Hey!


I'm setting up a Windows Server 2008 R2 Hyper-V cluster, with iSCSI connections to an Open-E machine. I've got it all up and running, but the problems start at the automatic failover of Open-e.

I've configured 2 Open-E machines in Failover mode, with all replication tasks running and the virtual ip configured. I've connected the 2008 nodes to the iSCSI targets, formatting the volume, adding it to the cluster and creating a Virtual Machine on that clusterdisk. When I trigger the manual failover with the Virtual Machine doing a disk stress test, I can see the MB/s drop to 0. After about 10 seconds, the virtual machine crashes, saying it can't connect to the storage. In the Failover Cluster management tool I can actually see the Cluster Disks failing and directly coming back online.

I've tried setting a higher iSCSI and Disk timeout in Windows, but all the timers seem to have zero effect. When I connect directly to an Open-e server, without failover or a virtual IP, and I pull out the cable while running a VM, nothing happens. When I plug it back in, the Virtual Machine continues, as if there was nothing wrong in the first place.

It's almost as if the Cluster Service of Windows 2008 'knows' it's not the same machine who is responding to the Virtual ip...

Allthough it continues directly after crashing, the virtual machine actually do crash. And that's not my idea of a failover



How to solve this problem?

Thanks in advance!

Peter