Capacity tier, are aborted uploads cleaned up?

[AM1]

Hi,

B&R v12.2, but it also did the same on v12.1.

SOBR with local perfromance tier, Azure hot capacity with immutability, move and copy policy.

When synthetic full backups are run to this SOBR, a incremental .vib file is first created, then it is fast cloned into a new vbk file. This is fine. But while the vbk full is being created, the SOBR offload job has already begun, and is proceeding to offload the incremental .vib file. Once the full is created, the incremental file is deleted.

This causes the offload job to abort the upload, normally after several GB of data is already sent to the capacity tier, with a message like:

Code: Select all

:: Unable to offload restore point [Machine name]D2024-09-03T121106_B558.vib: the restore point has already been deleted from the Performance Tier extent.  

A new offload job starts up, and then uploads the full vbk file to the capacity tier.

I would just like to get confirmation that the aborted partial .vib parts are removed from the Capacity tier and not orphaned in any way. I'm not sure if they would have had immutability set on thier parts or not.

Thanks, Andreas

[david.domask]

Hi AM1,

I'd actually ask you to please open a Support Case for this one as the behavior does not look expected. The offload process ought be aware of what's happening with the job and handle the Synthetic Full without such issues, so I think let's allow Veeam Support to review the logs and check the behavior.

Be sure to include logs for Support to review (use the 3rd radio button, and select the Veeam server itself to export logs from).

Please share the case number here once created.

Thanks!

[AM1]

Thanks David for the infomation, case created with logs: Case #07404602

[AM1]

Hi,

Support has checked into this, and said it is by design, the vib will be offloaded, and then aborted, and then the synthetic full will offload to the SOBR.

Two solutions they offered, active fulls instead of synthetics, and playing with the SOBR scheduling to avoid the active fulls.

I personally feel both of these options are non optimal, for a space and management perspective, surley this can be fixed??

Still trying to get an answer on whether those aborted uploads are cleaned up.

Thanks for any thoughts on this.
Cheers.

[Ivan239]

All blocks become immutable as soon as they are uploaded to the OS. Therefore, they remain immutable and can only be cleaned up after the immutability period expires. In your case, all incremental data will be present in the synthetic full backup, meaning the blocks will mostly be reused. So , it should not pose an issue in terms of used space.