Best practice hadoop implementation using Ansible
Switch branches/tags
Nothing to show
Clone or download
Pull request Compare This branch is 413 commits behind objectrocket:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


These Ansible playbooks will build a Hadoop cluster (Hortonworks Data Platform).

You can pre-build a Rackspace cloud environment or run the playbooks against an existing environment.

[Configuration files] (id:configuration)

To customize, change the variables under playbooks/group_vars folder:

  1. playbooks/group_vars/all: contains global cluster and cloud settings
  2. playbooks/group_vars/master-nodes: master-nodes configuration
  3. playbooks/group_vars/slave-nodes: slave-nodes configuration
  4. playbooks/group_vars/edge-nodes: edge-nodes configuration

For a one-node cluster, set cloud_nodes_count in master-nodes to 1 and cloud_nodes_count in slave-nodes to 0.

[Requirements] (id:requirements)

  • Requires Ansible 1.9 or newer

  • Expects CentOS/RHEL 6.x or Ubuntu 14 hosts

  • Building the cloud environment requires the pyrax Python module:

    Also recommended is to run pip install oslo.config netifaces.

  • The cloud environment requires the standard pyrax credentials file that looks like this:

    username = my_username
    api_key = 01234567890abcdef

    This file will be referenced in playbooks/group_vars/all (the rax_credentials_file variable).

    By default, the file is expected to be: ~/.raxpub

[Scripts] (id:scripts)

To provision a cloud environment, run the script after you've customized the variables under playbooks/group_vars:


###bootstrap* and hortonworks*

Similarly, run the bootstrap and hortonworks scripts (in this order), depending what type of environment you have.

Example for a cloud environment:


For dedicated / prebuilt environments, you'll need to manually add the nodes in the inventory/static file.

Accessing Ambari

Once you are at this point you can see progress by accessing the Ambari interface (the ambari-node will be the last host that ran a play).

The provided Ansible playbook will only open the firewall if you've added your workstation IP to allowed_external_ips variable in the playbooks/group_vars/all file.

Alternatively, you can access Ambari by either opening the firewall manually or by opening a socks proxy with the following command:

ssh -D 12345 root@ambari-node

You will need to modify your browser settings to use socks proxy localhost and port 12345.

You'll then be able to navigate to http://ambari-node:8080 in your configured browser and access all subsidiary links.

Provision a Rackspace Cloud Big Data cluster ( by running this script.

Customize it via the playbooks/group_vars/cbd file.


[Ansible-Hadoop History] (id:history)

As with many projects this code is the end result of a lot of effort from individuals not properly represented by a simple commit history.

Rackspace started deploying Hadoop on dedicated gear for customers more than a year ago in a very manual process. This process landed with myself and these Rockstars:

Joe Engel (Racker Emeritus)

Mark Lessel

Alexandru Anghel

All of whom wrote a lot of the automation for deploying Hadoop on customer gear at Rackspace.

Today with a pile of customers under our belt and many more all the time, we wanted to share our efforts with the world by publishing this code which you can also use to deploy Hadoop in various ways at Rackspace.

This of course is only the beginning!

I hope this project evolves and inspires even more Rockstars to find ways to contribute.