You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by "Konstantin Shvachko (JIRA)" <ji...@apache.org> on 2007/10/04 23:42:50 UTC

[jira] Updated: (HADOOP-1997) TestCheckpoint fails on Windows

     [ https://issues.apache.org/jira/browse/HADOOP-1997?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Konstantin Shvachko updated HADOOP-1997:
----------------------------------------

    Attachment: TestCheckpoint.patch

All tests are successful on both platforms now.

> TestCheckpoint fails on Windows
> -------------------------------
>
>                 Key: HADOOP-1997
>                 URL: https://issues.apache.org/jira/browse/HADOOP-1997
>             Project: Hadoop
>          Issue Type: Bug
>          Components: dfs
>    Affects Versions: 0.15.0
>         Environment: Windows
>            Reporter: Konstantin Shvachko
>            Assignee: Konstantin Shvachko
>            Priority: Blocker
>         Attachments: TestCheckpoint.patch
>
>
> The bug is in the test. TestCheckpoint opens edits files in order to verify its length but does not close it, so the
> secondary name-node attempt to delete it (in order to move edits.new into it on Win) fails.
> But this revealed 3 more bugs.
> - FSEditLog.purgeEditLog() in case of delete / rename failure calls FSEditLog.processIOError(), which should have
> failed because all FSEditLog streams are closed by that time. It does not because the condition for throwing the
> exception is incorrect.
> - In fact purgeEditLog() should call directly FSImage.processIOError() because again all streams are closed.
> - FSImage.processIOError() should also throw an exception if number of directories is <=1 rather than when it == 1

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