I'm studying for VCP5 exam and I'm playing around in a lab environment with SAN datastores.
I have 2 Hosts setup in a cluster with HA enabled using admission control policy: host failures cluster tolerates 1 . I have each host setup with:
2 NICs in a vswitch for management/vmkernel
2 NICs in a vswitch for a server VLAN segment.
A single VM running on each host using the VLAN segment
[Scenario 1]
If I unplug both NICs serving the VLAN for host1, should HA failover the VM to host 2? I know technically host1 isn't down, its simply lost the link status of the VLAN its serving from those 2 nics.....I setup a continuous ping of the VM and as expected when I unplugged the 2 NICs for the VLAN the pings timed out, however I expected HA would move the VM to the other host and power it up. It did not initiate a failiover at all.
[Scenario 2]
As a different test, I unplugged the 2 NICs for the management/vmkernel on host1, and although the VM stayed up as expected, I expected a HA failover since the host was isolated. It actually did try to initiate a failover, but got an error about not enough cluster resources being available. I'm scratching my head because each host server has 32 GB of RAM and dual quad core processors and my test VMs are single CPU and 4 GB of RAM, so nothing huge.
I am adding a 3rd host server into the mix soon, but I still expected HA to work with the 2 hosts in the cluster. Can anyone shed some light and or point me in the right direction?
