Feature request: MS-SQL Always On - Back up logs from the replica thats available if the secondary is not

[mkretzer]

Yesterday we had a migration of a non-clustered SQL Server to a new SQL cluster (always on). DB Size ~8 TB.
All worked well and while the initial DB sync in the cluster ran (took 2-3 hours) i already started a quick backup of the active cluster node so we would have log backups working again ASAP. This backup took ~6 hours because the data was not really compressable and our backup storage was still highly loaded by backup copies from the night before. I expected the log backups to start working as the SQL cluster backup preference was set to "prefer secondary" - but since the secondary node was not yet backed up (at least not a restore point which held the replicated databases) it should in theory switch to the primary node for log backup.
But what happened is that the backup on the first node told us "Skipping databases to be backed up from another SQL Server node" and the second node log backup did not yet know those databases.

We had to manually switch backup preference to "Primary", after that log backups started working immediately.

Since "Prefer Secondary" is the default and no binding setting (primary is allowed if secondary is not avaiable) Veeam should be able to switch to the other replica!

Markus

[PetrM]

Hello,

The current behavior you're seeing is by design, the reason is that we check backup preferences at the level of SQL Server but validate node availability at the level of server/virtual machine that is added to a job. As a result, our selection logic just ignored the primary node because it thought that the secondary one is available.

I fully agree that we should improve this mechanism in future versions, however, I cannot comment on ETA so far.

Thanks!

[mkretzer]

Thank you, its not a must-have but perhaps something for V13

[PetrM]

Okay, I can take a hint!

Thanks!