[HIVE-18157] Vectorization : Insert in bucketed table is broken with vectorization

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 3.0.0
• Component/s: Transactions, Vectorization
• Assignee: Deepak Jaiswal
• Reporter: Deepak Jaiswal
• Created: 28/Nov/17 01:55
• Updated: 22/May/18 23:12
• Resolved: 28/Nov/17 19:28

Description

create temporary table foo (x int) clustered by into 4 buckets;
insert overwrite table foo values(1),(2),(3),(4),(9);

select , regexp_extract(INPUT_FILE_NAME, './(.*)', 1) from foo;
OK
9 000000_0
4 000000_0
3 000000_0
2 000000_0
1 000000_0

set hive.vectorized.execution.enabled=false;
insert overwrite table foo values(1),(2),(3),(4),(9);
select , regexp_extract(INPUT_FILE_NAME, './(.*)', 1) from foo;
OK
4 000000_0
9 000001_0
1 000001_0
2 000002_0
3 000003_0

Issue Links

• duplicates: HIVE-17923 'cluster by' should not be needed for a bucketed table
• is broken by: HIVE-16557 Vectorization: Specialize ReduceSink empty key case
• links to: RB