[ 
https://issues.apache.org/jira/browse/HIVE-20319?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16629540#comment-16629540
 ] 

Alice Fan commented on HIVE-20319:
----------------------------------

Hi [~tiana528],
This issue will be resolved by applying this patch of 
[HIVE-12812|https://issues.apache.org/jira/browse/HIVE-12812]. As union remove 
optimizer will write the result directly to nested subdirs and it will require 
to enable mapred.input.dir.recursive. For this case, a workaround will be 'Set 
mapred.input.dir.recursive=true' in session level, then you will be able to see 
correct result. I will add your example as test case for HIVE-12812. Thanks.

> group by and union all always generate empty query result
> ---------------------------------------------------------
>
>                 Key: HIVE-20319
>                 URL: https://issues.apache.org/jira/browse/HIVE-20319
>             Project: Hive
>          Issue Type: Bug
>          Components: Query Processor
>    Affects Versions: 2.3.2
>         Environment: Run on MR, hadoop 2.7.3
>            Reporter: Wang Yan
>            Priority: Blocker
>
> The following query always generates empty results which is wrong.
> {code:sql}
> create table if not exists test_table(column1 string, column2 int);
> insert into test_table values('a',1),('b',2);
> set hive.optimize.union.remove=true;
> select column1 from test_table group by column1
> union all
> select column1 from test_table group by column1;
> {code}
> Actual result : empty
> Expected result: 
> {code:java}
> a
> b
> a
> b
> {code}
> Note that correct result is generated when set 
> hive.optimize.union.remove=false.
> It seems like the fix in HIVE-12788 is insufficient.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to