Scheduling two parallel machines with a single server: the general case