I have a query case with group by and UDF defined :
I use "create function explode_json_array
using jar 'hdfs:///hadoop-tools/hadoop-tools-0.0.1-SNAPSHOT.jar'; " to create udf.
It works fine on both mr and spark cluster mode without "group by" clause, but fails when running on spark cluster mode with "group by" clause defined.
Detail error message is below