I have zombie VMs on my alerts page. How can I get rid of these?
First, take the identifier provided on the alerts page and search it. This will give you the VM page so you can verify that it is actually hosted on a different HV than the one on the alerts page.
If it's different, SSH to the HV listed on the alerts page.
You can verify that the VM is running on that HV by executing:
xm list <VM identifier>
If it's running there you can then shut it down gracefully by executing:
xm shutdown <VM identifier>
Give it a minute or so to actually shut down.
While this is happening, go to the storage tab on the VM page in the UI and select disks.
Add the disk number to the end of the address to get to the disk identifier.
Verify that the VM has been shutdown on the HV by re-checking xm list.
If it's shutdown you can then deactivate the disks.
Find the proper path by running:
lvs | grep <disk identifier>
This will give you the disk identifier followed by the data store, which will look like onapp-xxxxxxxxxxx
Then deactivate the disks on the HV by running:
lvchange -an /dev/<data store>/<disk identifier>
Once this is done, you can refresh your alerts page and this VM should no longer show as a zombie.