In a previous post I described how to set up a single-node hadoop cluster in an ubuntu server running on a virtual machine. Sort story: it’s related to my course on Big scale analytics. Refer to the original post for the details. As software upgrade is a matter of fact, I decided to update that tutorial for the more recent 2.4.0 release of hadoop.

I opted for a VM-based solution, so that most of hardware and OS issues students would face would be limited to installing and configuring the VM manager. For the records, I am running Mac OS X 10.9.5 and relying on VirtualBox 4.2.8.

First of all, I downloaded the ISO image for Ubuntu server 14.04 at the Ubuntu server download page and created a Linux-Ubuntu based VM in VirtualBox with 1GB RAM (who read my previous post will note an increase in the server’s RAM, which is due to the fact that the default RAM amount of 512MB did lead to hadoop crashes during simple experiments), a 8GB VDI-based HD (dynamically allocated), and a DVD preloaded with the Ubuntu server 14.04 ISO image. Then I ran the VM and followed all default installation options, except for keyboard layout (I use an italian keyboard). I did not install any additional software, with the exception of manual package installation support.

Once the system was up and running, I installed Hadoop following a mix of the instructions in the tutorials provided by Michael Noll, BigData Handler, and Rasesh Mori, that is what follows.

Some details about the examples: the host name is manhattan, with an administrator user with login name boss (that is, boss is a sudoer); three points (...) in a console are used in order to skip verbose output. Finally, a dollar sign ($) occurring at the beginning of a line denotes the bash prompt.

Setting up the environment

First of all, we need to be sure to work on an up-to-date system. This will probably be the case if the ISO image refers to the current version of Ubuntu server. Just to be sure, log in as the boss user and type the following commands.

boss@manhattan:~$ sudo apt-get update
boss@manhattan:~$ sudo apt-get upgrade

Moreover, it is advisable not to run Hadoop services through a general-purpose user, so the next step consists in adding a group hadoop and a user hadoop-user belonging to that group (for the purposes of this tutorial, all information requested by adduser may be left blank, except the password.

boss@manhattan:~$ sudo addgroup hadoop
boss@manhattan:~$ sudo adduser --ingroup hadoop hadoop-user

Installing Java

The mentioned tutorials suggest a potentially unsafe procedure in order to install the jdk through apt-get, thus it’s advisable to opt for a manual installation.

boss@manhattan:~$ wget --no-cookies --no-check-certificate --header "Cookie: oraclelicense=accept-securebackup-cookie" ""
boss@manhattan:~$ tar -xvzf jdk-7-linux-x64.tar.gz
boss@manhattan:~$ sudo mkdir /usr/local/java
boss@manhattan:~$ sudo cp -r jdk1.7.0_45 /usr/local/java
boss@manhattan:~$ sudo update-alternatives --install "/usr/bin/javac" "javac" "/usr/local/java/jdk1.7.0_45/bin/javac" 1
boss@manhattan:~$ sudo update-alternatives --set javac /usr/local/java/jdk1.7.0_45/bin/javac

Finally, a couple of environment variables should be set up so that the java executables are in $PATH and hadoop knows where java has been installed: this is easily accomplished adding

export JAVA_HOME
export PATH

at the end of /etc/profile (to be edited through su). When these variable are in place it is easy to check that java has been properly installed.

boss@manhattan:~$ . /etc/profile
boss@manhattan:~$ javac -version
javac 1.7.0_45

Setup SSH

All communications with Hadoop are encrypted via SSH, thus the corresponding server should be installed:

boss@manhattan:~$ sudo apt-get install openssh-server

and the hadoop-user must be associated to a key pair and subsequently granting its access to the local machine:

boss@manhattan:~$ su - hadoop-user
hadoop-user@manhattan:~$ ssh-keygen -t rsa -P ""
Generating public/private rsa key pair.
The key's randomart image is:
hadoop-user@manhattan:~$ cat $HOME/.ssh/ >> $HOME/.ssh/authorized_keys

Now hadoop-user should be able to access via ssh to localhost without providing a password:

hadoop-user@manhattan:~$ ssh localhost
The authenticity of host 'localhost (::1)' can't be established.
Last login: ...

Disable IPV6

Hadoop and IPV6 do not agree on the meaning of address, thus it is adivsable to disable IPV6 adding the following lines at the end of /etc/sysctl.conf (after having switched back to the boss user):

# disable ipv6
net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.lo.disable_ipv6 = 1

After a system reboot the output of cat /proc/sys/net/ipv6/conf/all/disable_ipv6 should be 1, meaning that IPV6 is actually disabled.


