Hi Community, We are trying to test the HA Feature in cloudstack where all VMs which have HA Enabled will failover to a second node, AND cloudstack will fence off the affected node using IPMI when the node is no longer healthy.
Of course, the most common way to simulate a host failure is to abruptly kill the power. However, in our experiences, in most cases in production, there scenario is where the node is not dead (cause that would be easy to tell), but it is affected in some way (Dying or sick, but not dead, which is a much harder way to detect.) Does anyone have any recommendation how else I can simulate a host issue where the host is problematic but not fully dead? Just to be able to see if cloudstack does pick it up, fence off the node successfully and restart the VMs in the new host. Regards, Bryan
