Recover gracefully from failed backend services

[larsks]

There was recently a service interruption to the Redis and Galera (MySql) services that sit behind ESI LEAP (and other OpenStack services). While the backend services recovered, ESI LEAP required manual intervention. We should be more resilient in the face of this sort of failure and recover automatically when backend services are restored.