ambari-ansible-module

ambari-ansible-module copied to clipboard

Description

This project contains a custom ansible module for managing an ambari cluster. It supports creating a cluster, starting all cluster services, stopping all cluster services, and deleting the cluster. This assumes the existence of a running ambari server with a set of cluster nodes running the ambari agent. The custom module is located in extra_modules/ambari_cluster_state.py. The ambari_state.yml playbook and roles/ambari_state role demonstrate how to use the custom module.

Motivation

The power of dynamic inventory variables in ansible make it an excellent tool for managing cloud infrastructure. Ambari does much of the hard work in installing and configuring the hadoop ecosystem. This custom module helps us automate and simplify the management of dynamically provisioned hadoop clusters.

Defining a blueprint

The standard for ambari is to define a blueprint as a JSON file. This file maps ambari services (e.g. namenode, datanode) to user defined host groups. Then, you can map hosts to those same host groups as a separate JSON structure known as a host map. In the spirit of ansible, I combine these two structures into a more concise yaml file. See roles/ambari_state/files/blueprint.yml for an example. If useful, it would pretty simple to support the use of existing blueprint/host map files as well.

Running the sample playbook

Specify the appropriate vars in ambari_state.yml for your ambari server, and specify the appropriate services, groups, and hosts in roles/ambari_state/files/blueprint.yml. This sample playbook simple executes the main task in the ambari_state role. See roles/ambari_state/tasks/main.yml for how to create, stop, start, and delete an ambari cluster.

ansible-playbook ambari_state.yml

Dependencies

This project uses the requests and pyyaml python packages.