spark streaming data consumption kafka Times Java .lang.IllegalStateException: No current assignment for partition

At first you can run stable when consumption kafka data for a while, but could not pass a few minutes to run out of this exception program interruption, java.lang.IllegalStateException: No current assignment for partition

solved: because I ran the program on a cluster of consumption, local also with the same consumer code tests, the results appeared with a groupID at the same time repeatedly consume the same topic, the problem caused by offset record.