Until last week we were using cloudvirt1018 to host exactly two VMs:
toolsbeta-test-k8s-etcd-17
tools-k8s-etcd-13
cloudvirt1018 is having RAID problems (and due for refresh) so it's now drained and idle.
Each of the above VMs is part of a three-VM cluster that was spread over cloudvirt1018, 1019 and 1020. Now e have less redundancy as each cluster has two nodes on cloudvirt1019.
We will likely always need two 'fatvirt' hosts for the DB workloads on 1019 and 1020. Do we need a third fatvirt for those other two VMs? It seems like a lot of hardware to use on two VMs.