Rapid Recovery Error Transfer Failed, Timeout taking vss Snapshot

Recently we have been getting this error The transfer failed: 'Timeout taking VSS snapshot of 'servername'.'. This error isn't happening on every backup, around half of them per day though. It doesn't matter what time the backup runs. I've seen it succeed at 8 am then fail at 9 am, succeed at 3am, but fail at midnight.

Rapid recovery is giving a warning on one of the volumes in that server, the volume is below 30% free space (it has 29% free space left). I don't think that is causing the issue because VSS only needs 10-15% free space typically.

I did increase the VSS timeout in the registry to 6 minutes, and that looked like it succeeded when I did an on demand incremental snapshot, but then a couple hours later the scheduled snapshot failed with the same vss timeout error.

Any Ideas?

Thanks.

Parents
  • Is there an error in the windows event viewer (system or application) on the server that you are trying to backup (not the RR server)? 

    The RR GUI 'alert' about space is informational, VSS just needs enough space realistically to hold the changes that are made during the course of the backup. If it is a space issue you would see an event in the system event view with a source of volsnap (again on the server being backed up, not on the RR core). 

    Are there any other VSS tasks running on the server being backed up? Either shadow copies, or an SQL/Exchange task of any kind? 

  • There are no errors in the event log on the server in question.

    It could be an SQL task. I know that particular server is a SQL server that is heavily used/accessed during the day. Thanks, that points me in a direction to look at.

  • Ah, that could do it. If it is SQL then there could be other VSS tasks running. The other piece of the puzzle would be if you use RR to truncate the SQL logs, if not and you simply just need a block level backup without the SQL interaction, you could exclude some (or all) of the VSS writers to reduce the i/o load on the disks too. 

Reply
  • Ah, that could do it. If it is SQL then there could be other VSS tasks running. The other piece of the puzzle would be if you use RR to truncate the SQL logs, if not and you simply just need a block level backup without the SQL interaction, you could exclude some (or all) of the VSS writers to reduce the i/o load on the disks too. 

Children
No Data