In this paper, energy storage sharing among a group of households with integrated renewable generations in a grid-connected microgrid is studied. In such a microgrid, a shared energy storage management (SESM) system is operated by an aggregator aiming to minimize the long term time-averaged costs of all households, by jointly taking into account the operational constraints of the shared energy storage, the stochastic solar power generations and the time-varying load requests from all households, as well as the fluctuating electricity prices. We formulate this energy management problem, which comprises storage management and load control, as a constrained stochastic programming problem. A centralized real-time sharing control algorithm is designed based on the Lyapunov optimization theory. The performance of the proposed real-time sharing control algorithm is evaluated. By comparing with a greedy sharing algorithm, it is shown that the proposed sharing algorithm outperforms in terms of both cost saving and renewable energy generation utilization.