hadoop-common-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Allen Wittenauer (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HADOOP-11013) CLASSPATH handling should be consolidated, debuggable
Date Thu, 28 Aug 2014 15:42:10 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-11013?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14113867#comment-14113867

Allen Wittenauer commented on HADOOP-11013:

What's interesting to me in that output is the prevention of the dupe'ing of the HADOOP_CONF_DIR
in the CLASSPATH.  It could be (easily) argued that HADOOP_CONF_DIR should only be added once,
but I'm leaning towards leaving it in there like it currently is as a "safety" feature.

> CLASSPATH handling should be consolidated, debuggable
> -----------------------------------------------------
>                 Key: HADOOP-11013
>                 URL: https://issues.apache.org/jira/browse/HADOOP-11013
>             Project: Hadoop Common
>          Issue Type: Improvement
>            Reporter: Allen Wittenauer
>            Assignee: Allen Wittenauer
>         Attachments: HADOOP-11013-01.patch, HADOOP-11013.patch
> As part of HADOOP-9902, java execution across many different shell bits were consolidated
down to (effectively) two routines.  Prior to calling those two routines, the CLASSPATH is
exported.  This export should really be getting handled in the exec function and not in the
individual shell bits.
> Additionally, it would be good if there was:
> {code}
> echo ${CLASSPATH} > /dev/null
> {code}
> so that bash -x would show the content of the classpath or even a '--debug classpath'
option that would echo the classpath to the screen prior to java exec to help with debugging.

This message was sent by Atlassian JIRA

View raw message