If a job is removed from the job board mid-execution, the conductor should detect this and stop submitting tasks for execution to the worker. With the zookeeper board, it should be easy enough to add a watcher on the node being deleted, but not sure how to handle that on the redis board.