VM Guest destroys itself after reboot

[gconstantine]

Hi,

An administrator powered off a Suse guest. The ESX host then deleted all disks and configuration to do with that VM.

/var/log/vmkernel records:

[22:04:11 susevm vmkernel ....3864: Killing self with status 0x0:Success

/var/log/vmware/hostd.log recorded a successful powerdown. Then:

\[22:04:16 'vmsvc' ...error] Failed to get bool property: 16000:Error: The system returned an error. Communication with the virtual machine may have been interrupted

then

[22:04:36 'ha-eventmgr' ... Event 22: Failed to power on susevm on vmhost in ha-datacenter. A general system eror has occurred"

then

...marking virtual machine invalid.

then

22:09:30 'taskManager'...]Task Created : haTask-32-vim.ManagedEntity.destroy-217

22:09:30 'vm:/vmfs/volumes/...../Susevm'....Destroy VM Called.

The whole VM has disappeared. They're talking about pulling the plug on this project. Any help would be appreciated.

[GBromage]

Were the virtual disk files actually removed from the data store, or was it just taken out of VirtualCenter?

If you browse the data store, can you seee them?

[wila]

Hi,

An administrator powered off a Suse guest. The ESX

host then deleted all disks and configuration to do

with that VM.

Never seen that happen before and i don't believe that one is the result of the other. Just powering off a VM won't remove it.

As GBromage said, check your datastore (hint: double click on it in the VI client) to make sure it is really gone.

I'd expect the files to still sit there, in that case, select the ".vmx" file in the lot and register it back to your Host.

Okay, so i've experimented a bit and one way to get this behavior is by rendering the .vmx file invalid as this will indeed mark the VM in the list as "invalid" after powering off. There are 2 ways in resolving that:

1. If the .vmx file isn't butchered too hard, just unregister the invalid VM (option: Remove from Inventory, do NOT choose delete!)

2. Create yourself a new .vmx from scratch by creating a new VM and point it to the existing virtual disk(s). This should bring your VM back in its original state.

Message was edited by:

wila, added the xperiment bit

[gconstantine]

Thanks for the responses.

Not only are the files in the datastore gone but the whole directory. I trawled through /var/log/vmware/hostd.log and watched the files, directory and whole VM be removed from the host.

This is one for VM support.

Graham

[GBromage]

Hi graham!

Yes, if the whole directory is gone this is definitely one for support. If you find out the reason, please post it back here because now I'm worried about my machines.......