Test instance manager cleanup during uninstall

  1. Deploy Longhorn v1.1.2
  2. Launch some running volumes.
  3. Upgrade to v1.2.0. ==> All old unused engine managers should be cleaned up automatically.
  4. Make sure all running volumes keep state running.
  5. Upgrade all volumes. ==> All old replica managers should be cleaned up automatically.
  6. Detach all running volumes. ==> All old engine managers should be cleaned up automatically.
  7. do offline upgrade then reattach these volumes.
  8. Directly uninstall the Longhorn system. And use kubectl -n longhorn-system get lhim -w to verify that the system doesn’t loop in instance manager cleanup-recreation. ==> The uninstaller should work fine.
[Edit]