hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tsz Wo (Nicholas), SZE (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-3025) ChecksumFileSystem needs to support the new delete method
Date Tue, 18 Mar 2008 18:09:28 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-3025?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12579970#action_12579970

Tsz Wo (Nicholas), SZE commented on HADOOP-3025:

- I think it is better to change FilterFileSystem.delete(Path f) (or even FileSystem.delete(Path
f)) to make it calling FilterFileSystem.delete(f, true), instead of changing ChecksumFileSystem.delete(Path

- In ChecksumFileSystem.delete(Path f, boolean recursive), if f is a directory, it calls fs.delete(f,
recursive).  I think the checksum files won't be deleted.

- We need a test for deleting a tree for testing the recursive parameter.

- In RawInMemoryFileSystem.delete(Path f, boolean recursive), the recursive parameter is ignored.

> ChecksumFileSystem needs to support the new delete method
> ---------------------------------------------------------
>                 Key: HADOOP-3025
>                 URL: https://issues.apache.org/jira/browse/HADOOP-3025
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: fs
>    Affects Versions: 0.17.0
>            Reporter: Devaraj Das
>            Assignee: Mahadev konar
>            Priority: Blocker
>             Fix For: 0.17.0
>         Attachments: HADOOP_3025_1.patch, HADOOP_3025_2.patch, HADOOP_3025_3.patch
> The method FileSystem.delete(path) has been deprecated in favor of the new method delete(path,
recursive). Temporary files gets created in the MapReduce framework and when the time for
deletion comes, they are deleted via delete(path, recursive). This doesn't delete the associated
checksum files. This has a big impact when the FileSystem is the InMemoryFileSystem, where
space is at a premium and wasting space here might hurt the performance of MapReduce jobs
overall. One solution to this problem is to implement the method delete(path, recursive) in
the ChecksumFileSystem but is there is a reason why it was left out as part of HADOOP-771?

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

View raw message