Public | Automated Build

Last pushed: 2 years ago
Short Description
Base image for all cloudera hadoop nodes built on Centos6.6
Full Description

Docker container for Cloudera Base Nodes

This docker files shows how to build your own base image for various cloudera components running on Centos6.6

Docker instructions

  • Centos 6.6 - will be used as the base
  • Install the necessary packages like wget,sudo, ssh etc (all needed for cloudera manager)
  • Configure sudo for password key based login & create hadoop admin uid with keys
  • Disable IP Tables, Selinux ( by default selinux is disabled in centos6.6 & Cloudera doesn't like Selinux as of now )
  • Setup the Cloudera yum repo, import GPG Key
    • Install Java
    • Install core hadoop components
    • Install the cloudera manager agent - For ports used refer here
    • Clean up the repos,
    • Start SSHD
  • Build the image

Known Issues:

Cloudera does not like hyphen "-" in hostnames, some times it breaks during managed installations.

Build your image

Build your image using the dockerfile docker build --tag="local/clouderabasenode:latest" .

Run as container

  • I recommend running Weave before starting this node, as Weave DNS takes care of the Hostname/IP Constraints required by Cloudera & It likes the container to run in privileged mode
  • Below syntax only opens the cms server port, you might want add more ports as you need.
docker run -dti --name clouderabasenode \
                --privileged=true \
                mystique:clouderabasenode:latest
To Do
  • Expose the necessary ports ( there is a whole lot of them ).
  • Probably can add zookeeper client also.
Docker Pull Command
Owner
mystique
Source Repository

Comments (1)
mystique
2 years ago

Need to add a step to start the db once during the image build itself, would be interesting to see how much it adds to the image size.

  • Adding repos alone increases the image size by 70MB - Isn't that little strange.