Cinder-volume doesn't have to work if the backend is not initialized

[Andrey Grebennikov]

Fuel 6.1, Ubuntu, Ceph as the backend for Cinder-volume.
I found in the log that at the moment when the service was starting the backend was not available (could not connect to any monitor). At the same time, it still continues to pick up the tasks for creation/deletion from the queue even though it cannot work them out. All volumes it tries to delete get stuck and could not be deleted through the API anymore.

This particular service should be marked out as unhealthy in this case and it doesn't have to participate in the process.

[Vitaly Sedelnik (vsedelnik)]

Cinder team, please check if this is something that could be improved in MOS 8.0

[Yuriy Nesenenko (ynesenenko)]

This bug was not reproduced. The fix was committed to the Liberty.

[Yuriy Nesenenko (ynesenenko)]

This situation was checked out on the following environment:

VERSION:
  feature_groups:
    - mirantis
  production: "docker"
  release: "8.0"
  openstack_version: "2015.1.0-8.0"
  api: "1.0"
  build_number: "117"
  build_id: "117"
  fuel-nailgun_sha: "4ab6b1f994846fae5b14bcd7f892a621d21132bb"

[Andrey Grebennikov (agrebennikov)]

We may still want to check/validate it on 6.1/7.0 since it may affect existing customers