I trying to understand how I'm able to vMotion from ClusterA (HP Proliant G5 hardware) to ClusterB (Cisco UCS Hardware) without issue when a VM is @ hardware level 9, each cluster is sharing the same set of datastores.
If I attempt to perform the same test with a VM that's running hardware level 4 I get the message I would expect to see regardless:
But if I were to upgrade this same VM to hardware level 9 it would run vMOtion without issue. I feel like its a newbie question, but I'm stumped.
This problem is related to your servers processor. What is your Cisco processor model and HP processor model?
You can't migrate VM between different processor with different architecture, you can't do it when you have AMD on a cluster and INTEL on other cluster or servers.
You have to do cold migration. I mean, you have to move them when they are powered off.
If you have processor with same architecture but processor model is different, for example two Xeon processor from two different family, you should use EVC on your cluster or clusters.
Check these links:
This is related to EVC mode like Davoud pointed.
Refer this KB too
The hardware version of a virtual machine reflects the virtual machine's supported virtual hardware features. These features correspond to the physical hardware available on the ESXi host on which you create the virtual machine.
In your case, issue is with CPU compatibility. It seems CPU feature set available on source host (1st cluster) is updated as compared to destination host (2nd cluster) . Hence vMotion check is failing on destination host(target host) as some CPU features are not available on target host that source host has.
As per me, there is something more than only EVC issue that we need to understand as vMotion is working fine on the same VM when hw version is 9. I think, when hw version is 4, some of CPU features gets disabled on source host & it leads to CPU incompatibility between source & destination host. I will try to dig into this.
can you let us know:
1. are you using different versions of ESXi host in your cluster? If yes, let us know the details.
2. are you using different family/generation of CPUs.? Provide details.
3. are you performing vMotion when DRS is enabled, in this case DRS will choose the target host as part of initial placement ?
4. are you performing all these scenarios on same host (I mean, 1st host from 1st cluster & 1st on another cluster). It may be the case when you use different sets of hosts, vMotion some times works & some time does not due to CPU change across hosts.
However, it is always recommended to have latest VM hw version & vmware tools corresponding to ESXI host
Thanks for all the great feedback everyone. I'm well aware of the EVC mode benefits, and what it means not to have it enabled and thats why I'm so confused. vickyvision2020 in regards to your questions:
1) Version of ESX hosts is the same across each of the two clusters:
Source Host in Cluster A:
Destination Hose in Cluster B:
2) big difference in generation of CPU as seen above
3) Yes, DRS is enabled on the destination cluster
4) In my example, I've been moving everything to TSTESX06 on the source cluster (then performing vmotion) because it's a member of the destination distributed virtual switch.
Understood about always running latest hardware version, i'ts been slow since all my vm's end us creating a new NIC each time I upgrade hardware, so I have to go through the exercise of fixing that each time. From what I can tell, its a bug when going from HW4 to anything newer.
The bottom line question is why is it working with HW9 as I wouldn't expect it to work regardless since CPU levels are soo different in terms of age