SQL log truncation even though SQL data not backed up

[rasmusan]

Hello

Just wanted to ask a question regarding behavior in regards to SQL log truncation, which I think is a bit unfortunate.

We had one customer who set up their VM backup job, with SQL backup with default settings of log truncation on full backup (no periodic log backup). Job runs fine - no errors/warnings.

On 1 of the SQL servers, the data+log disks was attached as pass-through disks (hyper-v environment). And an SQL admin had created native SQL full backup + periodic log backup, to ensure proper backup of SQL data and to be able to roll back to point in time..... (of course its a mistake, that Veeam application processing was not adjusted accordingly, to use copy-only for this VM)

The pass-through disks of course prevented the Veeam job from backing up SQL data+log, but SQL transaction logs was still flushed by the Veeam backup job.... and no error/warning was to be seen in the job log.

Now this caused the customer, to be unable to roll back SQL DB to point in time, because the Veeam job did flush the SQL log, leaving a "black hole" in the SQL log trail...


My point is, shouldn't there be some kind of "check" in Veeam B&R, to ensure that if it is unable to backup the actual SQL data for some reason, then it should not touch the SQL server transaction logs? and maybe also a method to verify if any other kind of SQL backup exists for a particular SQL server/DB (more complex than it sounds I know) ?

[Rick.Vanover]

Was the pass-through disk thru the iSCSI initiator in Windows of the Guest VM?

[rasmusan]

just checked, but it's not - it's passthru disk on hyper-v host

[foggy]

Hi Rasmus, the behavior is currently expected, you should use db exclusions to avoid logs truncation in this case.