If Flume was used to populate Hive (througn HiveSink) then the data should have been written properly. Can you check the Flume log to see if the data was successfully getting written to Hive? Note that ACID needs to be enabled on Hive Metastore before streaming data from Flume into Hive.
@Deepesh. Thanks a lot for your quick response. I think data been written properly, as I can use hive CLI to query all data in my table simple4, no matter data written by flume or by SparkSQL program. The problem is when I use SparkSQL or beeline CLI to query data, only data written by SparkSQL or inserted in beeline, not those been written by Flume.