I'm seeing an interesting issue with 1.5.3, it looks like if a storage manager pod fails and goes to a completed state, the CRD does not recognize the change in state and restart it.