In satellite technology, a satellite cluster is an efficient resource-sharing approach in satellite communication, and the key to building it is how to achieve collaboration between satellites. However, in a distributed satellite cluster environment, the topology of the satellite cluster links changes dynamically, the network latency between nodes is high, and the difficulty of resource scheduling increases. Building a satellite cluster task planning system based on Kubernetes is an important direction for the development of satellite technology. However, during the scheduling process, the Kubernetes scheduler does not perceive the network status of the nodes and only considers CPU and memory resources, which can lead to scheduling failures. Therefore, to ensure efficient scheduling of Kubernetes in the satellite cluster environment, this paper improves the resource scheduling algorithm in Kubernetes and proposes a distributed satellite Pod scheduling algorithm based on DQN. The algorithm effectively enables Pod scheduling under complex inter-satellite link status while ensuring that limited satellite cluster resources are fully utilized.