[HIVE-5237] Incorrect group-by aggregation in 0.11.0

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Duplicate
• Affects Version/s: 0.11.0
• Assignee: Unassigned
• Reporter: Børge Svingen
• Created: 06/Sep/13 14:34
• Updated: 13/Sep/13 09:25
• Resolved: 13/Sep/13 09:25

Description

group by with sub queries does not correctly aggregate results in Hive 0.11.0.

To reproduce:

Put the file

1,b
2,c
2,b
3,a
3,c
4,a

in HDFS, and run

create external table abc (x int, y string) row format delimited fields terminated by ',' location '/data/';

The query

select
        x,
        count(*)
from
(select
        x,
        y
from
        abc
group by
      x,
      y
) a
group by
        x;

will then give the result

2	1
3	1
2	1
4	1
3	1
1	1

instead of the correct

1	1
2	2
3	2
4	1

In 0.9.0 and 0.10.0 this is all working correctly.

Issue Links

• duplicates: HIVE-5149 ReduceSinkDeDuplication can pick the wrong partitioning columns