Karen Coppage created HIVE-22902:
------------------------------------
Summary: Incorrect Spark/SparkOnYarn result for
auto_sortmerge_join_16.q
Key: HIVE-22902
URL: https://issues.apache.org/jira/browse/HIVE-22902
Project: Hive
Issue Type: Bug
Reporter: Karen Coppage
In files
{code:java}
auto_sortmerge_join_16.q.out_spark [TestMiniSparkOnYarnCliDriver]
auto_sortmerge_join_16.q.out [TestSparkCliDriver]
{code}
at the first run of:
{code:java}
select a.key , a.value , b.value , 'day1' as day, 1 as pri
from
( select key, value
from bucket_big_n17 where day='day1' ) a
left outer join
( select key, value
from bucket_small_n17
where pri between 1 and 2 ) b
on
(a.key = b.key)
{code}
the output is (beginning line 444):
{code:java}
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
103 val_103 val_103 day1 1
103 val_103 val_103 day1 1
103 val_103 val_103 day1 1
103 val_103 val_103 day1 1
169 val_169 NULL day1 1
172 val_172 val_172 day1 1
172 val_172 val_172 day1 1
172 val_172 val_172 day1 1
172 val_172 val_172 day1 1
374 val_374 NULL day1 1
{code}
Result should not include NULLs. It should match
llap/auto_sortmerge_join_16.q.out, beginning line 461:
{code:java}
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
0 val_0 val_0 day1 1
103 val_103 val_103 day1 1
103 val_103 val_103 day1 1
103 val_103 val_103 day1 1
103 val_103 val_103 day1 1
169 val_169 val_169 day1 1
169 val_169 val_169 day1 1
169 val_169 val_169 day1 1
169 val_169 val_169 day1 1
169 val_169 val_169 day1 1
169 val_169 val_169 day1 1
169 val_169 val_169 day1 1
169 val_169 val_169 day1 1
172 val_172 val_172 day1 1
172 val_172 val_172 day1 1
172 val_172 val_172 day1 1
172 val_172 val_172 day1 1
374 val_374 val_374 day1 1
374 val_374 val_374 day1 1
{code}
Looks like this was changed in HIVE-20915.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
