egoEconomotrics

This project uses Apache Spark run on a Single Node Hadoop/Yarn.

Warning:

This install will play with your ~/.ssh folder, more specifically the .ssh/authorized_keys file
It will allow hadoop to run the ssh localhost command using a DSA PassPhraseLess key

• Defintion:

  • Hadoop HDFS (Hadoop Distributed File System)
    
  • Yarn, MapReduce 2.0
    
  • Spark general engine for large-scale data processing

• Languages:

  • Scala and Python

Prerequisites:

• wget
• java (JVM)
• *nix - Darwin, Cygwin (not yet)
• Python (if running Python)
• sbt to run the Scala examples

Running the install:

./bootstrap/install.sh

Hadoop NameNode Daemons

• Set the Hadoop Home

HDFS_HOME=~/bin/local/bigdata/hadoop

• Starting the services

$HDFS_HOME/sbin/start-dfs.sh

Note: On MacOS, make sure SSH is started. System Preferences/Sharing/Remote Login [ON]

• Checking Services are running

jps

13049 NameNode (HDFS Name Node) -- Make sure this is running
13241 DataNode (HDFS Data Node)
22752 ResourceManager (Yarn Resource)
22894 NodeManager (Yarn Node)

---

Monitoring DFS Health

Browsing the File System's health

http://localhost:50070

Yarn Daemons

• Start ResourceManager daemon and NodeManager daemon:

$HDFS_HOME/sbin/start-yarn.sh

Monitoring Resource Manager

If you want to look at the running jobs or already executed (Jobwatch Equivalent)

http://localhost:8088

Hadoop Distributed File System (Hadoop DFS) handling

• Create and Mount a new Hadoop DFS

${HDFS_HOME}/bin/hdfs namenode -format

Note: You need to restart HDFS

• Create a directory in Hadoop DFS

Create the user directory along with the owner directory

${HDFS_HOME}/bin/hdfs dfs -mkdir -p /user/${USER}

Running the example

Package a jar containing your application

sbt package

... [success] Total time: ...

Set SPARK HOME

SPARK_HOME=~/bin/local/bigdata/spark

Use spark-submit to run your application

${SPARK_HOME}/bin/spark-submit --class "SimpleApp" --master local[4] target/scala-2.10/egoeconometrics_2.10-0.1-SNAPSHOT.jar

... Lines with a: 41, Lines with b: 17

Running Interactive Shell

• Scala

${SPARK_HOME}/bin/spark-shell

• Python

${SPARK_HOME}/bin/bin/pyspark --master local[4]

Running Spark Streaming

• You will first need to run Netcat (a small utility found in most Unix-like systems) as a data server by using

nc -lk 9999

• Then, in a different terminal, you can start the example by using

${SPARK_HOME}/bin/spark-submit --class "QuickStreamingApp" --master local[4] target/scala-2.10/egoeconometrics_2.10-0.1-SNAPSHOT.jar localhost 9999

Stopping the Services

When you're done, stop the daemons with:

$HDFS_HOME/sbin/stop-yarn.sh

$HDFS_HOME/sbin/stop-dfs.sh

AWS Ubuntu

• installing SBT
  http://www.scala-sbt.org/release/tutorial/Installing-sbt-on-Linux.html
  
• Install Java:
  
  • http://www.webupd8.org/2012/09/install-oracle-java-8-in-ubuntu-via-ppa.html
    
  • Add JAVA_HOME in profile (i.e. export JAVA_HOME=/usr/lib/jvm/java-8-oracle in .bashrc)
    
  • After the install change the ${JAVA_HOME} with /usr/lib/jvm/java-8-oracle
    in ~/bin/local/bigdata/hadoop/etc/hadoop/hadoop-env.sh
    
• Add a new spark user and allow remote connection
• Micro Instance out of memory - add swap
  http://developer24hours.blogspot.ca/2013/02/micro-instance-out-of-memory-add-swap.html
  Here are the commands to add a 1GB swap
  

sudo /bin/dd if=/dev/zero of=/var/swap.1 bs=1M count=1024
sudo /sbin/mkswap /var/swap.1
sudo /sbin/swapon /var/swap.1
To turn off the swap do the following:
sudo /sbin/swapoff /var/swap.1

Recommended reading for AWS http://www.michael-noll.com/tutorials/running-hadoop-on-ubuntu-linux-single-node-cluster/

Kwown issues:

• Not working under Cygwin
  • Install and run SSH Daemon on Cygwin
    http://docs.oracle.com/cd/E24628_01/install.121/e22624/preinstall_req_cygwin_ssh.htm
  • After the install change the ${JAVA_HOME} with i.e. /cygdrive/c/opt/bin/java
    in ~/bin/local/bigdata/hadoop/etc/hadoop/hadoop-env.sh
    

License

Copyleft © 2014 EgoOyiri [AfricaCoin]

Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.