When I execute the workflow it will alert the msg "MASTER_NOT_EXISTS(“无可用master节点”)"

[LZH-Work-Station]

I made a cluster version of dolphinscheduler with 2 masters and 3 workers. However, when I execute the job, sometimes(not all the time) it will alert the msg of “无可用master节点”. I check the zookeeper path of /dolphinscheduler, all sub-nodes are normal and the dead_nodes is empty.