Hello everyone,
After a network outage, we experienced a situation of chaotic
connectivity between the oVirt manager (located on a VM outside the
cluster) and the hypervisors themselves.
During the incident, while only the connectivity between the manager
and the hypervisors was difficult, during the recovery we had
fencing of several of our hypervisors.
I'm looking for a diagram or text that describes all the rules for
triggering fencing.
Specifically, I have questions about the following (just in case):
* Does the manager play an active role in the decision to fence a
hypervisor?
Regards,
Renaud