Deployment Examples Fargate (dxfar) Pipeline

The example is designed to be run from a blank AWS and Terraform account as no assumptions are made about pre-existing knowledge or infrastructure.

In this example we create a Fargate cluster, an Aurora Postgres database and launch a node server into the cluster.

The basic sequence of steps is as follows:

• Create an AWS and Terraform account and save credentials
• Create the deploymentuser in AWS so we are not using root credentials
• Create the VPC
• Create the database
• Create an ECR repo to push images to
• Create the reusable ECS infrastrcture like security groups, roles, the cluster itself etc.
• Build the image and push to the repo
• Deploy an image to the cluster
• See it running in the logs
• Create a deployment pipeline in Github Actions

Get Deployed

1. Prerequisites

• Get an AWS account and look up your root credentials. Login -> Your Name -> My Security Credentials -> Access Keys -> Create New Access Key.
• Register a domain name for your application in AWS and set the TF_VAR_dx_root_domain environment variable to the domain name. Route53 -> Registered Domains -> Register Domain -> Follow the prompts
• Create a Terraform account.
• brew install jq - used in the script: ./bin/docker-login

2. Setup Terraform Cloud

• Create a terraform account
• run terraform login in your console

Setup workspaces:

./bin/tf-workspaces-create

3. Create and Configure AWS DeploymentUser

Create a profile from your root credentials (this profile can be deleted after this section):

aws configure --profile root
# enter root AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY
# note that the profile name "root" is used by the infra/user terraform script

Create DeploymentUser:

./bin/tf-user-create # creates deploymentuser with permissions to deploy

Now configure the credentials of the DeploymentUser (outputted to the console from the last step) and configure them locally:

aws configure --profile deploymentuser
# enter the credentials printed from the Create DeploymentUser step
# note the profile name "deploymentuser" is used by all further terraform scripts

4. Install Base Infra

./bin/tf-vpc-create dev
./bin/tf-pg-create dev
./bin/tf-ecr-create
./bin/tf-cluster-create dev
./bin/docker-login
./bin/docker-app-build
./bin/docker-app-push
./bin/tf-app-create dev
./bin/logs dev

5. Github Action for Deployment

Go into your Repository -> Settings and add the following variables:

• TF_API_TOKEN - your terrform API token (create a new one using terraform login if you need to)
• AWS_ACCESS_KEY_ID - your deploymentuser access key ID from the earlier step
• AWS_SECRET_ACCESS_KEY - your deploymentuser secret key from the earlier step

You'll also need to edit .github/workflows/build.yml and change TF_VAR_dx_root_domain: deploymentexamples.com to your domain name.

Now push commits will deploy automatically to dev.

Local Docker Compose

Run the pg+node stack locally with:

docker compose up --build