Often when we use Apache Hadoop and would like to make a custom build (stock or forked) you'll have to rebuild the whole Hadoop, native libs, etc ... which takes 30+ minutes, and carries lots of dependencies (libraries, protobuf, etc - at a given version).
Hadoop 1.2.1 BaseInstall http://tecadmin.net/steps-to-install-hadoop-on-centosrhel-6/
Docker Build based on the Above installation: https://github.com/hughbrien/docker-hadoop-1.2.1
Install boot2docker and enable it.
Install Docker
boot2docker shellinit
git clone https://github.com/hughbrien/docker-hadoop-1.2.1
cd ./docker-hadoop-1.2.1
docker build -t <yourname>/docker-hadoop-1.2.1 .
docker run -d -i -t -p 50030:50030 -p 50070:50070 -p 50060:50060 --name "node1" hughbrien/docker-hadoop-1.2.1 /bin/bash
docker run -d -i -t -p 50030:50030 -p 50070:50070 -p 50060:50060 --name "node1" hughbrien/docker-hadoop-1.2.1 /bin/bash
docker exec -it nodeX /bin/bash
May have to run bin/hadoop namenode -format
May also have to manually start /etc/init.d/sshd start
docker build -t hughbrien/docker-hadoop-1.2.1 .
docker run -d -i -t -p 50030:50030 -p 50070:50070 -p 50060:50060 --name "node1" hughbrien/docker-hadoop-1.2.1 /bin/bash
docker exec -it node2 /bin/bash