Rebooting the platform causes TCPServiceDiscoveryBackend to fail as it requires a constant connection. This needs to be modified so it constantly reconnects. This will allow the the cloud to do rolling updates without causing issues like: #1060