Computer systems require high reliability in space environment. Contrary to limited storage space for embedded real-time systems, considering the memory reliability and Fault-tolerant task, we propose a two level of fault-tolerant solution. The program consists of a periodic memory detecting and an error correcting mechanism of system-level and an improved master/slave version scheduling mechanisms of task-level. The experimental verification of the program is carried out in an embedded operating system named MiniCore which based on a service body/execution flow model(SEFM). The introduction of fault-tolerant mechanism increases the code size of MiniCore kernel by 33% and ensures the accuracy of the memory data, with the system's time performance declining slightly and the success rate and scheduling performance of task execution improving significantly.