Generating Rollback files taking too long, OK to kill?

[TaylorB]

I have a synthetic full backup that has a few weeks worth of rollback files that it is trying to generate. The problem is that it is running at about 1% per hour thanks to high compression levels and a tape job on the same server. Is it OK to kill the job in the middle of rollback (using STOP in veeam), or will it leave my backups inconsistent? 1% per hour means it won't finish until Thursday which means no backups this week. That's not good.

[TaylorB]

Okay, I killed the tape job that was eating up IO on this server, but it is still very slow. I am still considering stopping the job.

[Gostev]

It will leave your backup files inconsistent yes, but all older restore points will be restorable.

[TaylorB]

I left it running one more night and it finished this morning at 32 hours. This is on a very high end server with fast storage, too. I don't know if the extra couple hundred GB is worth the hassle for the enabling the "transform previous backup chains into rollback" feature.

[Gostev]

Well, please keep in mind that this took very long long time because you have not had this option enabled right away on the job (my understanding from the above), so it had a few weeks of older backups to process during the first transform operation. Now that you have done this once, you will only have a few increments to process (1 week worth) during the following fulls, so this should be taking much less time. Does it make sense?

I recommend that you wait until the next periodic full and then make the decision based on its results.

Thanks.

[TaylorB]

Great advice on waiting until this week to make judgement on performance. It only took 3 hours this time. So I guess the lesson is to not allow the rollbacks to get too far off of schedule.