Application managed snapshots and HP Lefthand VSS Provider Backups on P4300

[jorgene]

Hi, i'm testing this on a 3-node W2012 Hyper-V Cluster. Everything updated to latest firmware/MPIO and SAN IQ 10.

Also i have tested with Veeam B&R Software as backup against the Cluster.

Now i have stubled into some strange behaivours.

If i backup a VM located on Cluster Node 3 (that node also is owner of the CSV/LUN which the VM holds it data on) (*.vhdx file) all works fine, and i see that i uses HP Lefthand VSS Provider for creating snapshots.

But if i Livemigrate the VM to Cluster Node 2 and runs the same backup job again it fails with

"Unable to create snapshot (HP LeftHand VSS Provider) (mode: Hyper-V child partition snapshot). Details: Unknown status of async operation
An expected disk did not arrive in the system.
--tr:Failed to create VSS snapshot.
--tr:Failed to perform pre-backup tasks."

And then it uses Microsoft CSV Shadowcopy Provider instead.

I have 2 dedicated NIC's by ISCSI with DPM for MPIO for each node in the cluster.

Also each hos has an ISCSI connection originating from each adapter IP to the VIP.

If i in CMC Edit the server object, there is a field named: Controlling Server IP Address,

here i only managed to put ONE of the Cluster Nodes IP addresses, tried to separate the IP addresses with . : and - but if i try to make a Application-Managed snapshot from CMC it only works with ONE ip, but according to the help you should put all IP addresses here:

(help needed for correct syntax with multiple IP)

Windows Failover Cluster Environment:

Physical IP address of each server hosting a VSS Provider in the Windows Failover Cluster.

I Suspect that this is what makes the backup fail from other nodes, ie snap shot request comes from wrong ISCSI IP address not listed in server object in CMC.

Has anyone got this to work?

[balblas]

I assume your nodes have a private IP address as well (not one of the iSCSI interfaces)? In the server configuration in the HP P4000 CMC you should use that address for each server node that you added to the Server Cluster config in the CMC.

[stp2]

I still cannot find comprehensive and clear information on how the 'controlling server IP address' property works when adding a new server to HP Lefthand. For example we have a 4 host Hyper-V cluster, each host server has 2 iSCSI NICs.

1. What is the syntax for adding multiple IP addresses in the 'Controlling server IP address' field? This not documented anywhere that I can find.

2. [balblas] Our Hyper-V cluster hosts do not have an additional private IP address in the same subnet as the SAN nodes or any otherroutable IP address other than their iSCSI NICs. Are they supposed to have this? If so where is this documented?

3. On our existing Hyper-V 2008 R2 failover cluster, none of the host servers have the 'Controlling server IP address' field populated on the CMC. It is left blank but yet they use HP Application aware snapshot manager 10.5 and the hardware VSS provider works ok. So why is this? Why is the 'Controlling server IP address' field required??

Thank you.