This may already be understood, but I have just recently reached an understanding of Cinder's HA status and felt it should be raised here. Using pacemaker, cinder-volume can be configured as active/passive using a shared storage backend and this configuration should work as expected. It was raised in this bug: https://bugs.launchpad.net/cinder/+bug/1322190 That attempting an active/active configuration where multiple cinder-volume instances that share the same host setting and operate together on the same shared storage should not be relied on and may stop working (depending on the backend driver). It is possible for the nodes to be out of sync on the current status of a particular resource or volume - and this could lead to problems. I think this may be relevant to Staypuft's HA deployment configuration.
Pull request upstream: https://github.com/redhat-openstack/astapor/pull/361 Detailed description of how i tested the fix with NFS backend: https://github.com/redhat-openstack/astapor/pull/361#issuecomment-53748283
Would like to pull in Fabio for review and make sure the HOWTOs are updated as well if this is needed.
Merged upstream.
openstack-foreman-installer-2.0.24-1.el6ost.noarch. cinder volume is A/P. openstack-cinder-volume (systemd:openstack-cinder-volume): Started mac848f69fbc643.example.com
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHBA-2014-1350.html