[one-users] VMM 'unknown' state not being set?

[Carlos Martín Sánchez]

Hi,

On Fri, Feb 20, 2015 at 1:15 PM, Andy Coates <andy.coates at gmail.com> wrote:

> We tried 'p' but it means we can't cancel the VM.  We have to resume the
> VM just to stop it, and then cancel it.
>
> Feels like we're going about something the wrong way if we're unable to
> monitor a VM's state to show it stopped, and thus cancel/clean up the VM.
>

Maybe the problem is with the nomenclature.

You said you wanted to report the VM is stopped. To OpenNebula that means
that the VM disks and the memory state have been saved back to the system
datastore, plus the networking is cleaned up, etc. That's something that
the VM can't do by itself, so we don't consider the option to report that
from the poll info.

If by stopped you mean that the VM was shut down, then the drivers can
report 'd', and the core will move it to the poweroff state.
In any case, a VM in the suspended or poweroff states can be deleted
without having to resume it first.

At least that's how it works for 4.10, it may be a bit different in 4.6.


Regards
--
Carlos Martín, MSc
Project Engineer
OpenNebula - Flexible Enterprise Cloud Made Simple
www.OpenNebula.org <http://www.opennebula.org/> | cmartin at opennebula.org |
@OpenNebula <http://twitter.com/opennebula> <cmartin at opennebula.org>



> On 20 February 2015 at 17:29, Carlos Martín Sánchez <
> cmartin at opennebula.org> wrote:
>
>> Hi,
>>
>> On Wed, Feb 18, 2015 at 11:37 AM, Andy Coates <andy.coates at gmail.com>
>> wrote:
>>
>>> Thanks Carlos - I thought 'd' would be marked as something else.   We're
>>> basically trying to flag a stopped VM but with those limited STATE options
>>> it doesn't seem possible.  How can we represent a 'stopped' VM in the poll
>>> information?
>>>
>>
>> If the driver returns 'p', the VM is moved to SUSPEND. But you can't tell
>> opennebula that the VM is STOPPED, this needs to be started from the core.
>>
>> Regards
>> --
>> Carlos Martín, MSc
>> Project Engineer
>> OpenNebula - Flexible Enterprise Cloud Made Simple
>> www.OpenNebula.org <http://www.opennebula.org/> | cmartin at opennebula.org
>>  | @OpenNebula <http://twitter.com/opennebula> <cmartin at opennebula.org>
>>
>>
>>
>>> On 18 February 2015 at 16:58, Carlos Martín Sánchez <
>>> cmartin at opennebula.org> wrote:
>>>
>>>> Hi,
>>>>
>>>> On Mon, Feb 16, 2015 at 9:51 AM, Andy Coates <andy.coates at gmail.com>
>>>> wrote:
>>>>
>>>>> Running OpenNebula 4.6.2
>>>>>
>>>>> We have a custom VMM driver that is set to report 'unknown' state if
>>>>> we can't determine the current VM state on the host.  According to the docs
>>>>> the '-' character indicates unknown, and as shown below ON is seeing and
>>>>> parsing the VMM data correctly:
>>>>>
>>>>> [VMM][D]: VM 1135 successfully monitored: USEDMEMORY=4194304
>>>>> USEDCPU=800 NETTX=0 NETRX=0 *STATE=-* IP=10.0.0.174
>>>>>
>>>>> Am I missing something?  Shouldn't the VM state be set to unkn/UNKNOWN
>>>>> in this case?
>>>>>
>>>>> Thanks.
>>>>>
>>>>
>>>> The '-' character indicates a monitoring unexpected error, like when
>>>> the monitoring probes could not parse the output. In this case OpenNebula
>>>> does nothing, the current VM state is kept.
>>>> A state='d' (disappeared) indicates that the monitoring could be done
>>>> (hypervisor is up), but the VM was not found in the list of VMs. In this
>>>> case OpenNebula will set the VM to unknown.
>>>>
>>>> See:
>>>> http://docs.opennebula.org/4.6/integration/infrastructure_integration/devel-vmm.html#poll-information
>>>>
>>>> Regards
>>>> --
>>>> Carlos Martín, MSc
>>>> Project Engineer
>>>> OpenNebula - Flexible Enterprise Cloud Made Simple
>>>> www.OpenNebula.org <http://www.opennebula.org/> |
>>>> cmartin at opennebula.org | @OpenNebula <http://twitter.com/opennebula>
>>>> <cmartin at opennebula.org>
>>>>
>>>>
>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20150225/9905fff7/attachment.htm>