Skip to content

Terraform modules repo in the style of "Terraform: Up & Running"

License

Notifications You must be signed in to change notification settings

percygrunwald/terraform-modules

Repository files navigation

terraform-modules

ci

Terraform modules repo in the style of Terraform: Up & Running. Modules from this repo can be referenced by an "infrastructure" repo that deploys live infrastructure, usually with terragrunt:

terraform {
  source = "${get_env("MODULES_REPO_GIT_URL", "")}//modules/aws/ec2_instance?ref=v1.2.3"
}

Development setup

The setup scripts are designed for asdf version manager. If you aren't using asdf, you can still use the setup script, but all tools in .tool-versions must be installed before running the script.

The setup script is idempotent and ensures all tools, global packages and hooks are installed.

# Set up dev environment
script/setup_development.sh

Running the examples and tests

In the spirit of Terraform: Up & Running, each module has all of the following:

  • A module definition at modules/PROVIDER/MODULE
  • A runnable example at examples/PROVIDER/MODULE
  • A test at test/examples_PROVIDER_MODULE_test.go

Run the examples

# Change to the example directory
cd examples/aws/ec2_instance

# Load AWS creds from ~/.aws/credentials
export AWS_PROFILE=percy_test-github_actions

# Initialize terraform
terraform init

# Launch infrastructure
terraform apply

# Destroy infrastructure
terraform destroy

Modifying the example vars:

terraform apply -var "aws_region=us-west-2"

Run the tests

# Load AWS creds from ~/.aws/credentials
export AWS_PROFILE=percy_test-github_actions

Run all the tests (-count 1 is required to prevent caching):

go test ./test/ -v -timeout 60m -count 1

Run an individual test:

go test ./test/ -v -timeout 60m -run TestExamplesAWSAppInstance -count 1

Standard tagging scheme

This repo adopts a standardized and flexible tagging scheme for infrastructure:

  • Org (org_name_underscore, all modules): The name of the organization/company
  • Team (team_name_underscore, all modules): The name of the team/project within the org
  • Environment (env, all modules): The name of the service/application environment (e.g. dev, prod, etc.)
  • Service (service_name_underscore, most modules): The name of the service/application within the env

The implicit logical hierarch is that:

  • A team/project exists within an org/company
  • An environment exists within a team/project
  • A service/application exists within an environment

Org > Team > Env > Service

Continuous Integration (CI)

Testing the CI pipeline locally

You can test the CI pipeline (Github Actions) locally using nektos/act. Requires docker.

The AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and GITHUB_TOKEN secrets must be passed or set in the current env. If any of these are not set, act will prompt you to enter them. GITHUB_TOKEN is required for some actions and can be a personal access token with empty permissions.

--reuse reuses the containers for each workflow job, keeping all installed tools/dependencies. This is recommended for frequent runs since act cannot make use of actions caching, which means all tools/dependencies must be downloaded each time. If you ever want to start again from scratch (empty container), just run without --reuse.

# Install act with go (see docs for other installation options)
go install github.com/nektos/act@latest

# Export required env vars
export AWS_ACCESS_KEY_ID=...github_actions user for test account...
export AWS_SECRET_ACCESS_KEY=...github_actions user for test account...
export GITHUB_TOKEN=...nektos/act personal access token...

# Run the CI pipeline locally
act -s AWS_ACCESS_KEY_ID -s AWS_SECRET_ACCESS_KEY -s GITHUB_TOKEN --reuse