Is there a way to force VMs from one host to come up on another host the next time the VM has a scheduled reboot?
We had a cluster with 10 blade hosts. It was not setup with EVC enabled. It's a long story. About 6 months ago, we added 2 new hosts. We'll call them hosts 11 and 12. Things always chugged along fine. We even upgraded all the hosts in the cluster along the way. Put any host in maintenance mode and the VMs migrate without issue. What we didn't notice is that a bunch of the VMs on host 11, would only migrate to host 12 and vice versa. If we try to move the VMs on host 11 and 12 to hosts 1 through 10, they fail. These VMs must have rebooted while on host 11 or 12 at some point and picked up the newer CPU instruction set. Not a huge problem. At least VMware knows enough to not move the VM to an older host.
We have 3 new hosts coming in for this cluster so, I need to fix this without rebooting any VMs if possible.
The least impactful way to do it I think, would be to somehow to force the VMs on host 11 and 12 to only come up on hosts 1 through 10 the next time they reboot. The next time the VM reboots with patching, the problem will solve itself. Pull host 11 and 12 out, turn on EVC for the cluster. Add hosts 11 and 12 back in.
Is there a better way to solve this problem?
Thanks in advance.
And by "We" I mean "Me"
Put hosts 1-10 in a group and use a VM-Host affinity rule?
Create a VM-Host Affinity Rule
What is the CPU platform of your Blade servers? are they same build and support same technologies? (all of them include of 1~10 & host 11 & 12)
Can you put them inside the a cluster when it's EVC-enable before adding the host?! If you have trouble, you may exprince same problem with the 3 new hosts too.
So I think the easier and faster way is wait for arrival of new servers, then create a new cluster object in your virtual datacenter and enable the EVC. Then move each of host one by one to this new cluster, and as you know you should power-off all the VMs of that host or migrate to the other hosts while you are moving it between the clusters.