An Evolvable Production System (EPS) is a complex and lively entity composed of intelligent modules that interact, through bio-inspired mechanisms, to ensure high system availability and seamless reconfiguration. The diagnosis of such dynamic systems, characterized by constant change, presents new diagnostic challenges and opportunities that can hardly be tackled by traditional approaches. On the one hand, given the decoupled nature of the system, fault interaction and propagation are harder to detect and contain, as is the development of a global diagnostic model, on the other hand local intelligence and careful characterization of the interactions, between the modules, can be explored to emerge the diagnostic functionalities. The impact of simple consensus mechanisms (majority voting) and fault context analysis (module and its current interactions states) is assessed in a multiagent-oriented application in the assembly domain to understand the validity and contribution of this approach in emerging useful self-diagnostic properties in EPS.