Running MapReduce v1 Jobs

Describes how to configure existing MapReduce v1 jobs to run in a MapR cluster.

If you have existing MapReduce v1 jobs from a hadoop 1.x cluster that you want to run in the MapR cluster, you may need to recompile the MapReduce jobs.

If your MapReduce v1 job contains the following API, you may need to recompile the job prior to running the job in the cluster.

MapReduce Job API Source Compatibility Binary Compatibility Recompile Required?
org.apache.hadoop.mapred Yes Partial If Job, JobControl, or JobContext classes are used, then the code requires minor updates and a recompile due to API changes.
org.apache.hadoop.mapreduce Yes No Recompilation is required.

To recompile a MapReduce v1 job, run the following command:

javac -cp $(/opt/mapr/hadoop/hadoop-0.20.2/bin/hadoop classpath) <java.program>