[HIVE-11761] DoubleWritable hashcode for GroupBy is not properly generated

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.2.0, 2.0.0
• Fix Version/s: 1.3.0, 2.0.0
• Component/s: Hive
• Hadoop Flags: Reviewed
• Assignee: Aihua Xu
• Reporter: Aihua Xu
• Created: 08/Sep/15 18:23
• Updated: 16/Feb/16 23:50
• Resolved: 11/Sep/15 15:05

Description

HIVE-11502 fixed the hashcode for LazyDouble. Additionally we should fix for DoubleWritable as well due to HADOOP-12217 issue. In some cases such as select avg(t) from (select * from over1k cross join src) t group by d; where d is double type, the data is actually in DoubleWritable, not LazyDouble. Thus, before HADOOP-12217 gets fixed, we need to fix hashcode for LazyDouble as well as DoubleWritable.

Issue Links

• is related to: HIVE-11502 Map side aggregation is extremely slow