In this scenario we have a VM running Windows Server 2008 R2 on a hypervisor that is part of a vSphere 5.1 cluster. The VM “lives” on a volume in the SAN (on a PS 6110E).
PS Group Manager is configured to take snapshots of this volume as per a predetermined schedule.
Power off the VM prior to initiating the restore from snapshot. If a volume is attached to a server via the iSCSI initiator, disconnect the volume before the restoration process.
First off, let’s create something we want to get rid of/roll back from. We are going to create a document on the test server.
In case you want to take a “manual” snapshot of the VM you can do so by right-clicking on the appropriate volume in the PS Group Manager, and then selecting Create Snapshot. Optionally, you can provide a description for the snapshot.
Power off the virtual machine.
Back in the PS Group Manager, select the appropriate volume, and in the Activities window pane, click on Restore Volume. Next, select the snapshot you want to restore the volume from.
When prompted to set the volume offline, click on Yes.
Confirm you want to restore the volume from the specified snapshot, as well as that you want the volume to be brought back online after the restoration process has been completed.
In your vSphere client the VM will be grayed out as the volume was not disconnected prior to restore from snapshot. Select the appropriate ESX host, click on the Configuration tab, then click on Storage in the Hardware section of the left and select Rescan All.
Once completed power on the virtual machine.
Don’t panic if during the boot up process you see the message below. Just Start Windows Normally. You may see this dependent on where Windows was at at the time the snapshot was taken. Snapshots taken at the array level are called Crash Consistent snapshots. At the time the snapshot was taken, the array is not aware of whether there are any open files on the server, if those files are being written to, etc. So when you go and restore from this snapshot, and boot the system up, the OS will look at the data and may interpret the state of the system as recovering from a crash.
Note that the volume was recovered from a snapshot (48 hours old), resulting in a desktop that is missing the TestDoc.txt file we initially created (and other folders present – that were there 48 hours ago).