[ 
https://issues.apache.org/jira/browse/YARN-8913?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16660947#comment-16660947
 ] 

Eric Yang commented on YARN-8913:
---------------------------------

Launch command looks like this in patch 002:

{code}
cd /tmp/mawo/hadoop-applications-mawo-0.0.1-SNAPSHOT/bin;export 
JAVA_HOME=/usr/lib/jvm/java-1.8.0;/tmp/mawo/hadoop-applications-mawo-0.0.1-SNAPSHOT/bin/mawo
 --config /tmp/mawo-config master -job_descripiton_file_path 
/tmp/mawo/mawo-payload.json -mawo_app_name mawo-hadoop-ut
{code}

This is trying to do 3 things at the same time.  
# Change working directory to 
/tmp/mawo/hadoop-applications-mawo-0.0.1-SNAPSHOT/bin
# Set JAVA_HOME environment variable
# Run Mawo script with configuration parameters

It will greatly simplify the complexity by using an entry-point.sh to 
bootstrap.  Mawo script can source mawo-env.sh to locate JAVA_HOME location 
like hadoop's libexec/hadoop-config.sh.  The working directory can be computed 
from basedir $0/.. in the entry-point.sh.  This will change the launch command 
to call a fixed entry-point.sh to improve readability and prevent path 
traversal security exploits in the container.


> Add helper scripts to launch MaWo App to run Hadoop unit tests on Hadoop 
> Cluster
> --------------------------------------------------------------------------------
>
>                 Key: YARN-8913
>                 URL: https://issues.apache.org/jira/browse/YARN-8913
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Yesha Vora
>            Assignee: Yesha Vora
>            Priority: Blocker
>         Attachments: YARN-8913.001.patch, YARN-8913.002.patch
>
>
> MaWo application can be used to run Hadoop UT faster in a Hadoop cluster.
>  Develop helper scripts to orchestrate end-to-end workflow for running Hadoop 
> UT using MaWo app.
> Pre-requisite:
>  * A Hadoop Cluster with HDFS and YARN installed
>  * Enable Docker on YARN feature
>  
> Helper-scripts
>  * MaWo_Driver
>  ** create a docker image with latest hadoop source code
>  ** create payload to MaWo app (This is input to mawo app where Each MaWo 
> Task = UT execution of each Hadoop Module)
>  ** Upload payload file to HDFS
>  ** Update MaWo-Launch.json to resolve RM_HOST / Docker Image etc dynamically
>  ** Launch MaWo app in Hadoop cluster



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org

Reply via email to