Failed to prepare guest for freeze

[perjonsson1960]

Folks,

We have a backup job that backs up our virtual SQL database servers. There are 23 VMs in the job.
Every time the job runs, one of the VMs fails with "Error: VSSControl: Failed to prepare guest for freeze, wait timeout 900 sec"
It is always the same VM. And when the job finishes, and the first retry is performed 10 minutes later, that VM always succeeds.

Is there anyone who has experienced something similar? Since it always succeeds during the first retry, then why doesn't it work the first time?
It is not a big problem, it is just a bit irritating. I have tried to double the wait timeout, but then it just waits 1800 sec instead.

Kind regards,
PJ

[PetrM]

Hello,

I don't believe it is possible to determine the exact cause of the issue without analyzing the debug logs and conducting a remote session. Most probably, the issue is related to the Microsoft SQL Server Writer(check VSSADMIN LIST WRITERS output), which may not be able to prepare the application for freeze fast enough. You can find more information in the System and Application event logs. Additionally, you can try collecting vsstrace output during failed and successful runs and compare them. However, I strongly recommend troubleshooting this issue with our support team.

By the way, you mentioned that only one VM is affected. Maybe you can protect the SQL Server databases on this VM using our plug-in for Microsoft SQL Server? It does not use VSS, it performs SQL Server native backups to Veeam infrastructure over VDI.

Thanks!

[perjonsson1960]

Thank you!
I will try the vsstrace approach first, and see if I am man enough to interpret the log...

[Andreas Neufert]

you can check with Microsoft on what VSS component (VSS writer) will take that long.
Usually it is an unmounted database in SQL or any left overs that lead to very long VSS preparation times.
If you have a very large database under heavy load or hundreds of SQL databases on the system, to go in consistency mode, can take hours in some cases. It is best to check with Microsoft SQL team on this. In the interent there are many examples on how to trigger a VSS consistency without Veeam in the mix. That way you can avoid the fingerpointing.

[perjonsson1960]

Well, as I said, the backup of the VM in question works on the first retry without any delays. So it is really strange that it doesn't work during the job's initial run. It seems that this VM enjoys to be backed up alone, when all the other VMs in the job have already been backup up...