chukwa-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jerome Boulon (JIRA)" <>
Subject [jira] Commented: (CHUKWA-297) Unbundling hadoop jar file
Date Tue, 16 Jun 2009 18:07:07 GMT


Jerome Boulon commented on CHUKWA-297:

So here my proposal: Keep hadoopJar for compile time but don't use it at runtime

1- keep hadoopJar for compile time, we need it there or we should get it from maven but still
in the hadoopJar directory
2- Remove hadoopJar directory from build/chukwa-xxx/ 
3- Remove this block "if [ d ${CHUKWA_HOME}/hadoopjars ]", just bail out instead of trying
to fix a configuration issue

This should work since HADOOP_HOME should be define in your

> Unbundling hadoop jar file
> --------------------------
>                 Key: CHUKWA-297
>                 URL:
>             Project: Hadoop Chukwa
>          Issue Type: Bug
>          Components: build and test code
>    Affects Versions: 0.1.2, 0.2.0
>         Environment: Redhat EL 5.1, Java 6
>            Reporter: Eric Yang
>            Assignee: Eric Yang
>            Priority: Blocker
> During the deployment, the most frequently issue is hadoop version mismatch.  When this
happens, the hadoop jar files needs to be copied to both CHUKWA_HOME/hadoopjars, and hicc/WEB-INF/lib.
 We should be able to work around both issues by using environment controlled class path.
 For build time, the HADOOP_JAR variable could be overwritten by and
to source in hadoop jar file.  If this is this sounds good, then I vote for removing the hadoop-*.jar
file from hadoopjars directory in the upcoming release.  This change would reduce the amount
of hand tweak on the deployment system.  Suggestion and feedback are welcome.

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

View raw message