What is wrong with my boto elastic mapreduce jar jobflow parameters?

Question

I am using the boto library to create a job flow in Amazons Elastic MapReduce Webservice (EMR). The following code should create a step:

step2 = JarStep(name='Find similiar items',
            jar='s3n://recommendertest/mahout-core/mahout-core-0.5-SNAPSHOT.jar',
            main_class='org.apache.mahout.cf.taste.hadoop.similarity.item.ItemSimilarityJob',
            step_args=['s3n://bucket/output/' + run_id + '/aggregate_watched/',
                       's3n://bucket/output/' + run_id + '/similiar_items/',
                       'SIMILARITY_PEARSON_CORRELATION'
                      ])

When I run the job flow, it always fails throwing this error:

java.lang.NoClassDefFoundError: org/apache/hadoop/mapreduce/JobContext
at java.lang.Class.forName0(Native Method)
at java.lang.Class.forName(Class.java:247)
at org.apache.hadoop.util.RunJar.main(RunJar.java:148)
at org.apache.hadoop.mapred.JobShell.run(JobShell.java:54)
at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:65)
at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:79)
at org.apache.hadoop.mapred.JobShell.main(JobShell.java:68)
Caused by: java.lang.ClassNotFoundException: org.apache.hadoop.mapreduce.JobContext

This is the line in the EMR logs invoking the java code:

2011-01-24T22:18:54.491Z INFO Executing /usr/lib/jvm/java-6-sun/bin/java \
-cp /home/hadoop/conf:/usr/lib/jvm/java-6-sun/lib/tools.jar:/home/hadoop:/home/hadoop \
/hadoop-0.18-core.jar:/home/hadoop/hadoop-0.18-tools.jar:/home/hadoop/lib/*:/home/hadoop/lib/jetty-ext/* \
-Xmx1000m \
-Dhadoop.log.dir=/mnt/var/log/hadoop/steps/3 \
-Dhadoop.log.file=syslog \
-Dhadoop.home.dir=/home/hadoop \
-Dhadoop.id.str=hadoop \
-Dhadoop.root.logger=INFO,DRFA \
-Djava.io.tmpdir=/mnt/var/lib/hadoop/steps/3/tmp \
-Djava.library.path=/home/hadoop/lib/native/Linux-i386-32 \
org.apache.hadoop.mapred.JobShell \
/mnt/var/lib/hadoop/steps/3/mahout-core-0.5-SNAPSHOT.jar \
org.apache.mahout.cf.taste.hadoop.similarity.item.ItemSimilarityJob \
s3n://..../output/job_2011-01-24_23:09:29/aggregate_watched/ \
s3n://..../output/job_2011-01-24_23:09:29/similiar_items/ \
SIMILARITY_PEARSON_CORRELATION

What is wrong with the parameters? The java class definition can be found here:

https://hudson.apache.org/hudson/job/Mahout-Quality/javadoc/org/apache/mahout/cf/taste/hadoop/similarity/item/ItemSimilarityJob.html

Answer 1

I found the solution for the problem:

1. You need to specify hadoop version 0.20 in the jobflow parameters
2. You need to run the JAR step with mahout-core-0.5-SNAPSHOT-job.jar, not with the mahout-core-0.5-SNAPSHOT.jar
3. If you have an additional streaming step in your jobflow, you need to fix a bug in boto:
   1. Open boto/emr/step.py
   2. Change line 138 to "return '/home/hadoop/contrib/streaming/hadoop-streaming.jar'"
   3. Save and reinstall boto

This is how the job_flow function should be invoked to run with mahout:

jobid = emr_conn.run_jobflow(name = name, log_uri = 's3n://'+ main_bucket_name +'/emr-logging/', enable_debugging=1, hadoop_version='0.20', steps=[step1,step2])

Answer 2

The fix to boto described in step #2 above (i.e. using the non-versioned hadoop-streamin.jar file) has been incorporated into the github master in this commit:

https://github.com/boto/boto/commit/a4e8e065473b5ff9af554ceb91391f286ac5cac7

Answer 3

For Some reference doing this from boto

import boto.emr.connection as botocon

import boto.emr.step as step

con = botocon.EmrConnection(aws_access_key_id='', aws_secret_access_key='')

step = step.JarStep(name='Find similar items', jar='s3://mahout-core-0.6-job.jar', main_class='org.apache.mahout.cf.taste.hadoop.similarity.item.ItemSimilarityJob', action_on_failure='CANCEL_AND_WAIT', step_args=['--input', 's3://', '--output', 's3://', '--similarityClassname', 'SIMILARITY_PEARSON_CORRELATION'])

con.add_jobflow_steps('jflow', [step])

Obviously you need to upload the mahout-core-0.6-job.jar to an accessible s3 location. And the input and out put have to be accessible.