Download and install Hadoop

Download hadoop-2.4.0.tar.gz, unpack it and move the results in /usr/local, adding a symlink using the more friendly name hadoop and changing ownership of the directory contents to the hadoop-user user:

boss@manhattan:~$ wget wget
boss@manhattan:~$ tar -xzvf hadoop-2.4.0.tar.gz
boss@manhattan:~$ sudo mv hadoop-2.4.0 /usr/local
boss@manhattan:~$ cd /usr/local
boss@manhattan:/usr/local$ sudo ln -s hadoop-2.4.0 hadoop
boss@manhattan:/usr/local$ sudo chown -R hadoop-user:hadoop hadoop-2.4.0

Setup the dedicated user environment

Switch to the hadoop-user user and add the following lines at the end of ~/.bashrc:

# Set Hadoop-related environment variables
export HADOOP_PREFIX=/usr/local/hadoop
export HADOOP_HOME=/usr/local/hadoop
export HADOOP_CONF_DIR=${HADOOP_HOME}/etc/hadoop
# Native Path
export HADOOP_OPTS="-Djava.library.path=$HADOOP_PREFIX/lib"
#Java path
export JAVA_HOME='/usr/local/java/jdk1.7.0_45'
# Add Hadoop bin/ directory to PATH

In order to have the new environment variables in place, reload .bashrc through source .bashrc get back to the administrator user, then open /usr/local/hadoop/etc/hadoop/ , uncomment the line setting JAVA_HOME and set its value to the jdk directory:

export JAVA_HOME=/usr/local/java/jdk1.7.0_45

Configure Hadoop

Before being able to actually use the hadoop file system it is necessary to modify some configuration files inside /usr/local/hadoop/etc/hadoop. All such files follow the an XML format, and the updates should concern the top-level node configuration (likely empty after the hadoop installation). Specifically:

  • in yarn-site.xml:
  • in core-site.xml:
  • in mapred-site.xml (likey to be created through cp mapred-site.xml.template mapred-site.xml):
  • in hdfs-site.xml:

This also requires to manually create the two directories specified in the value XML nodes:

hadoop-user@manhattan:~$ mkdir -p /usr/local/hadoop/yarn_data/hdfs/namenode
hadoop-user@manhattan:~$ mkdir -p /usr/local/hadoop/yarn_data/hdfs/datanode

Formatting the distributed file system

The last step consists in formatting the file system, operation to be executed as hadoop-user:

hadoop-user@manhattan:~$ hdfs namenode -format

the (hopeful) successful result of this operation is specified in one of the last output lines of previous command.

A few more steps and… that’s it!

Hadoop is now installed. Invoking the scripts and respectively start the distributed file system and the mapreduce daemons:

8/10/10 15:28:31 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
Starting namenodes on [localhost]
localhost: starting namenode, logging to /usr/local/hadoop-2.4.0/logs/hadoop-hadoop-user-namenode-manhattan.out
localhost: starting datanode, logging to /usr/local/hadoop-2.4.0/logs/hadoop-hadoop-user-datanode-manhattan.out
Starting secondary namenodes [] starting secondarynamenode, logging to /usr/local/hadoop-2.4.0/logs/hadoop-hadoop-user-secondarynamenode-manhattan.out
8/10/10 15:28:53 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
starting yarn daemons
starting resourcemanager, logging to /usr/local/hadoop/logs/yarn-hadoop-user-resourcemanager-manhattan.out
localhost: starting nodemanager, logging to /usr/local/hadoop-2.4.0/logs/yarn-hadoop-user-nodemanager-manhattan.out

Although it is possible to directly write on the hadoop file system root directory, it is more advisable to create the user directory for hadoop-user, because all relative paths will refer precisely to this directory:

hadoop-user@manhattan:~$ hdfs dfs -mkdir /user
hadoop-user@manhattan:~$ hdfs dfs -mkdir /user/hadoop-user

An absence of outputs from these command invokations means a successful directory creation, which also ensure that the distributed filesystem component of hadoop has been correctly installed. To test also the mapreduce component it is possible to run one of the example jobs distributed along with hadoop:

hadoop-user@manhattan:~$ hadoop jar /usr/local/hadoop/share/hadoop/mapreduce/hadoop-mapreduce-examples-2.4.0.jar pi 10 1000
Job Finished in 34.136 seconds
Estimated value of Pi is 3.1428571428571

Finally, to stop the hadoop daemons, simply invoke and

blog comments powered by Disqus