Resuming NDMP jobs?

[maxsven]

Hi,

I am trying to understand if there is a way that NDMP policy based jobs that fail can be restarted without having to backup the whole thing again.

I got multiple jobs that failed after running out of tapes (status96)

I am aware that NDMP policies can’t use be configured to use a checkpoint. So I am not sure how they would handle restarting a failed backup from a specific point.

For all failed jobs that I had, when I right click the job, I only see a “restart” option. The “resume” option is greyed out.

I have loaded new scratch media, how can I resume those jobs from the point they failed without having to restart the entire backup? (or how would I do that in the future).

Running Netbackup 7.5.0.4 on Windows Server 2008 R2

NDMP host is NetApp.

[watsons · Accepted solution]

"resume" option is tied to the "checkpoint" feature. So if your policy supports checkpoint, you can suspend and resume it later, obviously not NDMP type.

There is no way Netbackup can do that, not unless the NAS itself supports such feature because Netbackup is merely calling the NAS commands to perform backup/restore. 

To keep your scenario manageable, I would split the backup selection into a smaller directory (instead of volume) and this way if a backup fails, you can restart just the directory you need to.

[maxsven]

Thanks,

yes I am aware there is no checkpoint option. however, there should be some sort of a way that it keeps track of where the back is at... otherwise how would it swap tapes in the middle of a backup when oe tape is full and it needs to use a new one? 

[watsons]

Talking about the tape spanning mechanism, Netbackup uses the image fragments (breakdown of image), so a big image may span across many tapes by means of fragments.

Remember this is how Netbackup manages the image, but of how it applies checkpoint on top of this I am not sure - my guess is it may apply to typical filesystem data, but not to all. Given that NAS usually has its own "compression, deduplication" embedded into the filesystem, it could pose a challenge for the checkpoint.

[Dan_42]

NDMP tape changing doesn't rely on a checkpoint. NDMP implements a "pause" type command that the client can issue. 

[maxsven]

So are you saying that once an NDMP job fails it will have to be restarted from the beginning?

[Marianne]

Correct.

Status 96 is easy enough to prevent.
Use available_media script (in goodies folder) on a daily basis.