Open Stack

Hi,

I am working on an HA setup with multiple cinder-volume workers. However, I
have a single ceph cluster that I want both workers to front.

With relatively standard configuration, the volumes created seem to be tied
to a specific worker. This means, if a volume is created by cinder-volume-1
and that goes down, removing that volume would be stuck in delete state
until cinder-volume-1 comes back. It would be nice for cinder-volume-2 to
realize it's operating on the same pool as the other node, and thus remove
the volume right away.

Any thoughts?

-Simon
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20131122/911d8ff5/attachment.html>