A method, system, apparatus, and computer program product is presented for management of a distributed data processing system. Resources within the distributed data processing system are dynamically discovered, and the discovered resources are adaptively monitored using the network management framework. When the network management framework detects that certain components within the network management framework may have failed, new instances of these components are started. If duplicate components are later determined to be active concurrently, then a duplicate component is shutdown, thereby ensuring that at least one instance of these components is active at any given time. After certain failover events, a resource rediscovery process may occur, and a topology database containing previously stored information about discovered resources is resynchronized with resource information about rediscovered resources.