The code here bootstraps an infrastructure from a single entrypoint, the ./do script. Firstly, Ansible creates a state bucket for Terraform, and then Terraform can be run layer-by-layer to provision any kind of infrastructure. It works locally as well as through Jenkins, so long as the Environment variables are defined.
For Terraform, the ./do script runs both headlessly and interactively. Interactively, the script will offer options based on the available environments, available layers, and the three Terraform commands, plan, apply and destroy.
Updated the tags system to use Cloudposse's slick Terraform module here: https://github.com/cloudposse/terraform-null-label
To build a bare bones infra consisting on a single VPC with a single private subnet, do the following once the dependencies are installed and the environment variables are present.
./do bootstrap # this creates the S3 bucket for the Terraform state
./do
# And interactively choose an env, then the first layer, 001_vpc
# Then 'plan' or 'apply'
This will plan or build the infrastructure. When finished, run ./do again, but choose to destroy. Then,
./do teardown
This will remove the cleanup the state bucket in S3.
Python
Ansible
AWS Cli
Boto
Terraform
Bash
Copy envrc.template to .envrc and fill in the appropriate details.
Copy ./environments/999_example to a new directory under ./environments.
Create the S3 bucket required for Terraform state management:
./do bootstrap
Or:
ansible-playbook 000_bootstrap.yml
./do
And follow the prompts.
./do teardown
Or:
ansible-playbook 001_teardown.yml
This removes both the infra and then finally the S3 bucket for state management.
./do bootstrap
And
./do teardown
Then, the format is:
./do <command> -e <env_name> -l <layer_name>
Where available commands are plan, apply and destroy.
./do plan -e 999_example -l 001_vpc