Hi Viney,
Instead of adding sysout's and building every time, I will suggest you to
set up dev env for debugging.
As you are already built the mapreduce project, you can add debug conf's
in yarn-en.sh and put debug points in the code and start analyzing it.
1) Copy yarn-env.sh(
hadoop-dist/target/hadoop-0.24.0-SNAPSHOT/etc/hadoop/yarn-env.sh) into
YARN_CONF_DIR
2) Add following debug conf's
YARN_RESOURCEMANAGER_OPTS="-Xdebug
-Xrunjdwp:transport=dt_socket,server=y,address=5000"
YARN_NODEMANAGER_OPTS="-Xdebug
-Xrunjdwp:transport=dt_socket,server=y,address=5001"
YARN_JOB_HISTORYSERVER_OPTS="-Xdebug
-Xrunjdwp:transport=dt_socket,server=y,address=5002"
3) In eclipse, go to corresponding class file for example
ResourceManager.java -> debug as -> Remote Java Apllication ----> new --->
set corresponding port
Now it's ready to debug.
For testing your patch you can set-up separate testbed.
For contribution: http://wiki.apache.org/hadoop/HowToContribute
Thanks,
B Anil Kumar.
On Thu, Mar 8, 2012 at 2:58 PM, Viney Gupta <[email protected]> wrote:
> Hi all,
>
> I am new to Hadoop and just start coding in MapReduce. I've checked out
> the trunk and am able to build the MapReduce project. I also import the
> code to the eclipse. My very first goal is to add a few printout
> statements, locally build the MR jar, deploy it to a testbed, and run a
> test program on the testbed, watching how the flow go.
>
> In this scenario, besides the trunk code, should I prepare a test bed? How
> could I prepare such one, and where should I deploy the generated MR jar?
>
> It's a newbie question ... and really hope that people here could show me
> the ropes so that I could start. Any help will be appreciated.
>
> Thank you!
>
> --
> Regards,
> Viney Gupta
>
>
