Agent quit if promotion failed

The agent should stay working as agent even if promotion failed.

The promotion could fail due to various reasons like binding to an address already in use as following log shows.
"FATA[0076] listen tcp 0.0.0.0:4242: bind: address already in use"

Steps to reproduce:

$ swarmd -d /tmp/node-1 --listen-control-api /tmp/manager1/swarm.sock --hostname node-1
$ swarmd -d /tmp/node-2 --hostname node-2 --join-addr 127.0.0.1:4242
$ swarmctl service create --name busybox-top --replicas 2 --image busybox:latest --command top #optional step
$ swarmctl node promote node-2

The agent node quit with following error logs:

INFO[0076] Renewing TLS Certificate.
Error: listen tcp 0.0.0.0:4242: bind: address already in use
ERRO[0076] container status unavailable error=An error occurred trying to connect: context canceled module=taskmanager task.id=7jk8aj8v4h7f5bz5jm867wb06
WARN[0076] error fetching signed node certificate: rpc error: code = 1 desc = context canceled

Start by reproducing the failure with the listed swarmd and swarmctl commands, especially the occupied 0.0.0.0:4242 listener. Trace the promotion path and the agent lifecycle around the reported certificate and context-canceled logs. Done means a failed promotion leaves node-2 running as an agent rather than quitting.