[pve-devel] [PATCH qemu-server v2 1/5] add create and remove reboot trigger subs

[Thomas Lamprecht]

On 05.09.19 13:00, Dominik Csapak wrote:
> On 9/5/19 12:42 PM, Thomas Lamprecht wrote:
>> On 05.09.19 12:27, Dominik Csapak wrote:
>>> On 9/5/19 11:47 AM, Thomas Lamprecht wrote:
>>>> On 23.08.19 10:55, Dominik Csapak wrote:
>>> it would only reboot it in case we can remove the file though,
>>> otherwise we risk that the user cannot shutdown the vm...
>>>
>> yes, all else is an error, but I would warn and say that reboot was ignored
>> because of "blah".
>>
>> Another possible alternative, in combination with the separate vm_reboot
>> method from my 3/5 review, could be to remove this early in the locked context,
>> before doing the stop, and in the case of a error on removal just abort and
>> do nothing - then the VM would keep running, maybe a better option?
>> As a reboot always has the end goal to have a running machine again, if you
>> stop it in case of error the goal is violated, if you do nothing it isn't fully
>> violated (as it may still be outdated on pending changes)
> 
> but we have to somehow detect a reboot in the qm cleanup step
> in which the vm is already stopped so that we can start it again.
> i am not sure how this should work if we remove the file before
> stopping? or am i missing something here?
> 

Hmm, yes, true.. such errors are probably quite seldom and a cause of
bigger issues anyway.