cancel
Showing results for 
Search instead for 
Did you mean: 

Empty backup of LUN disks in cluster with Dfrs

Andrea_Gasparet
Level 3
My customer used to backup the Clustername successfully. Since he activated Dfsr the backups of the volumes with shares will succeed but the resulting backup set is empty, even when excluding the Dfsr-Private folder. The other volumes, without shares backup fine. According to the dubious note on this technote the behaviour experienced by the customer is expected. https://www.veritas.com/support/en_US/article.TECH225917 NOTE: Do not select the DFS or DFSR drive letters under the "File Server" or DFS /DFSR Cluster Name. If the drive are selected under Cluster Name the data is automatically excluded from backup due to Active File Exclusion on the DFS / DFSR folders. Is backing up through the nodes a good alternative? According to the customer he runs into problem when the disk switches between nodes. Any design suggestion?
1 ACCEPTED SOLUTION

Accepted Solutions

Colin_Weaver
Moderator
Moderator
Employee Accredited Certified

Just for reference it appears that because Shadow Copy Components is not available against the actual cluster that the correct way to backup clustered DFSR is via the active node

http://www.veritas.com/docs/000023896

This of course does mean tifthe cluster fails over then that job will fail and you will need a different job (perhaps kept on hold) for when the other node is active.

 

If you go down the route of disabling Active File Exclusion be aware that:

- the backup jobs could try to backup files that do not need backing up (Swap files etc), which will cause a larger overall byte count and a longer job.

- It may also attempt to backup open files that are locked by application in a way that the open file technology cannot get round, which will result in lots of skipped files being reported and a generally slower backup

- for some databases technologies the VSS request against the file system might interrupt log sequence numbers and therefore cause issues for your database agent backups

 

 

View solution in original post

5 REPLIES 5

VJware
Level 6
Employee Accredited Certified

Afaik, backing up via the nodes is the only option.

What problem does he encounter once the disk switches ?

Andrea_Gasparet
Level 3

Let's say he prefers to use the old backup through Clustername he used to perform safely.

I am going to redirect him here.

Andrea_Gasparet
Level 3

 

How to disable the Active File Exclusion feature in Backup Exec

We will check whether this regkey is still valid for BE15.

VJware
Level 6
Employee Accredited Certified
Should work for BE 15 as well.

Colin_Weaver
Moderator
Moderator
Employee Accredited Certified

Just for reference it appears that because Shadow Copy Components is not available against the actual cluster that the correct way to backup clustered DFSR is via the active node

http://www.veritas.com/docs/000023896

This of course does mean tifthe cluster fails over then that job will fail and you will need a different job (perhaps kept on hold) for when the other node is active.

 

If you go down the route of disabling Active File Exclusion be aware that:

- the backup jobs could try to backup files that do not need backing up (Swap files etc), which will cause a larger overall byte count and a longer job.

- It may also attempt to backup open files that are locked by application in a way that the open file technology cannot get round, which will result in lots of skipped files being reported and a generally slower backup

- for some databases technologies the VSS request against the file system might interrupt log sequence numbers and therefore cause issues for your database agent backups