cancel
Showing results for 
Search instead for 
Did you mean: 

Backup/duplication failure scenarios

NB-OPS
Level 4

Hi,

 

I am trying to understand the consequences of certain failure scenarios.

 

(1) I have to back up 3 volumes (vol1, vol2, vol3 in that order) via NDMP within the same policy which has weekly full on Saturday and daily incremental. Let's say during a weekly full backup, it failed in the middle of backing up vol2 and it's out of the full backup window.

 - Is the backup of the vol1 portion still good, ie, can it be used for future recovery?

 - If I re-run the backup policy manually by selecting weekly full schedule, will it do a full backup on vol1 again?

 - If I don't re-run the backup policy manually and wait for the next backup window which is a daily incremental backup, will it back up vol2 and vol3 in full?

 

(2) I have backups going to VTL which then get dupliated to physical tape library. If the tapes run out, duplication will fail. Will netbackup continue from where it left off? For example, if a duplication job is duplicating image x from VTL and it fails in the middle of it because there is no more tape to write to. After I put in more tapes, will it continue to write the data next in image x on the new tape? Or it will rewrite to the new tape from the beginning of image x?

 

Thanks,

 

2 ACCEPTED SOLUTIONS

Accepted Solutions

sdo
Moderator
Moderator
Partner    VIP    Certified

1) Yes the vol1 portion of the backup job is good for restore as long as that particular individual stream completes with status 0 - however if the job was not multistreamed and vol1+vol2 were consecutive in the same single job then the whole job is lost if it fails.

The next full willbe a full.

The next differential will still be a differential only as long as previous full still exists in the catalog.  If no previous full exists then the next differential will be an effective full.

2) If a duplication fails then the single image that was being duplicated at the time will not be duplicated.  If the duplication job contains multiple images then any images completely duplicated before the single image fails will be ok.  SLP will always keep retrying forever until eitheran image gets duplicated or it is manually suspended/cancelled from the CLI.  Any space used at the end of a tape by a failed image will effectively return to free, thus another smaller duplication later may succeed.  If no free tapes are available and the duplication of an image fails then the duplicated image is effecticely cancelled, and will have to restart from the beginning when more tapes are available.

View solution in original post

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified
1) It depends on whether all volumes are running as single stream or multiple streams (Allow multiple data streams in policy attributes ). If a separate stream is generated for each volume, then successful backup for vol1 will not be affected by vol2 failure. For single stream backup of all volumes, the entire backup will be discarded when vol2 backup fails. 2) Entire image x will need to be duplicated from scratch.

View solution in original post

2 REPLIES 2

sdo
Moderator
Moderator
Partner    VIP    Certified

1) Yes the vol1 portion of the backup job is good for restore as long as that particular individual stream completes with status 0 - however if the job was not multistreamed and vol1+vol2 were consecutive in the same single job then the whole job is lost if it fails.

The next full willbe a full.

The next differential will still be a differential only as long as previous full still exists in the catalog.  If no previous full exists then the next differential will be an effective full.

2) If a duplication fails then the single image that was being duplicated at the time will not be duplicated.  If the duplication job contains multiple images then any images completely duplicated before the single image fails will be ok.  SLP will always keep retrying forever until eitheran image gets duplicated or it is manually suspended/cancelled from the CLI.  Any space used at the end of a tape by a failed image will effectively return to free, thus another smaller duplication later may succeed.  If no free tapes are available and the duplication of an image fails then the duplicated image is effecticely cancelled, and will have to restart from the beginning when more tapes are available.

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified
1) It depends on whether all volumes are running as single stream or multiple streams (Allow multiple data streams in policy attributes ). If a separate stream is generated for each volume, then successful backup for vol1 will not be affected by vol2 failure. For single stream backup of all volumes, the entire backup will be discarded when vol2 backup fails. 2) Entire image x will need to be duplicated from scratch.