Hi folks,
This has gotten on my nerves twice this week and am wondering if anyone has any explanation.
Admittedly I have done two foolish things that result in networking to a host bouncing without placing the hosts in maintenance mode. The first one was reconfiguring vSwitch0 (the only one) to use MTU of 9000. The second was enabling flow control on the EtherChannel port group on the Cisco switch side.
What happened each time was this:
- vCenter sends out an alert that the host may be down
- Host marked as failed
- Host comes back up, vCenter decides that the rest are down even though connectivity was never lost
- HA reports cluster has insufficient resources to satisy HA and marks all others as down
Disabling/Enabling HA on the entire cluster puts things back to normal. However it seems to me that if there was a link blip to a host in the middle of the night the entire cluster would be without HA until someone manually intervened.
Does anybody have any idea what is going on here
Thanks