Operations in today's job-shop environment are highly uncertain due to uncertain events such as machine failure, job delay, urgent new job insertion, change of due date, lack of materials, and lack of tools. To respond to uncertain factors and improve flexibility, today's production scheduling systems are gradually moving from centralized to decentralized. In this research, we propose a decentralized scheduling model that requires no central scheduler; instead, the system's agents can collaborate to make their local schedules under uncertain job arrivals by employing a cooperative problem-solving mechanism. The experimental results show that the proposed model outperforms popular dispatching rules such as FIFO, EDD, and SPT in terms of resource utilization and tardiness.