I have several 2 node MS clusters that are virtualized under ESX across two physical ESX servers. When I shutdown the public interfaces of my VM Network on one ESX server these nodes never fail over to their counterparts. In the MS Cluster log it shows that the affected server pings the default gateway and fails, but for some magical reason it finds another address, in this case an IP for a cluster virtual name, and considers the network still up.

I have searched high and low for anything on this. VMware tech support is telling me I have to script the shutdown of the vSwitch that these nodes are connected to for this to work. I am having issues accepting this, and to add insult to injury I actually have one node that actually does fail over whereas 10 other nodes dont.

Ideas?

Steve