The Issue: The docker swarm manager node becomes useless after the tasks.db file explodes in size. This can be seen by worker nodes not being able to connect to the swarm, or manager nodes not seeing the other manager.
The Fix: Stop the docker service (service docker stop), delete or move the tasks.db file, start the docker service (service docker start). This seems too simple to be true, but it isn’t, the tasks.db file can be safely removed and regenerated by the docker swarm manager.