hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bryan Pendleton (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HADOOP-226) DFSShell problems. Incorrect block replication detection in fsck.
Date Thu, 03 Aug 2006 20:51:15 GMT
     [ http://issues.apache.org/jira/browse/HADOOP-226?page=all ]

Bryan Pendleton updated HADOOP-226:

               Status: Patch Available  (was: Open)
           Attachment: fsck.replication.fix.txt
    Affects Version/s: 0.4.0
                           (was: 0.2.0)

Here's a fix to make "fsck" count over/under replicated blocks according to the per-file value,
rather than the global one.

> DFSShell problems. Incorrect block replication detection in fsck.
> -----------------------------------------------------------------
>                 Key: HADOOP-226
>                 URL: http://issues.apache.org/jira/browse/HADOOP-226
>             Project: Hadoop
>          Issue Type: Bug
>          Components: dfs
>    Affects Versions: 0.4.0
>         Environment: All
>            Reporter: Konstantin Shvachko
>             Fix For: 0.5.0
>         Attachments: fsck.replication.fix.txt
> 1. We need to adjust Dfsck to the new per-file replication feature.
> fsck checks block replication based on the configured global replication parameter.
> Which is now just the default. The actual file replication is returned in DFSFileInfo.
> So at least the reporting is screwed by that, although I didn't check what will happen
> other options -move and -delete.
> 2. fsck throws NullPointerException if you type
> bin/hadoop fsck -files /doc
> instead of
> bin/hadoop fsck /doc -files
> 3. Unfortunately, there are several commands that throw different kinds of Exceptions
> rather than at least printing the usage info, when some of its arguments are missing
> misplaced. ArrayIndexOutOfBoundsException is one them. Try
> bin/hadoop dfs -mv
> bin/hadoop dfs -cp
> bin/hadoop dfs -rm
> 4. In general the shell is growing and getting more sophisticated.
> Should we work out a general convention on how the parameters should be structured, named,
> short/long version of the keywords, help, etc.

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira


View raw message