The configurations of an automotive embedded system are normally fixed in production and remain static over the vehicle lifetime. Future scenarios, however, call for more flexible configuration support. DySCAS (Dynamically Self-Configuring Automotive Systems) project aims to introduce context-awareness and self-management features into automotive embedded systems via middleware technologies. Contributing to online configuration decisions, this paper formalizes a fundamental self-configuration problem. It forms a basis for managing the cross interdependencies of configurational items, assessing the system-wide impacts of changes, and making dynamic decisions about new configurations.