ECS with Rails

This demo shows how to deploy a ruby application on ECS with the hoop agent.

Requirements

ECS cluster

Hoop Command Line

1. Login at Hoop and generate a client key

shell
hoop login
HOOP_KEY=$(hoop admin create agent ecs --mode embedded)

2. Create the application demo image

shell
# https://github.com/hoophq/ruby-getting-started
IMAGE=hoophq/ruby-getting-started

If you want to use your own image for this demo, build and push the image to your own registry. In a production setup use the Elastic Container Registry (ECR) service

Embeeded Agent

This demo requires that the hoop agent command is installed locally within your application. Refer to docker installation to learn how to install it.

1. Create the Task Definition

Before creating the task definition, make sure that you have:

The Task Role ARN

The log configuration group configured properly

shell
HOOP_CONNECTION=demo-rails
AWS_REGION=us-east-2
AWSLOG_GROUP=/ecs/webapp
ROLE_ARN=arn:aws:iam::200074533906:role/ecsTaskExecutionRole
cat <<EOF > /tmp/$HOOP_CONNECTION.json
{
    "family": "$HOOP_CONNECTION",
    "containerDefinitions": [
        {
            "name": "$HOOP_CONNECTION",
            "image": "$IMAGE",
            "portMappings": [{ "containerPort": 3000, "hostPort": 3000, "protocol": "tcp" }],
            "entryPoint": ["hoopstart", "--", "bundle", "exec", "puma", "-C", "config/puma.rb"],
            "environment": [
                { "name": "HOOP_KEY", "value": "$HOOP_KEY" },
                { "name": "HOOP_CONNECTION", "value": "$HOOP_CONNECTION" }
            ],
            "logConfiguration": {
                "logDriver": "awslogs",
                "options": { "awslogs-group": "$AWSLOG_GROUP", "awslogs-region": "$AWS_REGION", "awslogs-stream-prefix": "ecs" }
            }
        }
    ],
    "taskRoleArn": "$ROLE_ARN",
    "executionRoleArn": "$ROLE_ARN",
    "networkMode": "awsvpc",
    "requiresCompatibilities": [ "FARGATE" ],
    "cpu": "512",
    "memory": "1024"
}
EOF

shell
aws ecs register-task-definition \
    --region=$AWS_REGION \
    --cli-input-json file:///tmp/$HOOP_CONNECTION.json

2. Create the Service

Before you proceed, make sure to add:

The name of your cluster

The VPC, subnets and the security group for the AWS VPC Configuration

shell
CLUSTER=dev
VPC_CONFIG="awsvpcConfiguration={subnets=[subnet-xxx01,subnet-xxx02],securityGroups=[sg-xxxx],assignPublicIp=ENABLED}"

shell
aws ecs create-service --cluster $CLUSTER \
    --service-name=$HOOP_CONNECTION \
    --task-definition=$HOOP_CONNECTION \
    --region=$AWS_REGION \
    --desired-count=2 \
    --launch-type FARGATE \
    --platform-version LATEST \
    --network-configuration=$VPC_CONFIG

Wait for the service to start the web application

3. Connect to the application via console (bash)

Create the connection before proceeding

shell
hoop admin create connection $HOOP_CONNECTION

Now can to interact with it

Start a bash console

shell
hoop connect $HOOP_CONNECTION

this command is analogous to /bin/bash

Start a rails console

shell
hoop connect $HOOP_CONNECTION -- -c 'rails console'

this command is analogous to /bin/bash -c 'rails console'

Execute one off commands

shell
hoop exec $HOOP_CONNECTION -i 'pp Rails.env' -- -c 'rails runner -'

this command is analogous to echo -n 'pp Rails.env' | /bin/bash -c 'rails runner -'

Creating a connection that exposes /bin/bash as entrypoint could be danger. An alternative is defining directly which command to expose it, like rails console, or rails runner -