Feature #21216
closedMethod to release all rbd locks
0%
Description
I ran into an issue when upgrading from Kraken to Luminous with Openstack. Existing volumes would have I/O errors, however new volumes would work fine. I had forgotten to stop all I/O during the upgrade, and after some investigation what appeared to happen is that the old rbd volumes had locks on them that never were released, manually removing the locks resolved the issue. It would be nice to have a command to clear all locks on a pool in case such a similar situation was hit.
Updated by Jason Dillaman over 6 years ago
- Project changed from Ceph to rbd
- Category deleted (
librbd) - Status changed from New to Need More Info
@michael rodriguez: librbd will automatically delete old, stale locks when it attempts to acquire the lock. It sounds like your CephX user didn't have permission to blacklist (dead) clients (see step 6 here [1]).
[1] http://docs.ceph.com/docs/master/release-notes/#upgrade-from-jewel-or-kraken
Updated by Michael Sudnick over 6 years ago
My apologies, just came across that note a few days ago, missed it when doing updates. That fixed the issue, many thanks, this can be closed.
Updated by Jason Dillaman over 6 years ago
- Status changed from Need More Info to Closed