VMware Quiescing Error



Findings:

In vSphere/vCenter 6.0 manual initiated quiesced snapshots from vCenter and/or the backup solution Commvault initiated quiesced snapshots are failing on some VM's intermittently.

The Error:

The snapshots would fail with the following error:

An error occurred while saving the snapshot: msg.snapshot.error-QUIESCINGERROR.




Discovery:


A ticket was opened with VMware support and they deemed the issue to be that the VMware VSS writer included with the installation of VMware Tools was not powerful enough to freeze all the activity within the guest in a reasonable amount of time for the quiesced snapshot to take place. They were seeing higher than usual guest I/O activity at the time of the failed snapshots.

Resolution:

VMware’s work around is the following:



  • Uninstall VMware Tools
  • Reboot server
  • Reinstall VMware Tools with the custom option and disable the Volume Shadow Copy Service (Doing this forces the VM to use Windows Native VSS writer instead of VMware’s)
  • Re-enter IP information on adapter since uninstalling VMware Tools wipes that configuration




Additional Notes and Q/A:



Why is the VMware VSS Writer failing to quiesce and take the snapshots, but Microsoft’s VSS Writer is able to complete the process successfully?
The VMware VSS driver is significantly less powerful than the native Windows VSS driver and has more specific logic to freezing activity on the VM. In the case of a VM that simply has too much activity for the VMware VSS driver to freeze in a reasonable time, the Windows one can handle higher I/O more efficiently.

Will this issue be fixed in a later release of vSphere/vCenter?
The error that we are receiving is not a bug, but a limitation of the software on VMware Tools and the Windows box. Therefore it will not be fixed in a patch or future release, but most likely will be improved to increase the limits and speed of the VM.

If VMware’s VSS driver is enabled by default, what is the benefit of it, especially if the native Windows VSS driver is more powerful?
Not all versions of Windows have the VSS driver. VMware Tools installs it automatically to make sure that you can take a quisce snapshot regardless of the OS.

Does anything in particular rely exclusively on the VMware VSS driver as opposed to the Windows VSS driver?


No, the VMware Tools driver does everything the same as the Windows VSS driver. The host also doesn’t care what VSS driver does the snapshot, it just needs it to finish in a timely manner.

The below script will install/upgrade VMware Tools without the VSS driver, and the two NSX components that we don’t use. This one will not reboot the VM automatically.

/S /v "/qn REBOOT=R ADDLOCAL=ALL REMOVE=VSS,NetworkIntrospection,FileIntrospection"

The below script will do the same exact thing as the one above except this one will reboot the VM when it is complete.

/S /v "/qn ADDLOCAL=ALL REMOVE=VSS,NetworkIntrospection,FileIntrospection"
  
I used page 20 and 21 of the below link to give me the parameters/switches of the drivers.

Also the /S requests a silent installer, /v lets the application know that you’re going to pass switches directly to the MSI, /qn is MSI-speak for no user interface, and REBOOT=R is to ReallySupress the reboot. 





Comments