Endless tape backup - Case # 03511219

[rsouthwood]

I'm not sure how this happened but over the past couple of month's my end-of-month tape backup jobs are taking forever to complete and writing way too much data to tape. The repository is 14TB (12TB used) yet the tape job is writing nearly double that to tape. What happens is the tape job starts backing up the jobs, one or more jobs runs, and then the delta created causes the tape job to think that it needs to retry the backup, even if the source job was already backed up to tape.

The box to "Prevent this job from being interrupted by source backup jobs" is ticked but apparently doesn't work the way it reads. In my experience this only applies to the job currently being backed up to tape, not the overall tape job, which results in interminably long jobs and way too much data written to tape.

Support has told me that this behavior is "be design", which seems strange to me. To me the tape job should backup the source jobs, prevent them from interrupting tape backup, and not go back and retry if delta is detected. As in, Jobs 1/2/3 are backed up to tape. Job 1 gets written to tape, then it moves on to Job 2, and doesn't go back and retry Job 1 if it kicks off again.

[michael.pescuma]

Does your job ever stop? Or did it just keep getting larger and larger?

[rsouthwood]

Just kept getting larger and larger, like Leon in "Airplane!". Had to terminate the job, it had written 26TB to tape with no end in sight. Ended up breaking it into 5 smaller jobs, but performance really takes a hit for both disk and tape jobs if more than one is running at a time.

[Dima P.]

Hello rsouthwood,

Is that a simple backup to tape job or your are using GFS media pool as a target? Thanks!

[michael.pescuma]

Love the "Airplane!" reference. I'm encountering a similar issue and currently working with support to see if we can find a way around it.

[Dima P.]

michael.pescuma,

Please share the case ID in this thread as well. Thanks!

[Mgamerz]

I am not sure if I am encountering this error, but its been stuck at 99% and has been writing tens of TB's extra after updating to 4a this month as well. I also do end of month to tape.