While upgrading to ocata, we hit this bug: https://bugs.launchpad.net/cinder/+bug/1707936 Cu's ceph cluster is large. It has close to 400 volumes, some of which are very large. This is fixed upstream in: https://review.openstack.org/#/c/501325/4 Cu wants to update the openstack-cinder package to include this patch?
Again unsure about needed verification steps. 1. Deploy with Ceph (external in my case) 2. Create a few volumes, do we know many were needed to trip this 10 50 100 400 ? Does it matter if volumes are 1G or fewer volumes but larger sized? 3. Then see how long restart takes? systemctl restart openstack-cinder-volume
Verified on: openstack-cinder-10.0.6-4.el7ost.noarch Gorka suggested decrease these cinder.conf options below would speed things up a bit. 60->30 for periodic interval 60 -> 5 for periodic_fuzzy_delay Restarted service to update settings I've created several volumes 12+ volumes, totaling ~1T+ in provisioned capacity. Filled with: udev/random or large iso/qcow2. cloned and changed data nothing I did caused changes service state. Openstack-cinder-volume remains up since system was installed up time of 24H. All the volumes I added were created within the last 4h No glitch in service status while using watch -d -n 10.
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. https://access.redhat.com/errata/RHBA-2018:0306