Edge caching is a promising technologies, which enhances content delivery and reduces service latency by caching potentially interested content and services on edge nodes near users. Due to constraints such as limited edge server capacities and unstable communications in a highly dynamic MEC environment, caching and content delivery are prone to failures and inconsistencies . Consequently, redundant caching mechanisms are in high need as counter measures in this paper, we proposes a novel redundant service caching and task offloading Decision (FT-STD) method. Specifically, we utilize the primary backup (PB) approach for offloading fault tolerance and employ a learning-based algorithm for yielding redundant service caching schedules in MEC. Simulation results demonstrate thatclearly outperforms its peers across multiple performance metrics.