ESX3.5, VC2.5, with the following DFS/HA cluster:
6x Quad 2.0GHz, 16GB, iSCSI and FC Storage
2x Quad 2.6GHz, 16GB, iSCSI Storage
All of the machines have the CPU type, the same networks, and the iSCSI datastores are identical. DRS VMotions back and forth between all valid hosts without issue. I haven't set any CPU or Memory reservations on any VMs or resource pools. There are about 100 VMs total.
However, my cluster's "Current Failover Capacity" seems stuck at 1 host. When I put of the two 2.6GHz machines into Maintenance Mode, the capacity drops to zero and VC complains. When I put one of the 2.0GHz machines into Maintenance Mode, the capacity doesn't change.
I can't find any differences between those the two types of machines, but VC seems to think that the 2.0s are incapable of taking on the 2.6's VMs. Any ideas on where else to look?
Do you have any CPU or Memory reservations set?
You mean other than the obvious speed difference? I don't think that qualifies. On the surface it should work, but they are different model numbers and different speeds, so that's probably why.
You need almost exact machines for this to function.
Could this have to with primary and secondary cluster HA agents?
No reservations.
The CPUs are the same family, just different speeds. DRS works just fine in between those two sets of machines.
Not sure what you mean by primary and secondary cluster... I thought all machines in an HA cluster were equals?
Have you tried to remove the HA option from the cluster and then bring it back?
Regards,
David