I've seen a change in the behaviour of the Tape and Copy Jobs since upgrading to 9.5.u3. This is causing us issues and I'm hoping someone can explain the new behaviour so that we can sort out a work around. I may then log this with Veeam support if it looks like a bug instead of a feature.
Here's what we setup in original 9.0 install and has been running ever since:
Site 1 - Weekly Full with daily incremental
Copy Job copies each restore point to Site 2, with the weekly full set to be copy of original Active Full not a synthetic one.
Once a month, a tape job runs to backup the latest Active full from the copy job on Site 2. We achieved this by specifying 'Process latest full backup chain only' in the tape job and in the Schedule for creating synthetic fulls, set to "Weekly on seleted Days" but removed ALL days. Worked well, the tape job throughput was good as no merge was taking place just a straight copy of the latest Active Full.
We noticed that the Tape jobs were now producing merged backups not just copying the latest active full. Bad news because the job is much slower doing a merge and due to bandwidth limitations the Copy job does not always complete the incrementals before the next one is due. This results in an incomplete backup chain and a totally useless tape backup which is indicated by a 'warning' message long the lines of
"Some VMs in the synthesized full backup file will not be restorable because <VMname>.vib contains one or more incomplete or corrupted restore points"
The copy of the Active Full backup is always completed which is why we like to use just that for the tape copy.
Investigating the Tape job settings it now seems that the "Weekly on seleted Days" has changed to having Monday to Friday checked and we are now unable to uncheck all the days, at least one must be selected resulting in a mergeded tape file. Perhaps we could select the day that the 'Active Full' is copied so that there would be nothing to merge. I'm just experimenting with that now but in doing so I have found another copy job oddity.
In the Copy job you specify the day for the weekly active full to be copied. Up to now that is the day that the Full appeared in the copy job restorepoint chain. Now, post 9.5.u3 it shows up as the restorepoint the day before the active full was taken in the original backup. I assume that means it merges the active full with the incremental for the previous day but no idea why one would want to do that. We certainly do not want to merge the file we are then copying to tape.
One other point that may be relevant here is that the backup repositries at both sites are HP StoreOnce Catalyst stores.
Hope I have included enough detail to make some sense! Would be very grateful for any insights anyone has into the above behaviour as I need to get our tape backups back on track.