Open Stack

Hi,

we are running OpenStack Mitaka on CentOS + RDO and use Ceph Infernalis as our main (and only) storage engine (glance, cinder, cinder-backup).

As far as I can see, Change 205282 [1] made it into Mitaka (stable), so my current task is making snapshotting faster. Our hypervisors boot from 16GB disks, currently running out of memory when snapshots use the generic method.

Reading the specs / blueprint, the new method should be used automatically but I think there are missing permissions resulting in falling back to generic snapshots.

Do I need to give cinder-backup rwx to volumes, vms and images?

client.cinder
        key: ...
        caps: [mon] allow r
        caps: [osd] allow class-read object_prefix rbd_children, allow rwx pool=volumes, allow rwx pool=vms, allow rx pool=images

client.cinder-backup
        key: ...
        caps: [mon] allow r
        caps: [osd] allow class-read object_prefix rbd_children, allow rwx pool=backups

client.glance
        key: ...
        caps: [mon] allow r
        caps: [osd] allow class-read object_prefix rbd_children, allow rwx pool=images


And last but not least: Am I able to disable generic snapshotting completly? It has no use for us.

Best Regards,
Kevin

[1] https://review.openstack.org/#/c/205282/