hadoop-common-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Eli Collins (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-6501) RawLocalFileSystem.mkdirs does not throw IOException when it should?
Date Thu, 21 Jan 2010 21:05:54 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-6501?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12803484#action_12803484
] 

Eli Collins commented on HADOOP-6501:
-------------------------------------

Sounds like it should thrown an IOException. You might find HADOOP-6499 relevant.

bq. NOTE that a permission or other failure leaves a garbage partially-completed path in the
current hadoop impl.

Sounds like another good test to add.

> RawLocalFileSystem.mkdirs does not throw IOException when it should?
> --------------------------------------------------------------------
>
>                 Key: HADOOP-6501
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6501
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: fs
>    Affects Versions: 0.20.1
>         Environment: Fedora, Java 1.6.0_17, hadoop 20.1
>            Reporter: robert Cook
>            Priority: Minor
>
> public boolean mkdirs(Path f) throws IOException {
>     Path parent = f.getParent();
>     File p2f = pathToFile(f);
>     return (parent == null || mkdirs(parent)) &&
>       (p2f.mkdir() || p2f.isDirectory());
>   }
> none of the methods called throw IOException
>          yet FileSystemContract BaseTest.testMkdirsFailsForSubdirectoryOfExistingFile
fails
>     createFile(path("/test/hadoop/file"));   
>     Path testSubDir = path("/test/hadoop/file/subdir");
>     try {
>       fs.mkdirs(testSubDir);    //<---------returns false instead of throwing exception
>       fail("Should throw IOException.");
>     } catch (IOException e) {
>       // expected
>     }
> --------------------------------------
> further UNIX mkdir operates on a path argument.   why does hadoop mkdir create directories
in the path one at a time???
> NOTE that a permission or other failure leaves a garbage partially-completed path in
the current hadoop impl.

-- 
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