3 nodes makes up a cluster,close one of the nodes,and deletes the snapshot and journal files on that node.CLOSE the node as soon as possible AFTER reboot it.Finally,reboot the node and some shard data of that node cannot be successfully synchronized from other nodes.