Renewable energy dominated future power grids require enhanced system flexibility, in particular, activating the participation from various distributed energy resources (DERs). A coordinated two-stage flexibility trading framework for distribution system with microgrids (MGs) is proposed in this paper. At day-ahead stage, a peer to peer (P2P) trading mechanism and the associate leasing strategy of shared energy storage system are performed to solve the power variations caused by the wide spread integration of renewable resources, where asymmetric Nash bargaining is used to realize the fair revenue allocation according to the contribution of each MG in P2P trading. At intra-day stage, given the power imbalances from unexpected uncertainties, MGs exploit the adjustability of the DERs in responding to rapid flexibility requirements issued by distribution system operator. In particular, the average consensus based decentralized Newton method with super linear convergence is utilized to meet the requirements of flexibility while maintaining the information security. The feasibility, effectiveness and equity of the proposed trading strategies are verified through various simulation studies.