Hyper-V Cluster with 2 nodes: IBM x3690 X5 with Intel Gigabit ET2 Quad Port Server Adapter added.
The four Intel ports have been teamed in VMLB mode. The physical server has no access to the team - only VMs communicate through it.
Driver version (team adapter): 126.96.36.199
Driver version (physical NIC): 188.8.131.52
Use NLB on two VMs (info: VM roles are Exchange 2010 CAS/HUB)
In order to use NLB on the VMs you need to enable MAC spoofing on the virtual NIC.
When MAC spoofing is enabled on the virtual NIC the following happens:
1) When you try to shutdown a VM with MAC spoofing enabled, the VM gets stuck with the status "Shutting down"
It is not possible to force it down in any way through VMM, Failover Cluster Manager or Powershell.
2) When rebooting the host server, the server get stuck on "Shutting down Cluster Service service". Now the only option is to power it off
or wait a long time - eventually it will "blue screen".
Now, we have only found one "solution": If we remove one NIC from the team, create a new virtual switch connected to that single physical NIC, and connect the two VMs to the new virtual switch, MAC spoofing, NLB etc. works perfectly.
This suggests to me, it could be a problem with the Virtual team adapter driver, or the teaming software....
Any suggestions or solutions are welcome
EDIT: (extra info)
After shutting down the host server, the Cluster log file contains hundreds of lines like this:
The NOS is Windows 2008 R2 Datacenter Edition with SP1
Message was edited by: ralp