Open Stack

Hello, Juha

I can't reproduce this bug in a devstack environment freshly installed
today.

Instance's power state should be synced in a very short delay window since
the following bp.
https://blueprints.launchpad.net/nova/+spec/compute-driver-events

What is the exact version of nova are you using?
Could you file a bug?









On Thu, Sep 19, 2013 at 9:04 PM, Juha Tynninen <tyky72 at gmail.com> wrote:

> Hi,
>
> Ok, seems to be I can eliminate the delay by editing the value for the
> nova/compute/manager.py related configuration item in nova.conf:
>
> cfg.IntOpt('sync_power_state_interval',
>                default=600,
>                help='interval to sync power states between '
>                     'the database and the hypervisor'),
>
> In the grizzly code there is also this default 600 seconds interval
> defined for this check, but still there wasn't this kind of delay
> present... I wonder what has changed.
>
> How do you see, is there some negative effects to set the
> sync_power_state_interval very low in havana (e.g. 1s or even lower)...?
>
> Thanks,
> -Juha
>
>
>
> On 18 September 2013 12:41, Juha Tynninen <tyky72 at gmail.com> wrote:
>
>> Hi,
>>
>> Previously in grizzly if I killed the VM process (in case of devstack the
>> relevant qemu-system process) nova
>> almost immediately noticed this and the status of VM was updated to
>> SHUTOFF.
>>
>> But now in havana it takes about 5 minutes the VM status to change to
>> SHUTOFF.
>> Anyone knows what causes this delay?
>>
>> Thanks,
>> -Juha
>>
>>
>>
>
> _______________________________________________
> Mailing list:
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
> Post to     : openstack at lists.openstack.org
> Unsubscribe :
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>
>


-- 
Cheers,
Jian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130929/ac6e5be2/attachment.html>