awesome-release / terraform-express-slack

Terraform

This project demonstrates the usage of Terraform on Release. It handles the creation of an AWS Gateway API and Lambda function that forwards all requests to a dockerized Node.js API where it is then logged and sent to a slack channel.

To make this project run in Release, simply create a new application with this repository, and follow the instructions below when creating the application template and configuring the env variables.

Application Template Setup

In the services section inside the application template, add a service for the terraform script. Ensure that image and build.context point to the directory where terraform and its dockerfile live:

...
services:
- name: terraform
  image: awesome-release/terraform-demo/terraform
  build:
    context: terraform/
  has_repo: true
  static: false
  command:
  - "/bin/sleep"
  - infinity
...

NOTE: This terraform service is a "dummy" service used only for the job runner in the next step. If you enable this service in the workflows section below, you can use the container as a shell location to execute debuging/troulbeshooting commands (like terraform output as one example.)

Below the services section, add a jobs section with a job to run the terraform script. Make sure the from_services field points to the service name created from the previous step:

...
jobs:
- name: terraform-runner
  completed_timeout: 1800
  from_services: terraform
  args:
  - apply -auto-approve
- name: terraform-destroyer
  completed_timeout: 900
  from_services: terraform
  args:
  - destroy -auto-approve
...

Lastly, modify the workflows section and add the jobs that were created in the previous step:

...
workflows:
- name: setup
  order_from:
  - jobs.terraform-runner
  - services.terraform       # optional
  - services.backend
- name: patch
  order_from:
  - jobs.terraform-runner
  - services.backend
- name: teardown
  order_from:
  - jobs.terraform-destroyer
  - release.remove_environment
...

Environment Variables

Copy the following env variables and fill in their values.

...
services:
  backend:
  - key: NODE_ENV
    value: production
  - key: PORT
    value: '3000'
  - key: SLACK_ACTIVE
    value: '1'                  # Set to 0 to disable slack messaging.
  - key: SLACK_TOKEN
    value: <slack token>
  - key: SLACK_CHANNEL_ID
    value: <channel id>
  terraform:
  - key: AWS_REGION
    value: us-west-2
  - key: TF_VAR_tld_name
    value: <tld name>
  - key: TF_VAR_certificate_arn
    value: <aws:acm:cert>
  - key: TFAWS_ROLE_ARN
    value: <aws:iam:role>
  - key: TFSTATE_BUCKET
    value: <bucket name>
  - key: TFSTATE_BUCKET_REGION
    value: <bucket region>
mapping:
  TF_VAR_env_id: "${RELEASE_ENV_ID}"
  TF_VAR_backend_ingress_url: "${BACKEND_INGRESS_URL}"
...

After these changes are made, you are ready to deploy!

Expected result

Upon successful deployment, there should be 4 instances in the details tab of your environment:

• backend - This is the node api.
• health-checker - Regular Docker container health checker.
• terraform - This container is the context for the runner. The terminal can be used to interact with Terraform and tear down the created AWS resources, among other things.
• terraform-runner - This is the terraform script job.

The terraform-runner instance will output the URL of the newly created Gateway API in the logs:

You can then curl it with a POST request like so:

curl -X POST "https://<app id>.execute-api.<region>.amazonaws.com/v1/new-post?message=Watson+come+here+I+need+you"

The request will trigger the lambda function to then make a request to the node api. You can verify this by checking the backend logs for your message, and slack if you have enabled it in the env config: