krishnan-mani / ecs-for-us

Workshop to deploy containers to an ECS cluster with an ALB

Outline

• Create a container for a (simple) API
• Create an ECS cluster with application load balancer
• Deploy the API to ECS
• Make a change, and deploy a new version

Pre-requisites

• build and run Docker images locally. You may choose to use docker-machine, OR alternatively, native support for Dockers in Windows or Linux
• personal account on Docker Hub
• AWS credentials for a privileged user
• Software installed
  • git
  • curl
  • docker
  • docker-for-windows (optional)
  • virtualenv

Workshop

Note: Certain instructions below assume the use of docker-machine. Ignore them if they are not applicable to you.

• Run API locally

$ virtualenv venv
$ source venv/bin/activate # or `source venv/Scripts/activate` on Windows
$ pip install -r requirements.txt
$ python ./example.py
$ curl --silent localhost:5000/example/
# Expect a response from the API running locally

• Create Docker image and run container locally

# obtain the docker-machine name
$ docker-machine ls 
$ eval $(docker-machine env [DOCKER_MACHINE_NAME])

$ docker build . -t my-service:1
$ docker images
$ docker run --name my-service-container -d -p 5000:5000 my-service:1
$ docker ps
$ docker-machine ip [DOCKER_MACHINE_NAME]
$ curl --silent [DOCKER_MACHINE_IP]:5000/example/
# Expect a response from the API running as a container
$ docker stop my-service-container

• Push the image to a registry on Docker Hub

# [USERNAME] is your username on Docker Hub 
$ docker tag my-service:1 [USERNAME]/my-service:1
$ docker login
$ docker push [USERNAME]/my-service:1

• Deploy the ECS cluster. Use starter-template to deploy a CloudFormation stack

$ export AWS_REGION=eu-west-1
$ export AWS_PROFILE=nonprod
$ aws configure list
$ aws sts get-caller-identity

$ cp starter-parameters.example.json starter-parameters.json
# Edit starter-parameters.json and supply the required information (from the `dev` environment)

# Choose a name for the CloudFormation stack you will provision, say [`ecs-workshop-USERNAME`]
$ aws cloudformation create-stack \
    --stack-name [ecs-workshop-USERNAME] \
    --template-body file://starter-template.json \
    --parameters file://starter-parameters.json \
    --capabilities CAPABILITY_IAM

$ aws cloudformation describe-stacks \
    --stack-name [ecs-workshop-USERNAME] \
    --query "Stacks[0].Outputs"
$ curl --silent [LOAD_BALANCER_DNS]

• Deploy our API to the cluster. Refer snippet.json to add the following for the API:
  • task definition
  • service description
  • target group
  • listener rule

$ cp starter-template.json template.json 
# Edit this template.json file for all subsequent changes

# Add the information needed for stack parameters
$ cp starter-parameters.json parameters.json

# Update the CloudFormation stack    
$ aws cloudformation update-stack \
    --stack-name [ecs-workshop-USERNAME] \
    --template-body file://template.json \
    --parameters file://parameters.json \
    --capabilities CAPABILITY_IAM
$ aws cloudformation describe-stack-events \
    --stack-name [ecs-workshop-USERNAME]
$ curl --silent [LOAD_BALANCER_DNS]/example/

• Modify service (bump DesiredCount) and witness scaling activity

$ aws ecs update-service \
    --service-arn [SERVICE_ARN] \
    --cluster [CLUSTER_NAME]
    --desired-count 3

• Deploy new version of the API

# Change application code, build new Docker image version, and push
$ docker build . -t my-service:2 -t [USERNAME]/my-service:2
$ docker push [USERNAME]/my-service:2 

# update the template and the parameters, then update the stack
$ aws cloudformation update-stack \
    --stack-name [ecs-workshop-USERNAME] \
    --template-body file://template.json \
    --parameters file://parameters.json \
    --capabilities CAPABILITY_IAM  

• Remember to delete your stack!

$ aws cloudformation delete-stack --stack-name [ecs-workshop-USERNAME]
$ aws cloudformation describe-stacks --stack-name [ecs-workshop-USERNAME]