We have a cluster based on OpenStack Rocky, and it was working fine until we added some compute nodes and ceph.
Now we have some issue with Cinder stability. Cinder is crashing during every rally test and even when the rally test is done or not executed. We searched in the log files and found no signs of error or failure messages. Any ideas why? Thanks.
Can you enable verbose logging on Cinder services and post the output? You can set debug = True in /var/log/cinder.conf in the controller and restart cinder services (e.g. cinder-api and cinder-scheduler) before monitoring the logs again; likewise, set debug = True in /var/log/cinder.conf in the compute nodes (restart cinder-volume). You might then be able to see error messages. If you see some error messages, post your results here to help others in debugging.
