[one-users] VM shutdown - rescue from error state

[Shantanu Pavgi]

On Aug 13, 2012, at 4:19 PM, Shantanu Pavgi wrote:

> 
> A VM got into an error state during shutdown operation as shown below. 
> 
> {{{
> ERROR=[
>  MESSAGE="Error shuting down VM: Timeout reached and VM one-52 is still alive",
>  TIMESTAMP="Mon Aug 13 15:03:08 2012" ]
> }}}
> 
> The vm.log for the VM shows following messages: 
> {{{
> Mon Aug 13 14:57:57 2012 [LCM][I]: New VM state is SHUTDOWN
> Mon Aug 13 15:03:08 2012 [VMM][I]: Command execution fail: /var/tmp/one/vmm/kvm/shutdown one-52 kvm-02 52 kvm-02
> Mon Aug 13 15:03:08 2012 [VMM][E]: Timeout reached and VM one-52 is still alive
> Mon Aug 13 15:03:08 2012 [VMM][I]: Timeout reached
> Mon Aug 13 15:03:08 2012 [VMM][I]: ExitCode: 1
> Mon Aug 13 15:03:08 2012 [VMM][I]: Failed to execute virtualization driver operation: shutdown.
> Mon Aug 13 15:03:08 2012 [VMM][E]: Error shuting down VM: Timeout reached and VM one-52 is still alive
> Mon Aug 13 15:03:08 2012 [LCM][I]: Fail to shutdown VM. Assuming that the VM is still RUNNING (will poll VM).
> 
> }}}
> 
> The VM was shutdown successfully and it's not running on the hypervisor host. It seems like VM was shutdown, but OpenNebula didn't receive exit code in given timeout value. 
> 
> I haven't had such issue before and I am not sure what would cause such delay in response. I will look into libvirt logs as well for more details. 
> 
> Is there any way to change this VM's state from RUNNING to SHUTDOWN? Also, how do I update it's state on the host where it's being counted in running VMs list?
> 
> 

This VM was already shutdown on the hypervisor and after a proper shutdown through OpenNebula it would have entered 'done' state. Since it was already shutdown on the host, I decided to 'delete' it from OpenNebula and get it in the 'done' state. I am sure there are other possible approaches, but above seems to have worked fine for me. 

--
Shantanu