Dear all,

I finally have set up 2.0.3-alpha in pseudo distributed mode as per
http://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-common/SingleCluster.html

This is evidenced by having yarn resourcemanager and nodemanager running,
no exceptions in the logs and I am able to access their web interfaces. The
node manager has registered in the resourcemanager.

Instead of setting all the listed environment variables, I opted for
setting HADOOP_HOME correctly, and used the supplied start-yarn.sh script.

When running randomwriter as indicated on the above website, I notice that
the local job runner is used, which means only one map task is ever used.

Am I missing a configuration setting to have examples run on yarn?
Is there a way to fail Jobs, when they are using LocalJobRunner?

Thanks in advance,

Jens

Ok, got it:

http://hadoop.apache.org/docs/r2.0.3-alpha/hadoop-project-dist/hadoop-common/SingleCluster.html

didn't tell to set

<property>
  <name>mapreduce.framework.name</name>
  <value>yarn</value>
</property>

This does the trick.

Jens

Do you have a mapred-site.xml configured with mapreduce.framework.name
set to yarn?

On Wed, Mar 20, 2013 at 12:28 AM, Jens Scheidtmann
<[EMAIL PROTECTED]> wrote:
> Further notes:
>
> I may be mistaken here:
>
> 2013/3/19 Jens Scheidtmann <[EMAIL PROTECTED]>
>>
>>
>> When running randomwriter as indicated on the above website, I notice that
>> the local job runner is used, which means only one map task is ever used.
>
>
> ...but examples do not register as applications in the resourcetracker and
> randomwriter does not write out 10 maps/node x 1 GB but instead does only 1
> map and 1 GB.
>
> Any hints on how to debug this?
>
> Jens

--
Harsh J