data-driven-manifests

Data Driven Manifests

This is a slightly advanced example, making use of the iteration features available in the future parser.

What

The provided diagram shows what this example will create, namely:

• 2 security groups, named web-sg and db-sg
• 6 instances, split between the two security groups

         +-------------------------------------------------+
         | +---------+ +---------+ +---------+ +---------+ |
         | |         | |         | |         | |         | |
 web-sg  | | web-1   | | web-2   | | web-3   | | web-4   | |
         | |         | |         | |         | |         | |
         | +---------+ +---------+ +---------+ +---------+ |
         +-------------------------------------------------+

         +-------------------------+
         | +---------+ +---------+ |
         | |         | |         | |
 db-sg   | | db-1    | | db-2    | |
         | |         | |         | |
         | +---------+ +---------+ |
         +-------------------------+

What we're really demonstrating is the use of data to drive our infrastructure. At the top of the example manifest is the following:

$instances = {
  'web' => 4,
  'db'  => 2,
}

The rest of the manifest uses this data to automatically generate the puppet resources. If you increase the numbers, or add new types, and rerun the manifest, the new security groups and instances will be created.

How

To run the examples first install the module as described in the README. Then run the following in this directory.

puppet apply init.pp --test --parser future

Discussion

This example shows one approach to using data to drive the structure of your infrastructure, but other elements could obviously be pulled out from the manifests and into a Hiera backend.

A known limitation of this specific code is that reducing the number os instances, or removing a type, from the hash won't result in those instances being deleted.