This is a wonderfully frustrating error. Veeam have a few options for fixes, including registry modifications and updating to the latest patch releases. This is generally caused by the temporary VSS service (in this instance) used by Veeam to capture application aware processing.
Well, if none of this works, go back to the crazy option of:
Editing the job and un-ticking the Enable application aware image processing option, which is under the Guest processing section.
Retry the failed job. This should complete the failed VM’s successfully.
Edit the job again and re-tick the above option.
This error happened after a power failure, where my hosts powered down without the guests being powered down first.
Veeam was basically unable to freeze the VM’s to take a hot backup. The backup would complete however, but be extremely slow in the process. This is a result of Changed Block Tracking not telling Veeam what blocks have changed in an incremental (or full) backup.
A lot of sites, including Veeam’s KB page, have suggested changing the CBT parameters in the offending VM’s, but this involves shutting down the guests.
What worked for me was much less invasive.
Edit the existing Veeam job, and click on Virtual Machines
Click the Recalculate button to the right. I initially saw the calculated size was indeed zero, so it had no idea of the machine sizes it was backing up
Once the machines/folders have recalculated and you have sizes populated, click Finish. This should then re-populate the Veeam database.
Hopefully you’ll have no more warnings, and also have a much faster backup timeframe.