hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chaoyu Tang (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HIVE-7441) Custom partition scheme gets rewritten with hive scheme upon concatenate
Date Tue, 05 Aug 2014 22:07:16 GMT

     [ https://issues.apache.org/jira/browse/HIVE-7441?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Chaoyu Tang updated HIVE-7441:
------------------------------

    Attachment: HIVE-7441.1.patch

Thanks, Szehon. I made changes and uploaded the new patch to review board. 
Also I fixed the tests and the failures I think were from the order of the selected rows.

> Custom partition scheme gets rewritten with hive scheme upon concatenate
> ------------------------------------------------------------------------
>
>                 Key: HIVE-7441
>                 URL: https://issues.apache.org/jira/browse/HIVE-7441
>             Project: Hive
>          Issue Type: Bug
>          Components: CLI
>    Affects Versions: 0.10.0, 0.11.0, 0.12.0
>         Environment: CDH4.5 and CDH5.0
>            Reporter: Johndee Burks
>            Assignee: Chaoyu Tang
>            Priority: Minor
>             Fix For: 0.14.0
>
>         Attachments: HIVE-7441.1.patch, HIVE-7441.patch
>
>
> If I take a given data directories. The directories contain a data file that is rc format
and only contains one character "1".
> {code}
> /j1/part1
> /j1/part2
> {code}
> Create the table over the directories using the following command:
> {code}
> create table j1 (a int) partitioned by (b string) stored as rcfile location '/j1' ;
> {code}
> I add these directories to a table for example j1 using the following commands:
> {code}
> alter table j1 add partition (b = 'part1') location '/j1/part1';
> alter table j1 add partition (b = 'part2') location '/j1/part2';
> {code}
> I then do the following command to the first partition: 
> {code}
> alter table j1 partition (b = 'part1') concatenate;
> {code}
> Hive changes the partition location from on hdfs
> {code}
> /j1/part1
> {code}
> to 
> {code}
> /j1/b=part1
> {code}
> However it does not update the partition location in the metastore and partition is then
lost to the table. It is hard to find this out until you start querying your data and notice
there is missing data. The table even still shows the partition when you do "show partitions".



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message