Terraform Controller

Introduction

This is a custom Kubernetes controller designed to run in the Scipian namespace. It watches for changes on incoming Custom Resources and launches Jobs to create/update/destroy infrastructure using Terraform.

It is built with Kubebuilder 2.0, with full documentation found here.

Setting Up the Cluster

The Scipian Terraform Controller expects a few things to be set up in the cluster it will run in:

1. A scipian namespace
2. A secret named scipian-aws-iam-creds with AWS IAM secret accesss key and access key ID as aws_access_key_id and aws_secret_access_key respectively. These creds are for Scipian's S3 bucket where it will access Terraform State, and should be for that AWS account. NOTE: These should be base64 encrypted. In order to avoid new line characters in the base64 encrypted string, use the following flags when encrypting: echo -n <aws_cred> | base64 -w 0.
3. An S3 bucket and corresponding DynamoDB table. Set these in config/manager/manager.yaml in the ConfigMap section. NOTE: The DynamoDB table should have the same name as the S3 bucket, but with -locking appended to it.
4. make install - installs Custom Resource Definitions (CRDs) into the cluster

Running Locally

To run the project locally for developing:

1. Using Direnv, set up your .envrc file with SCIPIAN_STATE_BUCKET and SCIPIAN_STATE_LOCKING pointing to your desired s3 bucket and DynamoDB table respectively.
2. go get
3. make install
4. make run (this will run against the cluster defined in $HOME/.kube/config)

Deploying in Cluster

To deploy the controller in a cluster:

1. make docker-build
2. make docker-push
3. make deploy

Testing

This project uses Ginkgo as a BDD testing framework. Make sure to have Ginkgo installed locally.