Keep the Cronicle job scheduler running when its master server fails: a master and a backup server on Docker, behind Nginx as the load balancer.