jamesalbert / deploy-unity

Deploy your Unity WebGL game to Amazon's Elastic Container Service

deploy-unity

Deploy your Unity WebGL game to Amazon's Elastic Container Service using this handy-dandy virtual environment.

How to Deploy Unity Builds

Dependencies

• aws-cli
• ecs-cli
• jq

Things you need:

• a domain (guide)
• a TLS certificate (guide)
• an ECS cluster (guide) and service (guide)
  • name your service with the name you will set in config.json
  • recommended: set your service behind an ALB (guide)

Create a CNAME under your domain that points to the ALB (guide). Use the TLS certificate to encrypt the traffic between the client and the ALB (guide). Then configure a target group to forward traffic to your service's tasks (guide). This tool doesn't touch on any of these things, but you should have this setup before hosting a game.

Last thing to note is to ensure that the ECS task role you use have ecr:* and logs:* permissions. It doesn't have to be that open, but ¯\_(ツ)_/¯.

Configuration

Before you begin, be sure to have aws-cli and ecs-cli configured. If you don't already, follow this guide for aws-cli, and this guide for ecs-cli.

Once that's done, the first thing you need to do is fill out config.json:

• build_dir: build directory (where you built to from Unity)
• name: the name for your ECR repository, task definition, and service
• tag: tag your deploy (usually "latest")
• cluster: name of your cluster
• network_mode: bridge, host, awsvpc, or none
• task_role_arn: role arn the containers will assume
• execution_role_arn: role arn that container agents and docker daemon will assume
• network_mode: bridge, host, awsvpc, or none (not implemented, uses default)
• cpu: CPU units
• memory: MiB units

Deploying

Reminder: you must have a cluster and service ready before deployment.

If you're ready to go, here are the steps:

• In Unity:
  • click File>Build Settings...
  • click WebGL
  • configure your build
  • click Build and choose this repo as the destination
• Here:
  • run . deploy_env.sh
  • run deploy
  • when done, run deactivate

deploy will:

• login to ECR
• build the docker image (./Dockerfile)
• push the image to ECR
• create a new task definition revision
• update the service with the new task definiton

While in the deploy environment, you can:

• ecr-login: login to ECR
• build: build the local docker image
• push: push image (takes name from config.json.name) to ECR
• revise: create new task definition revision
• update: update the service with the new task definition
• deactivate: deactivate deploy environment

That's the basic flow of things. You can always version control your builds and use this in something like Jenkins.