v9 SQL Transaction Logs 'Transaction Log Was Not Backed Up'

[lightsout]

All,

Got an issue, and not quite sure how to deal with it.

I've got a bunch of jobs going to a SOBR, and the extents have a limitation on the max number of jobs running. And they all run at 22:00. Several of these jobs have SQL transaction log backups set every 60 minutes.

These SQL transaction log jobs run OK, but they all go to the warning state. Looking at the logs, the SQL transaction log backup job starts at 22:03. As the main back jobs are running already, they consume all resources on the SOBR and so it is eventually approx 11:20 before they actually start. This means they miss the first 60 minute window, all jobs are then marked as "transaction log was not backed up". Not a big deal, but all of my SQL transaction log jobs are marked as being in a warning state each day because of it as a result.

I didn't believe this happened in v8 either. Is there anyway I can fix this? Beyond trying to juggle my scheduling, or allowing a lot of jobs to SOBR I am not really sure how.

Any ideas? Thanks!

[foggy]

Need to confirm, however, my understanding was that transaction logs backup jobs do not need free repository slots for their execution. Seems this is not the case with scale-our repositories, this could be simply overlooked. I will ask someone from QA to verify this on Monday.

[veremin]

According to QA team, SQL log job should not take the specified counter into account. Thus, the described behaviour looks unexpected. We'd appreciate if you open a ticket and post its number here. Thanks.

[lightsout]

Case #01700670 submitted.

[veremin]

Information has been passed to QA team. Let's wait and see what support engineer says after logs' investigation. Thanks.

[lightsout]

OK, so the issue seems to be not on the repository, but on the proxy.

In v8, the SQL transaction logs didn't count towards the proxy's job counts it seems. In v9 RTM, they currently do. So I left my proxy concurrent task limit alone, but lowered my concurrent repository tasks, and that fixed the issue by making sure there was always a task slot available for my SQL transaction log jobs.

The investigation continues if this is the intended behaviour or not!