This paper proposes a novel energy management framework by combining proactive and reactive approaches to efficiently address the uncertainties associated with generation and demand in islanded and interconnected operation of residential microgrid (MG). The MG considers renewable energy sources (RESs), a diesel generator (DG), and a storage device with the possibility of power exchanges with the grid. The proposed formulation is cast as stochastic mixed-integer linear programming (MILP) model with 24-hour rolling horizon, simulated periodically by updating input data and advancing on an hourly basis for a one year scheduling period. The objective is to minimize the total MG cost while preserving comfort priorities of individual households. The energy balance is potentiated by smart central management system of heating, ventilation and air-conditioning and electric water heater loads in conjunction with exploiting the building thermal inertia of well-insulated buildings. Random outages of MG components are modeled via a two-state Markov chain process, which is considered in the reactive approach. Extensive numerical results are presented to reveal the benefits obtained by coordinating the demand response and supply sources for a practical case of Helsinki, Finland. Finally, the impact of various system parameters on the optimal expected MG cost and risk indices is investigated.