SMB Share connections lost during Veeam backup

[svp007]

Hello
We have problems backing up a large Hyper-V fileserver (12 TB - Windows 2012); Veeam B&R is patched to last version.
During the create snapshot phase and before any data is being backed up all SMB share connections are lost as a side effect; all other services such as RDP, PING and so on continue to work. This is of course unacceptable; restarting the Hyper-V host and Veeam backup server is the only way to restore operations to normal.

I opened a support case (support ID# 00993152) more then 1 week ago but they have no idea at all. Being passed last Friday from first level to second level support but 2nd level support starts over again without taking into account the steps already taken during last week. Not happy at all with how this is handled by support. We migrated from BackupExec to Veeam as they were known for there excellent support but very frustrating until now.

Have any of you seen the same problem ? SMB connections lost during backup ? any ideas/leads to solution ?

Best regards

Stefaan

[foggy]

Stefaan, the first question would be whether you can observe similar behavior while manually copying data via diskshadow, without Veeam B&R in picture. I believe your 2d tier engineer will be able to verify this with you.

[svp007]

The diskshadow tests have already been done during last week as requested by Veeam support.
During these tests SMB connections are not lost.

[Vitaliy S.]

What about Windows Event log for this VM and the Hyper-V host? What errors do you see there?

[svp007]

All Windows Event logs have been uploaded to support. I also checked them manually but no related Event when the problem occurs.
What is really strange that he problem is occuring on the VM guest; the VM guest snapshot works fine and from that point on backup communication is taking place between Veeam backupserver and Hyper-V host (backing up from Hyper-V host snapshot).
The VM guest is no longer playing a role in my opinion ... but the side effect is happening on the guest (all SMB connections lost - SERVER service hung).
Support ask me last night additional command output and I've uploaded it this morning.
I will wait for their reaction.

[svp007]

Veeam backups (full backups - based on host snapshots) were causing SMB connection loss in the guest (host OS = Windows 2012 Hyper-V) on our 12 TB fileserver; the problem appears after 10-120minutes (varies). Also strange was that only SMB2/3 connections were lost; SMB1 connections continue to work.

After long debugging the problem was caused by using virtual disks in the guest connected to the Hyper-V virtual scsi controller ! Switching to IDE-disks (connected to virtual IDe controller) avoids the problem but no real solution.

BUT swithching to IDE disks is not what is recommended by Microsoft when virtualizing e.g. Exchange 2013.
Extract from the "Best practices for Virtualizing & managing Exch 2013" (update July 2015) :
"The virtual IDE controller must be used for booting the virtual machine; however, all other drives should be attached to the virtual SCSI controller. This ensures optimal performance, as well as the greatest flexibility."
The same logically applies to large fileservers.

So switching to IDE-disks results in performance loss ... but that's unfortunately the only solution to avoid problems caused by Veeam Backup.

I personnally don't have a spare server (with sufficient diskspace) to check wether the problem is solved on a Windows 2012R2 Hyper-V host. Veeam support also has no intention to do this.

Veeam should take this very seriously and look into that but they have no intention to do this.
I suggested Veeam adding an article to their knowledge base but they refuse it.
I do not understand at all why Veeam is refusing to add a knowledge base article helping their paying customers to avoid side effect problems caused by their own software.

This forum seems to be the only way to get other customers informed.

[PTide]

Hi Stefaan,

As Tier2 support engineer has already said, a further discussion with a QA team is needed in order to find out if QA team can assign an engineer to reproduce the described behaviour. Without that done we can not know for sure if the issue is Veeam related or Hyper-V related, or both, therefore placing a KB article is not possible at the moment. You ticket is still opened and, in the last line of their latest email, support has asked you to perform one more test. Kindly, keep working with your support engineer.

Thank you for your patience.