You are viewing a plain text version of this content. The canonical link for it is here.

Posted to common-issues@hadoop.apache.org by "shanyu zhao (JIRA)" <ji...@apache.org> on 2014/01/15 00:57:19 UTC

[jira] [Commented] (HADOOP-10234) "hadoop.cmd jar" does not propagate exit code.

    [ https://issues.apache.org/jira/browse/HADOOP-10234?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13871398#comment-13871398 ] 

shanyu zhao commented on HADOOP-10234:
--------------------------------------

+1

This patch fixed the problem that a Java program starts a process with command line "hadoop.cmd ..." but always read 0 as exit status even if the process failed. Thanks for the fix Chris!

> "hadoop.cmd jar" does not propagate exit code.
> ----------------------------------------------
>
>                 Key: HADOOP-10234
>                 URL: https://issues.apache.org/jira/browse/HADOOP-10234
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: scripts
>    Affects Versions: 2.2.0
>            Reporter: Chris Nauroth
>            Assignee: Chris Nauroth
>             Fix For: 2.3.0
>
>         Attachments: HADOOP-10234.1.patch
>
>
> Running "hadoop.cmd jar" does not always propagate the exit code to the caller.  In interactive use, it works fine.  However, in some usages (notably Hive), it gets called through {{Shell#getRunScriptCommand}}, which needs to do an intermediate "cmd /c" to execute the script.  In that case, the last exit code is getting dropped, so Hive can't detect job failures.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)