You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Sergey Shelukhin (JIRA)" <ji...@apache.org> on 2014/10/29 21:19:35 UTC

[jira] [Commented] (HIVE-8654) CBO: parquet_ctas test returns incorrect results

    [ https://issues.apache.org/jira/browse/HIVE-8654?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14188926#comment-14188926 ] 

Sergey Shelukhin commented on HIVE-8654:
----------------------------------------

[~brocknoland] do you guys have any feedback? We will probably look more at this, but I wonder why query would break on Parquet specifically. Does writing to Parquet table depend on something in addition to what other output formats do? CBO can change query plan, column names, etc.

> CBO: parquet_ctas test returns incorrect results
> ------------------------------------------------
>
>                 Key: HIVE-8654
>                 URL: https://issues.apache.org/jira/browse/HIVE-8654
>             Project: Hive
>          Issue Type: Sub-task
>          Components: CBO
>            Reporter: Sergey Shelukhin
>            Assignee: Sergey Shelukhin
>             Fix For: 0.15.0
>
>
> I am investigating right now. 
> The issue is specific to Parquet:
> {noformat}
> set hive.cbo.enable=true;
> drop table staging;
> drop table parquet_ctas;
> create table staging (key int, value string) stored as textfile;
> insert into table staging select * from src order by key limit 10;
> select * from staging;
> create table parquet_ctas stored as parquet as select * from staging;
> select * from parquet_ctas;
> create table orc_ctas stored as orc as select * from staging;
> select * from orc_ctas;
> create table txt_ctas stored as textfile as select * from staging;
> select * from txt_ctas;
> {noformat}
> The parquet query returns all NULLs with CBO on.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)