hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Pengcheng Xiong (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HIVE-13988) zero length file is being created for empty bucket in tez mode (I)
Date Wed, 22 Jun 2016 23:46:16 GMT

    [ https://issues.apache.org/jira/browse/HIVE-13988?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15345425#comment-15345425
] 

Pengcheng Xiong edited comment on HIVE-13988 at 6/22/16 11:45 PM:
------------------------------------------------------------------

[~ashutoshc], your comments are valid. Could u take another look? I tried to only use move
task but it seems more complicated than i thought. Move task is followed by stats task and
we also need to make stats work. Thus, I only make very limited optimization, i.e., when there
is only one "insert into", we skip the task compilation. Please see attached q files for examples.
Thanks.


was (Author: pxiong):
[~ashutoshc], your comments are valid. Could u take another look? I tried to only use move
task but it seems more complicated than i thought. Move task is followed by stats task and
we also need to make stats work. Thus, I only make very limited optimization, i.e., when there
is only one "insert into", we skip the task compilation. Please see attached q files for examples.

> zero length file is being created for empty bucket in tez mode (I)
> ------------------------------------------------------------------
>
>                 Key: HIVE-13988
>                 URL: https://issues.apache.org/jira/browse/HIVE-13988
>             Project: Hive
>          Issue Type: Bug
>            Reporter: Pengcheng Xiong
>            Assignee: Pengcheng Xiong
>         Attachments: HIVE-13988.01.patch, HIVE-13988.02.patch
>
>
> Even though bucket is empty, zero length file is being created in tez mode. 
> steps to reproduce the issue:
> {noformat}
> hive> set hive.execution.engine;
> hive.execution.engine=tez
> hive> drop table if exists emptybucket_orc;
> OK
> Time taken: 5.416 seconds
> hive> create table emptybucket_orc(age int) clustered by (age) sorted by (age) into
99 buckets stored as orc;
> OK
> Time taken: 0.493 seconds
> hive> insert into table emptybucket_orc select distinct(age) from studenttab10k limit
0;
> Query ID = hrt_qa_20160523231955_8b981be7-68c4-4416-8a48-5f8c7ff551c3
> Total jobs = 1
> Launching Job 1 out of 1
> Status: Running (Executing on YARN cluster with App id application_1464045121842_0002)
> ----------------------------------------------------------------------------------------------
>         VERTICES      MODE        STATUS  TOTAL  COMPLETED  RUNNING  PENDING  FAILED
 KILLED  
> ----------------------------------------------------------------------------------------------
> Map 1 ..........      llap     SUCCEEDED      1          1        0        0       0
      0  
> Reducer 2 ......      llap     SUCCEEDED      1          1        0        0       0
      0  
> Reducer 3 ......      llap     SUCCEEDED      1          1        0        0       0
      0  
> Reducer 4 ......      llap     SUCCEEDED     99         99        0        0       0
      0  
> ----------------------------------------------------------------------------------------------
> VERTICES: 04/04  [==========================>>] 100%  ELAPSED TIME: 11.00 s   

> ----------------------------------------------------------------------------------------------
> Loading data to table default.emptybucket_orc
> OK
> Time taken: 16.907 seconds
> hive> dfs -ls /apps/hive/warehouse/emptybucket_orc;
> Found 99 items
> -rwxrwxrwx   3 hrt_qa hdfs          0 2016-05-23 23:20 /apps/hive/warehouse/emptybucket_orc/000000_0
> -rwxrwxrwx   3 hrt_qa hdfs          0 2016-05-23 23:20 /apps/hive/warehouse/emptybucket_orc/000001_0
> ......
> {noformat}
> Expected behavior:
> In tez mode, zero length file shouldn't get created on hdfs if bucket is empty



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

Mime
View raw message