Adding the YARN Application CLASSPATH to the Configuration File for MapR Distributions

Two main configuration properties specify the application CLASSPATH: yarn.application.classpath and mapreduce.application.classpath. If you do not specify the YARN application CLASSPATH, MapR takes the default CLASSPATH. However, if you specify the MapReduce application CLASSPATH, the YARN application CLASSPATH is ignored. The SAS Embedded Process for Hadoop requires both the MapReduce application CLASSPATH and the YARN application CLASSPATH.

To ensure the existence of the YARN application CLASSPATH, you must manually add the YARN application CLASSPATH to the yarn-site.xml file. Without the manual definition in the configuration file, the MapReduce application master fails to start a container.

The default YARN application CLASSPATH for Linux is:

$HADOOP_CONF_DIR,
$HADOOP_COMMON_HOME/share/hadoop/common/*,
$HADOOP_COMMON_HOME/share/hadoop/common/lib/*,
$HADOOP_HDFS_HOME/share/hadoop/hdfs/*,
$HADOOP_HDFS_HOME/share/hadoop/hdfs/lib/*,
$HADOOP_YARN_HOME/share/hadoop/yarn/*, 
$HADOOP_YARN_HOME/share/hadoop/yarn/lib/*

The default YARN application CLASSPATH for Windows is:

%HADOOP_CONF_DIR%, 
%HADOOP_COMMON_HOME%/share/hadoop/common/*,
%HADOOP_COMMON_HOME%/share/hadoop/common/lib/*,
%HADOOP_HDFS_HOME%/share/hadoop/hdfs/*,
%HADOOP_HDFS_HOME%/share/hadoop/hdfs/lib/*,
%HADOOP_YARN_HOME%/share/hadoop/yarn/*,
%HADOOP_YARN_HOME%/share/hadoop/yarn/lib/*