roland_geiser
Enthusiast
Enthusiast

unable to migrate vm though the underlying esx is powered off

Hello

I had the following issue on my farm (6 ESX 3.5U4 Hosts, managed by VC 2.5, HA+DRS enabled):

A host was going down due to a defective NIC. HA has moved the VMs to the other ESX hosts. But one single VM remained on that host. The host was marked as "not responding". The only available menu entry on that ESX host was "remove host". Every other entries have been grayed out. The vm was grayed out also (I hope I remember correctly) and on the vm's menu I could also not choose some helpful operations. Everything was grayed out - I think except the "Snapshot" point.

OK I was able to fix the server. And when the server was up and running again, it was visible in VC normally and the VM has become a normal status etc. So I was then able to manipulate the vm again.

But the question is: What, if I am not able to fix the defective ESX Server in an acceptable time frame? How can I shut down this "running" vm (on a shut down ESX!) and move it to another ESX in my farm? I tried to restart VirtualCenter-Service on my VC. This was not helpful. Also I opened a putty session to the sc on my other hosts to find out, if the vm was hosted on one of them. I tried this with the command cat /proc/vmware/vm/*/names, but this VM was never listed...

Has anyone ever had this issue also? How to solve the problem?

Best regards

0 Kudos
2 Replies
jrenton
Hot Shot
Hot Shot

If you browse the datastore where the VM resides and right click the .vmx file and Add to Inventory you will be able to register the VM on one of your hosts which is powered up.

roland_geiser
Enthusiast
Enthusiast

Hello

Yes. This is probably a possibility. But what if the vm is still registered in the cluster somewhere? In that case, I assume there would be an error message that means "the machine is already registered".

When I restarted the fixed ESX host, the vm was "green" and starting ... everything was ok automatically. So I think the vm was registered always "somewhere" in the cluster. If the machine would not have been registered, would'nt it be grayed out after the ESX host's reboot and probably be reported as "orphaned"?

0 Kudos