Dear All,
I have 4 Node Cluster, All are having same build version and EVC mode is enabled.
Whenever I am trying to Migrate some of VM’s from Node No 2 to Node No 3.
I am getting below error.
The target host does not support the virtual machine's current hardware requirements.
To resolve CPU incompatibilities, use a cluster with Enhanced vMotion Compatibility (EVC) enabled. See KB article 1003212.
com.vmware.vim.vmfeaature.cpuid.ssbd
The warnings is saying that Node2 and Node3 have different cpu generations (for instance Haswell vs Nehalem) and or even brands (amd + intel)
The quick&dirty way to resolve this is to powerdown the VM and try again.
The other much more nice way is to configure EVC on the cluster, so all hosts in that cluster run the same virtual processor generation
It might require to powerdown some vm's on that cluster though...
Derek.
Have you fullfil requirment for Vmotion? Kindly explain what you have done for vmotion.
Try changing the EVC mode to the mode it hardware support
Hi,
Were your ESXi hosts updated recently? Are all hosts in cluster on exactly same ESXi build version ? Did you apply Spectre/Meltdown fixes ?
Match CPU features from source and target host by comparing output of below command:
cat /etc/vmware/config | wc -l
If the output is different, then it seems like your EVC needs to be refreshed by disabling EVC and re-enable it on ESXi cluster level.
Even if the output of command above is same, you can try refreshing EVC.
Dear Pragg12,
Thanks for the response. I have checked on all the nodes in the cluster and its observed that issued node having -1 count.
As you said I need to re-enable EVC mode on the cluster. But its say "To re-enable EVC, it might be necessary to power off running virtual machines."
This cluster is a Production and can't power off my any VM's.
Please suggest any alternative solutions.
Regards,
Premnath Bangar.
Yes, I can understand the pain here. Please ensure that all your ESXi is having same hardware/cpu model. Please go to Summary and check this out,
Unfortunately either you have to evacuate the VMs out of the cluster which will be bit difficult considering it the production environment.
You are left with only option to shutdown the VMs and try to do vMotion again.
EVC considerations has to be taken during the cluster creation itself. Once you onboard Vms on the cluster it becomes difficult to reset the EVC settings.
The missing feature is Speculative Store Bypass Disable (SSBD) which - as mentioned by pragg12 - is related to Intel Microcode patches.
If all your hosts are running the same Microcode (updated by either a vendor's BIOS update, or through VMware's microcode patch that's included in recent updates), then what you may try is to open the EVC settings, select another EVC mode (even an unsupported one), then switch back to the current one, and if it shows as compliant save the settings (otherwise click Cancel). This may already be sufficient.
André
Works perfectly if all the hosts are of same make/model.