Recommend configuration for backing up CSV on MSA2324sa

[amonje]

Hi everybody,

I am new to Veeam and the forum.
I've been using Veeam B&R to backup a 2 node Hyper-V cluster (W2k8 R2) with 2 CSVs on a SAS attached MSA2324 for a couple of months.
I am using the HP VSS hardware provider, and everything works fine. Backups finish without errors and I can see on the MSA that snapshots are created and deleted before and after the backup of each VM.
However, I believe this stresses the MSA, and after a few weeks, it crashes badly. I have a case open with HP, and they are looking into the issue.
In the meantime, what would be your recommendation?
Just using the Microsoft SW provider to take the snapshots?
I upgraded from v6.5 to 7.0, so parallel processing is disabled. I guess activating it would stress even more the MSA, isn't it?
Would be any way to configure the backup jobs in a way that only 1 volume snapshot were created for all VMs in the job (providing that all VMs reside in the same volume ?
Thanks in advance,
Antonio

[Vitaliy S.]

Hello Antonio,

1. Using hardware VSS provider is still a preferable option. See our online help center for complete list of differences.
2. Enabling parallel processing will reduce the backup window time, so the snapshot will be removed much quicker. BTW, what kind of errors do you have on the MSA when it crashes? Are you using the latest version of the hardware provider?
3. I don't think that it is possible, since Hyper-V itself creates multiple snapshots for VMs during backup operation.

Thank you!

[amonje]

Hello Vitaliy,

I will test enabling parallel processing then...
I am using M114R01 firmware on the MSA controllers and HP CAPI Proxy and HP VSS Provider 2.8.4.39 at
default behaviour, so my volumes are standard, and the provider converts them to master and creates the snap-pool before taking each snapshot. When the snapshot is removed, the provider converts the volumes back to standard.
The error I am getting is

Code: Select all

107 Critical Error: Fault Type: NMI p1: 0x0135838, p2: 0x0000000, p3: 0x0000000, p4: 0x0000000 CThr: Null 

When this happens, both controllers restart, so the volumes go down and the VMs are killed and restarted a few minutes later. Not nice...

Thanks very much!

[Vitaliy S.]

Do you have any error events in the Windows Event log? What is the state of the writer when this issue happens? Also I would appreciate if you could update this topic with HP support team response on the error message above, as I haven't seen it before.

[amonje]

I have the expected errors about the MSA going down (source HP Storage Agent), just a couple minutes later than the events 20003 and 20001 (source UserPnp) informing that the snapshots were mounted on the host.
Unfortunately, I didn't check the status of the writers (I guess you mean running vssadmin list writers) since the host rebooted from bugcheck 0x3b a few hours later. Anyways I believe the writers were stable, since I can see the entries of the CSV volumes going into redirected mode and back to normal afterwards.
I have sent all logs to HP, keep you informed...

[amonje]

Hello Vitaliy,

Finally, HP has replaced the storage controller A, they believe the problem is not related to firmware but faulty HW.
Let's hope it is the case...
Thank very much!

[Vitaliy S.]

Thanks for the update!

[amonje]

You are welcome!