how to analyze cinder volume service CPU utilizations
In my environment cinder volume service CPU utilization is more than 100%. i am using ceph as backend for cinder. I would like to know where exactly high cpu used in cinder volume. Is there any tool which can help me. And also cinder volume open TCP connections are keep on increasing at high rate with the time. Is there any way to see where exactly these connections are being created.
Have you figured out, what caused that CPU load? I recently upgraded to Ocata, we also use Ceph as backend and since that upgrade we are seeing the same problem as described above, obviously caused by lots of TCP connections to the cluster. Any pointer is appreciated!