Backup Copy to rotated drives retention problems

[AndraZ]

Since upgrading to V11 we have some strange behavior on our offsite backup copy jobs.
I've opened case 04875802, but it is a strange problem and want to share also here for possible input.

We have 2 external USB drives that make up rotating drive repository "Offsite Repository". We make backup copy of all our backup jobs every afternoon. Retention policy was between 7 and 12 restore points. We have only one of the drives connected at a time. We exchange the drives once per week.
Expected result and how it worked for the past 2-3 years:
On every drive there is an independent forever incremental chain. When drive was changed, the new drive's next incremental point is a bit bigger, because it holds changes from the last week (when this drive was last connected). It then keeps doing daily incrementals for a week and it is also merging old ones into the full at the end of the chain. It does not matter to one chain, that one week of incrememntals is on the other drive.

The new behavior: we had instances that Veeam deleted the whole chain on the drive after drive change. We suspected that retention policy now takes into account that on the old drive there is a chain that is long enough and so it doesn't need this chain on the new drive from 2 weeks before. And it deletes it whole and makes a new full backup, which takes too much time. But after chatting with support we concluded that this was not the issue. Retention policy on a connected drive works just as it did before and merging to the number of files on this specific drive independently. But after drive swap it makes some unexplainable steps and sometimes wipes whole chain on the newly connected drive.

With support we then suspected that after upgrade to V11 the first swap to a drive that was not yet seen in V11 causes wipe + full backup. Not really logical, but OK. It now wiped our drive for the third time - it was not a new drive. So this is also not the cause.

It gets stranger. Yesterday we swapped a drive and Veeam didn't notice. It was throwing errors that file is missing and stating the file from the other drive! After this failed step we did "sync now", and now it detected the drive change, but it wiped it, "per retention policy". However, this is the content of the drives (days of June):
Drive 1: 12-14, 22-28 - so we have 10 points of retention.
On 29th (yesterday) we changed the drive and before that:
Drive 2 had: 4-8, 15-21 and that was all wiped. It hold 12 points because during testing we were changing retention from 10 to 13 and back.

Looking at the logs I get the feeling that Veeam evaluates retention policy before it gets that drive was rotated, and based on the chain from previously inserted drive, but it deletes files on new drive...

Is there something we are missing regarding retention policy of rotated drives?

Thank you for the input and best regards,
Andraž

[foggy]

Hi Andraz, the backup copy retention behavior for rotated drives has changed in v11 - now you specify the total number of restore points that you want to retain on all the used drives.

[AndraZ]

This is the answer I got from support, yes. But it doesn't work like that for us. For instance, previous week we had drive1 in, that had these days of June: 9-14 and 22-27 (12 restore points on this drive). Drive2 had days 4-8 and 15-21 (12 restore points on this drive).

On 28th we changed retention policy from 12 to 10 to test it out merging and it merged drive 1 into 12-14 + 22-28 (10 restore points on this drive).

On 29th we changed the drive to Drive2 and it didn't notice so backup was not made.
Then on 30th we did a "sync now" and it noticed the drive change but it wiped 4-8 + 15-21.
Why? Because there were 10 restore points (12-14 + 22-28) on Drive1? But it didn't wipe Drive1 on 22nd, even though 12 restore points were on drive2 (4-8 + 15-21). I don't get this. And it doesn't make any sense to keep retention of 12 per drive and per set at the same time, because then everytime one drive hits retention count, it wipes all other drives?

This scenario happened 3 times now.
Here are screenshots of settings of the Copy Job and of the repository: https://drive.google.com/drive/folders/ ... sp=sharing
If you spot any weird setting.

Thank you and best regards
Andraž

[foggy]

The settings look correct but the described behavior does indeed look unexpected. Please ask support to investigate the reasons.

[Brad-Translucent]

Watching this. Is there any update?