I think the first steps here would be to either file a bug, or a blueprint. Blueprint seems better because this is probably more than a couple of lines of code and has some interesting operational impacts.
Cheers, Michael On Fri, Nov 8, 2013 at 7:15 AM, Chris Friesen <[email protected]> wrote: > On 11/07/2013 01:56 PM, Daniel Speichert wrote: >> >> There is compute_nodes file in /var/lib/nova/instances with the >> hostnames of all compute nodes using that storage so it should be >> possible and easy to figure out if that is a shared storage or not. > > > Presumably this wouldn't help if you've only configured one compute node in > the system so far but plan on adding more. > > Also, this doesn't handle the "nova host-evacuate" case with a mix of local > and shared storage. > > If we were going to put a shared-storage entry in /etc/nova/nova.conf, maybe > it would make sense to give it a name as well....that way you could indicate > that a subset of machines share instance storage, and the scheduler could > limit itself to attempting migration/evacuation to other compute nodes that > use the same shared storage. > > Chris > > > _______________________________________________ > Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > Post to : [email protected] > Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack -- Rackspace Australia _______________________________________________ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : [email protected] Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
