Software defined network(SDN) controller system is the brain of the whole network. In order to reduce the influence of single point fault of controller, this paper designs a simple, reliable and more available master-slave distributed system and master-slave election algorithm based on priority and non preemption. The election process is no longer limited by the number of healthy nodes. The log replication system of Raft algorithm is simplified, and a simple and efficient master-slave incremental/full data replication process is designed. Combined with the data persistence storage and recovery function, the data can not be lost after the system restart. The experiment results show that the system has realized the main functions of the main and subordinate distributed system, and the performance is better, and it can adapt to various complex application scenarios of the distributed SDN controller system.