESXi 8.0u2