Can anyone firm that Hardware snapshots work with the IBM V7000/SVC VSSVDS drivers using DPM? I have tried just about every configuration possible and nothing will reliably work.
I have a 4 node Server 2012 Hyper-V Cluster that was just brought online. I have applied all of the hotfixes required which at this point looks to only include (KB2838669).
The DPM configuration is DPM 2012 SP1
This CSV key is set like this, I have also tried ... (See Below)
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft Data Protection Manager\2.0\Configuration\MaxAllowedParallelBackups]
"Microsoft Hyper-V"=dword:00000003
This is just a piece of the DataSourceGroups file but I though maybe if I created 3 different Serialized groups I would see better results.
<Group GroupName="Group1-Sky-Cluster.Staff.local">
<Datasource DatasourceName="32A1DE45-4EBF-444D-B6FA-9892F82A09B2" ProtectedServerName="TESTVM1.Sky-Cluster.Staff.local" WriterId="66841cd4-6ded-4f4b-8f17-fd23f8ddc3de" />
</Group>
<Group GroupName="Group2-Sky-Cluster.Staff.local">
<Datasource DatasourceName="D234FD58-5EC3-49F6-B521-82CF065E10CF" ProtectedServerName="TESTVM2.Sky-Cluster.Staff.local" WriterId="66841cd4-6ded-4f4b-8f17-fd23f8ddc3de" />
<Datasource DatasourceName="62F2E05F-1848-493D-B541-C7DBFF594F70" ProtectedServerName="TESTVM3.Sky-Cluster.Staff.local" WriterId="66841cd4-6ded-4f4b-8f17-fd23f8ddc3de" />
</Group>
<Group GroupName="Group3-Sky-Cluster.Staff.local">
<Datasource DatasourceName="314380C4-EAEB-4BDB-B86A-30DAFFA61228" ProtectedServerName="TESTVM4.Sky-Cluster.Staff.local" WriterId="66841cd4-6ded-4f4b-8f17-fd23f8ddc3de" />
</Group>
Below was tried with and without datasourcegroups.xml
In my current test environment I have 4 VM's Using 2 Volumes each, I have 1 VM running on each of the nodes.
I have installed the latest VSSVDS drivers from IBM for the V7000 and VSC, I have verified that if ran on a single host they work. I had initially Created the FREE POOLS and RESERVED POOLS and used the same WWPN Reservations on each Node, this was as per my understanding of the documentation from IBM. After looking at the IBMVSS logs on the nodes I could see that when DPM started to take a backup that it would try to reserve a random disk that met the requirements from the pool. It looked as if the first one in would get the reservation and then the next ones in line were so close in line that they would also receive the same FREE POOL list and thus would end up in many cases pulling the same volume that was used be the first one, this will then cause it to fail. So in most cases one backup would complete but all other would fail.
I have since removed the datasourcegroups.xml and tried creating a separate pools of volumes for each of the nodes and mapping the pools with different WWPN's to the SAN. This also did not work, DPM started all 4 backups and the FREE volumes were reserved, but then DPM started failing all but one of the nodes and left the IBMVSS provider with the mapped volumes and nothing to do with them.
My next step is to add the VM's back to the datasourcegroups.xml and try that. I don't mind serializing the backups per node if possible but I can't possibly serialize the entire Cluster. This is why I created the 3 groups thinking this would stagger things enough that it would be able to complete 3 at a time. ???
Anyway if anyone has any insight into this please, help.
Thanks,
David