hadoop-hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Prasad Chakka (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HIVE-141) drop table partition behaving oddly - does not create subdirectories
Date Wed, 24 Dec 2008 00:02:44 GMT

    [ https://issues.apache.org/jira/browse/HIVE-141?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12659000#action_12659000
] 

Prasad Chakka commented on HIVE-141:
------------------------------------

this seems to be result of a bug/feature in Hadoop Trash.moveToTrash() function

The comment for the class Trash (see below) says that the deleted directory should retail
the full path in .Trash/Current directory but the actual code is only the name of the directory
is used. I will try to check the actual policy of the deletes and see whether we need to implement
our own moveToTrash function.

/** Provides a <i>trash</i> feature. Files are moved to a user's trash
 * directory, a subdirectory of their home directory named ".Trash". Files are
 * initially moved to a <i>current</i> sub-directory of the trash directory.
 * Within that sub-directory their original path is preserved. Periodically
 * one may checkpoint the current trash and remove older checkpoints. (This
 * design permits trash management without enumeration of the full trash
 * content, without date support in the filesystem, and without clock
 * synchronization.)
 */ 

> drop table partition behaving oddly - does not create subdirectories
> --------------------------------------------------------------------
>
>                 Key: HIVE-141
>                 URL: https://issues.apache.org/jira/browse/HIVE-141
>             Project: Hadoop Hive
>          Issue Type: Bug
>          Components: Metastore
>            Reporter: Hao Liu
>   Original Estimate: 4h
>  Remaining Estimate: 4h
>
> for example, I have a table, which has two partitions:
> tmp_table_name/dt=2008-11-01
> tmp_table_name/dt=2008-11-02
> When we use hive metastore to drop the first partition (as root), I expect the data file
will be moved to user/root/.Trash/081103xxxx/tmp_table_name/dt=2008-11-01 by default. However,
directory tmp_table_name was not created, the data was moved to user/root/.Trash/081103xxxx/dt=2008-11-01,
which makes data recovery a very difficult task.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message