An administrator has discovered a failure that has affected network connectivity on the
management network of a vSphere HA cluster. As a result, half of the hosts in the cluster are able
to communicate with each other but not the other half. The cluster was configured with default
settings.
What happens as a result of the failure?
A.
The VMs will be powered off on slave hosts in the cluster.
B.
All VMs and hosts in the cluster are unaffected.
C.
The VMs are unaffected, a second master will be elected.
D.
The VMs will be powered down on all hosts in the cluster.
Although host isolation response by default is power off, Wouldn’t the default datastore heartbeating allow all VMs to stay powered up?
Actually the default host isolation response is “Leave Powered On” according to “Mastering VMware vSphere 5” and my cluster in my dev environment which I just checked.
But a bunch of VMware docs state the default is “power off”…so, I’m not really sure what to believe.
This scenario is called Network Partition. As long as the other half are able to communicate among themselves, they elect a new master as management network communication within the partitions functions but between the partitions fail. The Original master continue to exist and performs it role in one of the partitions along with its slaves. The other partitions instead of declaring them down, vsphere HA, responds by electing a secondary master on the second partition. THis is partly due to the datastore they both connect, making it another form of communication among them though management network is down.