Engerybox Devops

I attached the files I created for this test spring-petclinic.zip, or you can pull the project I forked to test all together: https://github.com/acevedomiguel/spring-petclinic

I used ECS for deployment since your company uses it but I have no experience using it, so I will give it a try.

Requirements

Docker daemon running (to build and dockerize the project)
AWS keys configured with permission to create the infrastructure (for simplicity it will use the default profile on ~/.aws/credentials)
Latest terraform installed

Installation

Unzip the files on the root folder of the demo project (not necessary for terraform but good to have all in the same place)

Define region to create

Update the file /terraform/versions.tf to define the default region for the resources, and can also set up the AWS if you don't want to use the default ones.

provider "aws" {
  region = "us-east-2"
  # access_key = "my-access-key"
  # secret_key = "my-secret-key"
}

Create the infrastructure with terraform

cd terraform/
terraform plan
terraform apply

Output

After apply, the output will give information to update the build script (build.sh) later:

  + AWS_ECR          < ECR where the dockerize image will be uploaded
  + ECS_CLUSTER_NAME < ECS cluster name created
  + ECS_SERVICE_NAME < ECS service name created
  + load_balancer_ip < URL/IP to access the site

When destroying the infra I found an issue, sometimes you need to go to the ECS console and stop the current task otherwise terraform will keep waiting forever.

Update build script

Whit the infra information please update the build.sh script, assign the ECR and ECS information to make the build and deploy

AWS_ECR="" 
AWS_REGION="" # optional, will try to guess it from the ECR URL
ECS_CLUSTER_NAME=""
ECS_SERVICE_NAME=""

Updating the project code

Once created the infra and updated the build script can be free to modify the code.

Once updated can just run the build code:

./build.sh

About the files

Terraform

This will be a simple explanation of the tf files, how to split could be a never ending discussion, I just tried to tied them up by functionality

Files	Short Description
`variables.tf`	I just allow the number of instances, could be more options but I didn't want to overengineer.
`ecr.tf`	Create the ECR where to push the docker image
`ecs.tf`	Creates the cluster, the service name, the task
`gateway.tf`	internet internet gateway, nat gateway, routing, and elastic ip.
`loadbalancer.tf`	Load balancer and target group to the ECS tasks
`main.tf`	VPC and AZ
`policyrole.tf`	role and policy for the ECS task and task executor.
`securitygroup.tf`	security rules for the load balancer access and the communication between the LB and the task
`subnets.tf`	creates 2 public subnets and 2 private subnets
`versions.tf`	AWS configuration and terraform versions

Docker

Multistage docker, after building it will use an alpine version to run the jar file.

I put the name of the JAR file name on the entry point and changing the version will require to change the Dockerfile:

one option will be to create and script to get the latest build jar
or for second option I know the Dockerfile can be created from the JAVA project instead of having a Dockerfile.

Build script

Just in a few words what the script it's doing:

Using the information from the infra (if not using terraform still can assign any other existing AWS infra)
First check the required variables to run otherwise will fail
Login to docker using AWS credential (required to push to ECR)
Docker build and push to ECR
With the ECS information it will update teh deploy to get the latest build

Comments

In an ideal scenario terraform should use it’s own keays with it’s own permissions.
No build.sh script, I always prefer a proper CI/CD, there are a lot of hardcoded values, not a visible status of the pipeline when working with a team, it was actually harder for me to do this way.
The terraform options (values.tf) it’s kept at the minimum, should be discussed and what values should be parametric.
I missed the autoscaling for ECS, since I have no experience. I still had some to learn there and I didn’t want to hack it or just copy and paste, this is what I could do in the the time dedicated to the test, hope is still enough.

acevedomiguel / spring-petclinic