Scalability and privacy concerns have created significant interest in decentralized coordination of distributed energy resources (DERs) within microgrids. Previously proposed approaches, however, fail to achieve feasible solutions under flexible demand (FD) and energy storage (ES) participation. After justifying and demonstrating this challenge, this paper develops a novel Lagrangian relaxation-based mechanism achieving feasible, nearoptimal solutions in a decentralized fashion, considering both active and reactive power. A two-level iterative algorithm eliminates the infeasibility effect of FD and ES nonstrict convexities, and prevents the creation of new demand peaks and troughs by the concentration of their response at the same low-and high-priced periods. Tradeoffs associated with the design and operation of the mechanism are analyzed, and the value of additional information submission by the DER, in enabling the quantification of an optimality bound of the determined solutions and significant improvements in communication requirements, is assessed. These contributions are supported by case studies on an LV microgrid test system.