In this article, the problem of optimal reconfiguration control for large-scale satellite cluster with collision avoidance constraints and uncertainties is considered. Based on receding horizon control and successive convexification (SC) methods, a decentralized closed-loop control algorithm is proposed. First, the optimal control problem is converted into a series of convex optimization subproblems by linearization and discretization. These subproblems can be efficiently solved in polynomial time through the SC approach. Second, to alleviate the communication and computational burdens on satellites and enhance scalability, the SC technique is integrated with the receding horizon framework. This approach enables the continuous updating of optimized control commands, resulting in high-precision closed-loop control for satellite cluster reconfiguration. Third, to maintain collision-free satellite operations throughout the mission, an anticollision strategy is proposed to address potential collisions between adjacent discrete points. Furthermore, the concepts of trajectory freezing and parallel computing are utilized to improve computational efficiency. Finally, numerical simulation results are presented to demonstrate the effectiveness, robustness, reliability, and real-time applicability of the proposed method under constraints and uncertainties.