reverse proxy failover during re-deployment

The current deployment has a flaw: While docker containers are restarted during redeployment nginx will return 503 errors because the backend service is down.

The probably simplest way to mitigate this is to start the new container before destroying the old one and configure nginx to try to serve data from either one of those before failing.

https://serverfault.com/questions/140990/nginx-automatic-failover-load-balancing