Thanks to Martin Bukatovič for this excellent writeup !

This page expects that you have already provisioned cluster of CentOS 6.x machines where you would like to deploy 0.8.0 release of Bigtop Hadoop distribution using Puppet recipes from bigtop-deploy.

Requirements of 0.8.0 release:

Quick overview

To install Bigtop Hadoop distribuion on your cluster of machines running CentOS 6.x, you would need 2 Bigtop deliverables:

So where do I find the packages? Bigtop home page contains link to bigtop-0.8.0/repos/centos6/bigtop.repo file which uses


You can use this url when you will configure puppet config file (as shown in detail below), or you can go to Bigtop jenkins and download all packages so that you can setup you own local repository. Another option would be to build it yourself, but this topic is not covered in this page.

Puppet deployment scrips are not packaged and you can get it from either:

While it's safer to use recipes from the release in theory, most people uses puppet scripts from git. The reasoning behind this is that while current recipes changed a lot, it still can deploy Bigtop Hadoop from 0.8.0 release and so it makes sense to learn and use the newest version from the start.

Note: the only 0.8.0 Bigtop Hadoop component which can't be deployed with scripts from git is sqoop.

Installation of requirements

There are basically only two requirements:

For the CentOS 6.x machines, this means:

Note that:

The deployment

Make sure that content of Bigtop 0.8.0 release tarball or bigtop repo (as discussed above) is available in /opt/bigtop on all machines. You may like to use NFS share for this.

Now it's the time to configure puppet site.csv file (the only puppet file you need to touch). It defines cluster roles, hadoop components installed and other details:



Bigtop uses puppet in masterless mode, so you need to distribute new version of site.csv and then run puppet locally on all machines.

So on each machine of the cluster:

  1. push new site.csv /opt/bigtop/bigtop-deploy/puppet/config/site.csv
  2. run puppet:
cd /opt/bigtop/bigtop-deploy/puppet
puppet apply -d --modulepath=modules --confdir=. manifests/site.pp

Check java version

If you have multiple versions of java installed, check which one is active:

# alternatives --display java

The puppet wouldn't notice that different version of java is active.

And install and/or change it to 1.6 if needed:

# yum install java-1.6.0-openjdk java-1.6.0-openjdk-devel
# alternatives --set javac /usr/lib/jvm/java-1.6.0-openjdk.x86_64/bin/javac
# alternatives --set java /usr/lib/jvm/jre-1.6.0-openjdk.x86_64/bin/java