[HIVE-7023] Bucket mapjoin is broken when the number of small aliases is two or more

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.13.0
• Fix Version/s: 0.14.0
• Component/s: Query Processor
• Assignee: Navis Ryu
• Reporter: Navis Ryu
• Created: 07/May/14 04:21
• Updated: 13/Nov/14 19:39
• Resolved: 09/May/14 23:27

Description

From auto_sortmerge_join_11.q,

-- small 1 part, 2 bucket & big 2 part, 4 bucket

CREATE TABLE bucket_small (key string, value string) partitioned by (ds string) CLUSTERED BY (key) INTO 2 BUCKETS STORED AS TEXTFILE;
load data local inpath '../../data/files/smallsrcsortbucket1outof4.txt' INTO TABLE bucket_small partition(ds='2008-04-08');
load data local inpath '../../data/files/smallsrcsortbucket2outof4.txt' INTO TABLE bucket_small partition(ds='2008-04-08');

CREATE TABLE bucket_big (key string, value string) partitioned by (ds string) CLUSTERED BY (key) INTO 4 BUCKETS STORED AS TEXTFILE;
load data local inpath '../../data/files/srcsortbucket1outof4.txt' INTO TABLE bucket_big partition(ds='2008-04-08');
load data local inpath '../../data/files/srcsortbucket2outof4.txt' INTO TABLE bucket_big partition(ds='2008-04-08');
load data local inpath '../../data/files/srcsortbucket3outof4.txt' INTO TABLE bucket_big partition(ds='2008-04-08');
load data local inpath '../../data/files/srcsortbucket4outof4.txt' INTO TABLE bucket_big partition(ds='2008-04-08');

load data local inpath '../../data/files/srcsortbucket1outof4.txt' INTO TABLE bucket_big partition(ds='2008-04-09');
load data local inpath '../../data/files/srcsortbucket2outof4.txt' INTO TABLE bucket_big partition(ds='2008-04-09');
load data local inpath '../../data/files/srcsortbucket3outof4.txt' INTO TABLE bucket_big partition(ds='2008-04-09');
load data local inpath '../../data/files/srcsortbucket4outof4.txt' INTO TABLE bucket_big partition(ds='2008-04-09');

set hive.auto.convert.join=true;
set hive.ignore.mapjoin.hint=false;

set hive.auto.convert.sortmerge.join=true;
set hive.optimize.bucketmapjoin=true;
set hive.optimize.bucketmapjoin.sortedmerge=true;

select /* + MAPJOIN(a,b) */ count(*) FROM bucket_small a JOIN bucket_big b ON a.key = b.key JOIN bucket_big c ON a.key = c.key;

The last query produces 0 row, instead of 180 rows, which is correct.

Issue Links

• blocks: HIVE-4790 MapredLocalTask task does not make virtual columns