Imagine, that we want to split a Docker Swarm cluster into two parts to operate them independently with a minimal downtime