Backups Hang on Merge at 96% or 97%

[cgracie@sygnet.ca]

Hello,

We just upgraded to v11 from v10, and most of our jobs are fine, however we have several jobs that run multiple times a day (one hourly, and others every 4 hours) and they take forever to finish. The problem is they are no longer finishing in time to get the frequency we need. This all worked totally fine under v10. The weird part is the jobs initially run and perform normally. The initial backup of each VM processes at normal speeds, and the merges start fine, however near the end the jobs just hang. There is no activity that I can detect for these jobs. It sits for hours at 96 or 97% complete on the merge.

I do have Case #04898149 open, but so far we do not have a solution.

Anyone else come across this?

Thank you.

[foggy]

Thanks for opening a case, I can see that our engineers have suggested trying a few optimizations - let's see whether they help. What kind of repository is it, btw?

[cgracie@sygnet.ca]

Thank you Foggy,

The suggestions have not helped so far, hence why I posted here. All of the problem jobs are running on Windows Server 2012 R2 systems and HP Proliant GEN9 servers. The repos are on NTFS formatted drives. Initially they did not have the "Align backup file data blocks" setting enabled, but I did so yesterday based on info in the forum here. That did not help. I enabled the two registry keys recommended last night for UseUnbufferedAccess (set to 0) and DisableHtAsyncIo (set to 1), but so far no difference.

Thanks, Cam.

[foggy]

I can see that the case has been escalated and there's also another case where similar behavior is observed. In that other case, the issue was in the number of encryption keys stored in the database and how Veeam B&R handles those during merges/transforms. R&D has already a hotfix for this situation so please continue working with the support engineer - they will verify whether you're seeing the same issue and the hotfix applies. Thanks!

[cgracie@sygnet.ca]

Thank you Foggy. We are working through the case at this time. A possible work around was also provided to run an Active Full backup on those jobs. I have done so since I had the space to do so, so now I just need to wait a day or so before merges start again. Otherwise I will have to count on that Hotfix.

[cgracie@sygnet.ca]

Just to update and wrap up this post, our problem was solved by a hotfix provided by Veeam support. Performing an Active Full was also a successful workaround for a few jobs we tried that on. Thank fully we had the space to do that. Now we can really enjoy the V11 goodness. It is performing quite a bit better than V10.