A customer had 17 ESX servers in stress, after a thermal error 3 blades went off line and disk 0 was gone. HA gave an error on all servers and reconfigure HA didn’t do the job.
An error occurred during configuration of the HA Agent on the host was the error. I disabled HA on the cluster and enabled it, afther this action ervy server was back online and no error’s. I suppose that the setting that only 2 servers can fail for HA was course of this error, 3 servers went down at the same time and VMware ESX did’t see the old servers online therefor it was not allowed to reconfigure the hosts for HA. Disable HA and enabling HA pushes an new HA settings set to the ESX servers and makes it possible to reconfigure it for HA.