Cluster with two CPU versions and EVC disabled

[MBouillon]

I have recently inherited a VMware environment (ESXi 5.5 / VCenter 6.0) that contains 9 hosts and approximately 500 vm's.  6 hosts have an older Intel E7-8870 CPU and 3 hosts have the Intel E7-8860 v3 CPU.  In this environment, approximately 200 vm's are stuck on the 3 newer hosts and cannot vmotion to the older hosts.

Since I have plenty of capacity to support the 200 vm's on the 6 older hosts, I created an affinity rule that should move a rebooted/power cycled vm from the newer hosts to one of the 6 older hosts.  I was hoping that when these servers when through their monthly patch cycle, they would migrate during the reboot.  Unfortunately this is not working and will only work if the vm is power cycled.

Is there a way to automate the power cycle when the vm reboots?

Thanks for any input.

Marty

[vHaridas · Accepted solution]

Am not sure if you can do this but you can try to automate Windows patch installation and power Cycle VM using VMware Orchestrator+PowerShell.

something like Jason did for updating VM template but you need to adjust it for your running VMs.

Virtually Jason: Updating VM Templates with vRealize Orchestrator, Part 1

If you can afford downtime for VMs running on those 3 hosts, I would suggest enabling EVC in Cluster -

Disable HA in existing cluster.

Disconnect those 3 hosts and then remove from vCenter.

Add disconnected 3 hosts in vCenter under datacenter ( not in same old Cluster)

Enable HA and EVC in existing Cluster which has 6 hosts and VMs running.

Now Power Off VMs running on One of the three ESXi Host and move those VMs to old Cluster and Power on those VMs again.

Once all VMs migrated from that hosts, add that ESXi host in DRS Cluster which has EVC enabled.

You can automate these task using PowerCLI.

Power Off VMs

Migrate VM

Power On VM

Put host in MM

Add host in Cluster

you can accomplish this task with maximum 5/10minute downtime for each VM.

At end of this migration, you will have HA/DRS Cluster with EVC enabled and then DRS will take care of balancing load across all hosts.

-

Haridas

[MBouillon]

Thanks Haridas.  Unfortunately Orchestrator has not been implemented in this environment and I am not familiar enough with it to deploy without first having a better understanding of it. 

With this scenario, I figured I would still have to schedule an outage for those VMs stuck on the 3 newer hosts.  Thanks again for your input!  I will continue looking into Orchestrator and will eventually have it implemented in this environment..

Marty