Mixed incremental and active full in one backup??

[pshute]

We run daily incremental backups with an Active Full backup on a Tuesday. Last night (Monday 9/7/18) one VM had difficulties and the job retried it later. It looks lo me like the retry ran as an Active Full, maybe because it ran after midnight Monday.

In the job history, I see "Backup job: Backup Job 1 - main (Incremental)" starting 9/7/18 8pm, finishing with a Failed status 10/7/18 2:24am,
then "Backup job: Backup Job 1 - main (Active Full) (Retry 1)" starting 10/7/18 2:34am, finishing with a Success status 10/7/18 6:56am.

In the backups folder, the vbk file is dated 4/7/18 10:12am, and the most recent vib file is dated 10/7/18 6:56am, and is about 10 times the size of the other vib files. I think it contains an Active Full of one of the backed up VMs.

Is this how it's supposed to work? I would have though a scheduled incremental of a VM should retry as an incremental, even if it happens to get retried on the day an Active Full is scheduled.

What will happen tonight when the Active Full runs? Will that VM get backed up again as an Active Full?

[pshute]

What will happen tonight when the Active Full runs? Will that VM get backed up again as an Active Full?

And if it doesn't get backed up in tonight's Active Full, will it end up on the tape backup we do tomorrow? That's configured to only include the latest full backup. Will it simply copy the new vbk file to tape, or will it somehow include this VM's full backup from the previous vib file too?

[pshute]

Looks like the same thing happened with a different server last week, and it looks like it did get included in the Tuesday night Active Full, so that's l good.

My issue then is just the amount of extra time and disk space used by the unexpected full backup of that server on Tuesday morning.

[foggy]

We run daily incremental backups with an Active Full backup on a Tuesday. Last night (Monday 9/7/18) one VM had difficulties and the job retried it later. It looks lo me like the retry ran as an Active Full, maybe because it ran after midnight Monday.

Correct, it was started on Tuesday, where the active full is scheduled at. Anyway, full backup is performed only once a day, so even if it is a full run during retry, the next (scheduled) run during the same day should be incremental.

[pshute]

It looks like it did another full backup to me. Perhaps it behaves differently like this when there are other VMs in the backup?

[foggy]

No, it performs full for the entire job. The fact that it creates VIB instead of VBK doesn't seem expected to me, though. I'm passing this to QA for investigation, but you can also contact support for a closer look.

[pshute]

I don't think it's entirely unexpected that the morning full backup ended up in a vib file. It's a late retry of a single VM that should have been included in the incremental with the other VMs the night before, so why would it create a separate vbk file just for that one VM? What's unexpected is that one VM in what's supposed to be an incremental gets a full backup just because it's doing a late retry.

[foggy]

If it were a full for just that single VM, but apparently it makes full for all VMs in the job. Anyway, we will investigate this.

[pshute]

Thnaks. To be sure you've completely understood, here's a summary:
- VMs a, b, c and d have a backup job scheduled with daily incrementals, and an active full on Tuesdays.
- On Monday evening a, c, and d got an incremental backup, as expected, including those that ran into Tuesday morning.
- VM b's backup failed, and it retried later Tuesday morning. Instead of appending an incremental to the vib file, it appended a full.
- The Tuesday night active full ran as normal, with all VMs including b got a full backup as they should.

End result:
- vib file started Monday night has incremental of a, c and d, plus a full of b.
- vbk started Tuesday night has full backups of a, b, c and d.

I feel the only issue is that a retry of a single VM should use the job type of the rest of the job, rather than determining it independently.

[foggy]

Thanks, Peter. This actually differs from how I initially understood what was happening. Let's wait for the results of our internal tests.