chaostoolkit-terraform 0.0.7

A Chaos Toolkit module to deploy terraform stacks

chaostoolkit-terraform

A Chaos Toolkit driver to extend chaos experiments with Terraform

Package Installation

From Python package index

To install the latest chaostoolkit-terraform stable release:

pip install -U chaostoolkit-terraform

Edge version from the GitHub repository

To install the edge version of the chaostoolkit-terraform package directly from the repository source code:

pip install -U "git+https://github.com/mcastellin/chaostoolkit-terraform.git#egg=chaostoolkit-terraform"

Usage

chaostoolkit-terraform provides a control to deploy terraform modules. To activate the chaosterraform.control for your experiments you need to define it in your experiment files (or settings):

title: My experiment
description: ...

controls:
  - name: "Deploy Terraform module"
    provider:
      type: python
      module: chaosterraform.control

steady-state-hypothesis: {...}

method: []

By default the chaosterraform.control will reference the Terraform module found in the current working directory.

The control will execute Terraform command in the following phases of the experiment execution:

Phase	Actions
Configure control	Initialize the Terraform driver
Before experiment	Initialize and apply the selected Terraform module
After experiment	Run terraform destroy unless specifically asked to retain the created resources

Configuration

You can configure the Terraform control either via control arguments or using Chaos Toolkit parameters with the tf_conf__ prefix:

Configuration with control arguments

controls:
  - name: "Deploy Terraform module"
    provider:
      type: python
      module: chaosterraform.control
      arguments:
        silent: false
        retain: true

Configuration using Ctk parameters

configuration:
# parameters prefixed with `tf_conf__` will configure chaosterraform driver
  tf_conf__silent: false
  tf_conf__retain: true

controls:
  - name: "Deploy Terraform module"
    provider:
      type: python
      module: chaosterraform.control

When both options are supplied configuration parameters supplied via the experiment configuration will be used.

Parameter Name	Usage
silent	Suppress Terraform console output to avoid verbose experiment logs, defaults to true
retain	Do not run terraform destroy at the end of the experiment to retain resources, defaults to false
chdir	Instruct Terraform to change its working directory before running subcommands

Set Terraform Input Variables

You can override input variables defined in the Terraform module from within the experiment using Chaos Toolkit configuration parameters prefixed by tf__:

configuration:
# parameters prefixed with `tf__` will set terraform input variables for the module
  tf__vpc_id: "vpc-0000000000"
  tf__number_of_azs: 2
  ...

controls:
  - name: "Deploy Terraform module"
    provider:
      type: python
      module: chaosterraform.control

Use Terraform Outputs In Chaos Experiments

If your Terraform module exports some output variables you can use them in the Chaos Toolkit experiments as regular experiment parameters. Such variables are added to the configuration context with the tf_out__ prefix.

For example, this Terraform module exports a load balancer DNS name:

terraform {
    ...
}

output "alb_dns_name" {
    value = aws_lb.application_lb.dns_name
}

We can use the exported DNS name in our chaos experiment like so:

controls:
  - name: "Deploy Terraform module"
    provider:
      type: python
      module: chaosterraform.control

steady-state-hypothesis:
  title: "Application is available"
  probes:
    - type: probe
      name: "should-respond-200"
      tolerance: 200
      provider:
        type: http
        url: "http://${tf_out__alb_dns_name}"
        method: "GET"
        timeout: 3