When a virtual machine with thin disk has a snapshot, the redo log is incorrectly created as thick disk

Products

• VMware vSphere ESXi

Issue/Introduction

If you create a snapshot of a virtual machine that has thin-provisioned disk with ESX/ESXi 4.x, vSphere (the Client and the underlying disk library) identifies the disk as thick. This is because the disk chain starts with the snapshot, which is always thick not thin. Most software accessing VMDK storage opens the snapshot (also called redo-log).

Backup software might misbehave due to this issue. For example, when saving a thin-provisioned disk that has a snapshot, the storage is actually backed up as thick disk. A similar problem occurs when trying to restore a thin disk.

Note: In VMware terminology, sparse disk versus flat disk is not the same as thin versus thick. To see an example, refer to the attached screenshot.

Environment

VMware ESXi 4.1.x Installable
VMware ESX 4.0.x
VMware ESXi 4.0.x Installable
VMware ESX 4.1.x
VMware ESXi 4.1.x Embedded
VMware ESXi 4.0.x Embedded

Resolution

This issue was fixed in vSphere 5.

In 4.x releases, you can use the vSphere Client to remove the snapshot (retaining current virtual machine state) or revert to the snapshot (returning the virtual machine to its state at snapshot time), and the thin provisioned disk will be properly reported.

The workaround for developers writing backup software is to determine the base disk type before taking a snapshot of it.