Hadoop Workflow Example

Hadoop is a scalable, distributed computing solution provided by Apache. Similar to queuing systems, Hadoop allows for distributed processing of large data sets.

This example sets up hadoop on a single node and processes an example data set.

Installing & Running Hadoop

This example uses the flight environment which will need to be activated before the environments can be created so be sure to run flight start or setup your environment to automatically activate the flight environment.

• Install dependencies for Hadoop:

  [flight@chead1 (mycluster1) ~]$ sudo yum install -y java-1.8.0-openjdk.x86_64 java-1.8.0-openjdk-devel.x86_64

• Set default Java version:

  [flight@chead1 (mycluster1) ~]$ sudo alternatives --set java java-1.8.0-openjdk.x86_64
  [flight@chead1 (mycluster1) ~]$ sudo alternatives --set javac java-1.8.0-openjdk.x86_64

• Download Hadoop v3.2.1:

  [flight@chead1 (mycluster1) ~]$ flight silo software pull --repo openflight hadoop 3.2.1

• Add the hadoop installation to the user’s path along with the Java home (replacing SILO_SOFTWARE_DIR with the software directory used by silo in the download above, this can be done temporarily in the CLI or by adding to the user’s ~/.bashrc):

  export HADOOP_HOME=SILO_SOFTWARE_DIR/hadoop/3.2.1
  export PATH="$PATH:$HADOOP_HOME/bin/:$HADOOP_HOME/sbin/"
  export CLASSPATH="$HADOOP_HOME/share/hadoop/mapreduce/hadoop-mapreduce-client-core-3.2.1.jar:$HADOOP_HOME/share/hadoop/mapreduce/hadoop-mapreduce-client-common-3.2.1.jar:$HADOOP_HOME/share/hadoop/common/hadoop-common-3.2.1.jar:~/MapReduceTutorial/SalesCountry/*:$HADOOP_HOME/lib/*"

If the above has been set in the ~/.bashrc then a new session for the user will need to be started for the environment changes to take effect, otherwise the below commands will not be located

• Start the Hadoop distributed file system service:

  [flight@chead1 (mycluster1) ~]$ start-dfs.sh

• Start the resource manager, node manager and app manager service:

  [flight@chead1 (mycluster1) ~]$ start-yarn.sh

Downloading the Hadoop Job

These steps help setup the Hadoop environment and download a spreadsheet of data which will Hadoop will sort into sales units per region.

• Create job directory:

  [flight@chead1 (mycluster1) ~]$ mkdir MapReduceTutorial

• Download job data:

  [flight@chead1 (mycluster1) ~]$ cd MapReduceTutorial
  [flight@chead1 (mycluster1) MapReduceTutorial]$ flight silo file pull openflight:hadoop/hdfiles.tar.gz
  [flight@chead1 (mycluster1) MapReduceTutorial]$ tar xf hdfiles.tar.gz

• Check that job data files are present:

  [flight@chead1 (mycluster1) MapReduceTutorial]$ ls
  Manifest.txt  SalesCountryDriver.java  SalesCountryReducer.java  SalesJan2009.csv  SalesMapper.java  desktop.ini  hdfiles.tar.gz

Preparing the Hadoop Job

• Compile java for job:

  [flight@chead1 (mycluster1) MapReduceTutorial]$ javac -d . SalesMapper.java SalesCountryReducer.java SalesCountryDriver.java

• Compile the final java file for job:

  [flight@chead1 (mycluster1) MapReduceTutorial]$ jar cfm ProductSalePerCountry.jar Manifest.txt SalesCountry/*.class

Loading Data into Hadoop

• Create directory for processing data and copy sales results in:

  [flight@chead1 (mycluster1) MapReduceTutorial]$ mkdir ~/inputMapReduce
  [flight@chead1 (mycluster1) MapReduceTutorial]$ cp SalesJan2009.csv ~/inputMapReduce/

• Check data can be seen in distributed file system:

  [flight@chead1 (mycluster1) MapReduceTutorial]$ hdfs dfs -ls ~/inputMapReduce

Running the Hadoop Job

• Execute the MapReduce job:

  [flight@chead1 (mycluster1) MapReduceTutorial]$ hadoop jar ProductSalePerCountry.jar ~/inputMapReduce ~/mapreduce_output_sales

• View the job results:

  [flight@chead1 (mycluster1) MapReduceTutorial]$ hdfs dfs -cat ~/mapreduce_output_sales/part-00000 | less