vSphere: VM “freezes” during the removal of a snapshot

If your VM is running on a NFS datastore and Changed Block Tracking (CBT) is enabled, you might experience the following symptoms:

When removing the snapshot of the virtual machine residing on the NFS storage in an ESX/ESXi 4.1 host, you may experience these symptoms:

• The virtual machine appears to be in a hung state within the console view
• The virtual machine stops responding to ping requests
• The virtual machine is inaccessible via remote desktop

But why is this VM freezing? The explanation is the locking mechanism which NFS uses:

This issue may occur if you are using a backup application that utilizes Changed Block Tracking (CBT) and the ctkEnabled option for the virtual machine is set to true. The virtual machine becomes unresponsive for up to 30 seconds as NFS locks on the .ctk file are changed.

The workaround is to disable CBT:

1. Wait for the virtual machine to become responsive and for the snapshot operation to complete.
2. Schedule an outage window for the affected virtual machine.
3. Shut down the virtual machine.
4. In the Inventory pane, right-click the virtual machine and click Edit Settings.
5. Click the Options tab and click General.
6. Click Configuration Parameters.
7. If the ctkEnabled  parameter is not listed, click Add Row, add ctkEnabled, and set it to false.
8. Power on the virtual machine.

This workaround can impact the backup performance because you can’t use CBT. To Disable CBT in a Veeam job, you have to edit the job and disable CBT in the Advanced Settings under the vSphere tab:

Or if you’re using PHD Virtual Backup you can change the CBT settings in the Options tab on the properties page of a Backup Job: