openstack team mailing list archive

[Dennis Jacobfeuerborn <dennisml@xxxxxxxxxxxx>]

On 08.05.2013 02:34, Steve Heistand wrote:
> > 2013-05-08 02:13:17  WARNING [cinder.scheduler.host_manager] service is
> > down or disabled.
> > 2013-05-08 02:13:17    ERROR [cinder.scheduler.manager] Failed to
> > schedule_create_volume: No valid host was found.
> >
> > I'm running the api and scheduler services on the controller node and
> > the volume service on a compute node. Any ideas why cinder doesn't find
> > a valid host?
>
> Ive found that even small time differences between nodes can cause issues
> like this.

Yep, that was it. Ntpd was running on the systems but apparently it 
couldn't reach any timeservers. I set up another system as the time 
source to synchronize the two nodes to and now the volume is created 
properly.

Thanks!

Regards,
  Dennis