A method for monitoring a data processing system is proposed. The method involves the measuring of state parameters of the system. Indicators of the performance of the system are then inferred from the state parameters by applying fuzzy-logic rules. The proposed solution is based on the idea of estimating a trust value, based on the effectiveness of the corrective actions. If the previous corrective actions prove to be effective than the trust value is enhanced and the system is allowed a higher level of autonomy. Otherwise the intervention of an operator might be invoked.