Quick Start

In a hurry? No sweat! Here’s a quick start to using ufo that takes only a few minutes. For this example, we will use a Sinatra app from tongueroo/demo-ufo. The ufo init command sets up the ufo directory structure in your project. The ufo ship command deploys your code to an AWS ECS service. The ufo ps and ufo scale command shows you how to verify and scale additional containers.

gem install ufo
git clone https://github.com/tongueroo/demo-ufo.git demo
cd demo
ufo init --image=tongueroo/demo-ufo # NOTE: use your own account
ufo current --service demo-web
ufo ship
ufo ps
ufo scale 2

Note: These quick start commands assume that you have push access to the repo. Refer to the Notes “Repo Push Access” section below for more info. It also assumes that you’d like to use ECS EC2 and assume that you have an ECS cluster with some EC2 Container instances running. Refer to the Notes “ECS EC2 vs ECS Fargate” section below for more info.

What Happened

The ufo ship demo-web command does the following:

  1. Builds the Docker image and pushes it to a registry
  2. Builds the ECS task definitions and registry them to ECS
  3. Updates the ECS Service

You should see something similar to this:

$ ufo init --app=demo --image=tongueroo/demo-ufo
Setting up ufo project...
      create  .env
      create  .ufo/settings.yml
      create  .ufo/task_definitions.rb
      create  .ufo/templates/main.json.erb
      create  .ufo/variables/base.rb
      create  .ufo/variables/development.rb
      create  .ufo/variables/production.rb
      create  Dockerfile
      create  bin/deploy
      append  .gitignore
Starter ufo files created.
$ ufo ship demo-web
Building docker image with:
  docker build -t tongueroo/demo-ufo:ufo-2017-09-10T15-00-19-c781aaf -f Dockerfile .
....
Software shipped!
$ ufo ps
+----------+------+-------------+---------------+---------+-------+
|    Id    | Name |   Release   |    Started    | Status  | Notes |
+----------+------+-------------+---------------+---------+-------+
| f590ee5e | web  | demo-web:85 | 1 minutes ago | RUNNING |       |
+----------+------+-------------+---------------+---------+-------+
$ ufo scale 2
Scale demo-web service in development cluster to 2
$

Congratulations! You have successfully deployed code to AWS ECS with ufo. It was really that simple 😁

Note: This quick start requires a working Docker installation. For Docker installation instructions refer to to the Docker installation guide.

Notes

Repo Push Access

The quick start example pushes the Docker image to a Dockerhub repo. You need your own repo with push access. This repo could be your own Dockerhub account or ECR repo. You can control the setting with the --image option. Example:

ufo init --image=yourusername/yourrepo # use your own account

Also, if you are using ECR instead, you can specify an ECR repo with the --image option. Example:

ufo init --image 123456789012.dkr.ecr.us-west-2.amazonaws.com/myimage

For more info, refer to the ufo init reference docs.

ECS EC2 vs ECS Fargate

Ufo does not create the EC2 servers themselves to run the ECS tasks. If you use ufo ship to deploy an application to ECS EC2 and have not set up the EC2 servers, then the CloudFormation update will not be able to provision the ECS tasks and eventually roll back. Essentially it cannot create the ECS tasks because there are no EC2 servers to run them.

Refer to the AWS Creating a Cluster docs to create an ECS cluster. Also refer to ECS Terms Tutorial for an explanation of ECS terms.

If you would like not to manage the EC2 server fleet, you are looking for ECS Fargate instead of ECS EC2. ECS Fargate allows you to run ECS Tasks and AWS will manage the EC2 server fleet for you. Refer to the Ufo Fargate docs and use those quick start like commands instead. The pricing for Fargate is more because AWS manages the server fleet for you. Refer to Heroku vs ECS Fargate vs EC2 On-Demand vs EC2 Spot Pricing Comparison for a pricing comparison.

Learn more in the next sections.

Pro tip: Use the <- and -> arrow keys to move back and forward.

Edit this page

See a typo or an error? You can improve this page. This website is available on GitHub and contributions are encouraged and welcomed. We love pull requests from you!