Open Stack

Hello Eugen 

Thanks for the response. I am pretty sure none was playing with virt-manager. Wil check libvirt logs and update back. 

Even enabled crashdump in vm which didn’t generate any at the time of vm shutdown.

In qemu logs there was just one message stating “ 2020-11-13 07:00:52.164+0000: shutting down, reason=crashed”


Sent from my iPhone

> On 18-Nov-2020, at 11:49 AM, Eugen Block <eblock at nde.ag> wrote:
> 
> Hi,
> 
> from my own experience the mismatch happened when someone was playing with virt-manager (or other tools) to reboot/shutdown VMs through libvirt and not from inside the VM or via openstack. And since nova has to catch up at some point this led to a cycle of unwanted shutdown events. In my case I know that users were trying to do that via virt-manager, does that apply to your scenario, too? Or does libvirt show any error logs that would explain unexpected shutdowns?
> 
> Regards,
> Eugen
> 
> 
> Zitat von Deepa KR <deepa.kr at fingent.com>:
> 
>> Thanks for the response.
>> Any idea what would have caused power state to mismatch in DB. Hypervisors qemu versions are upto date. We have many VMs and seeing this issue with only few VMs(say 3)
>> 
>> Sent from my iPhone
>> 
>>>> On 18-Nov-2020, at 8:57 AM, Mohammed Naser <mnaser at vexxhost.com> wrote:
>>> 
>>> This really doesn’t mean that the VM is shutting itself down. This just means that Nova has noticed that the power state of the VM doesn’t match the database one.
>>> 
>>> You would also see this error message in any scenario where the VM disappeared from libvirt
> 
> 
>