[jira] Updated: (HADOOP-71) The SequenceFileRecordReader uses the default FileSystem rather than the supplied one

["Owen O'Malley (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HADOOP-71?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Owen O'Malley updated HADOOP-71:
--------------------------------

    Attachment: non-default-fs-input.patch

This patch also fixes the TextInputFormat and the framework's handling of the output path too. I will file a follow-up but to get deprecate the current FileSystem.get(Configuration) method and replace it like FileSystem.getDefault(Configuration). There are many place still left in the code where the default file system is gotten. However, with this patch, I was able to run map/reduce jobs (including a unit test) with the input and output going to non-default file systems. 

> The SequenceFileRecordReader uses the default FileSystem rather than the supplied one
> -------------------------------------------------------------------------------------
>
>                 Key: HADOOP-71
>                 URL: https://issues.apache.org/jira/browse/HADOOP-71
>             Project: Hadoop
>          Issue Type: Bug
>          Components: mapred
>    Affects Versions: 0.1.0
>            Reporter: Owen O'Malley
>            Assignee: Owen O'Malley
>             Fix For: 0.14.0
>
>         Attachments: non-default-fs-input.patch
>
>
> The mapred.TestSequenceFileInputFormat test was failing when run with a conf directory that pointed to a dfs cluster. The reason was that SequenceFileRecordReader was using the default FileSystem from the config, while the test program was assuming the "local" file system was being used.

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