ECS Service Module

A simplification to get ECS Services up and running, without having to think about the details.

This module helps you get up and running with an ECS Service. It sets up your service, and links it with VPCs, load balancers, monitoring, and more.

• Examples
• Arguments
  • application_container and sidecar_containers
  • lb_health_check
• Considerations
  • Platform differences
  • Migrating from terraform-aws-ecs-fargate and terraform-aws-ecs-anywhere
• FAQ

Examples

The examples folder is the home of example usages of this module. These are the examples you can find there.

Simple

The minimum usage of this module.

On Prem

Run your service directly in an on-prem datacenter using ECS Anywhere.

Autoscaling

A simple service using autoscaling to handle increased load.

Autoscaling with custom metrics

A service using autoscaling to handle increased load based on custom metrics.

Arguments

These are the arguments that can not be expressed by the terraform module

application_container and sidecar_containers

These arguments are where you specify everything ECS needs to know about your container and how to run it.

Both use the same format, but application_container will automatically be set as an essential container, while sidecar_container will not be set as essential.

Name	Description	Required	Default
name	The name of the container, must be unique.	yes	n/a
image	The image (and tag) of the container to run	yes	n/a
command	A command to run on the container	no	null
essential	If the container is essential for the service	no	true if application_container, false otherwise
environment	A map of environment variables	no	Empty map
secrets	A map of secrets	no	Empty map
port	A port to expose from the container.	depends (required for application_container)	null
protocol	The application layer protocol for the exposed port.	depends (required for application_container)	null
health_check	Container health checks. Not to be confused with LB health checks.	no	null
cpu	The number of CPU units reserved for this container.	no	null
memory_hard_limit	The max amount of memory that the specific container can consume.	no	null
memory_soft_limit	A soft memory limit, that ECS will do best effort to follow when memory is lacking for the service.	no	null
extra_options	Configure any other options for container definitions	no	None (Though log_configuration is automatically configured for CloudWatch)
placement_constraints	A list of placement constraints for the service. Not valid for FARGATE launch_type. SEE: https://docs.aws.amazon.com/AmazonECS/latest/developerguide/cluster-query-language.html and https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/ecs_service#placement_constraints	no	[]

lb_health_check

These are health checks that will be executed by the loadbalancer.

See the health_check documentation for aws_lb_target_group.

Considerations

Platform differences

The different platforms are quite similar, but some considerations have to be taken.

Fargate

You must make sure that the following arguments are present:

• lb_listeners

• private_subnet_ids

• vpc_id

ECS Anywhere

Please keep the considerations from AWS in mind. Only use ECS Anywhere if it is really neccessary. https://docs.aws.amazon.com/AmazonECS/latest/developerguide/ecs-anywhere.html#ecs-anywhere-considerations

Main things to note:

• There is no VPC networking

• No loadbalancers

Migrating from terraform-aws-ecs-fargate and terraform-aws-ecs-anywhere

A guide is available if you were previously using terraform-aws-ecs-fargate or terraform-aws-ecs-anywhere.

Go check out the documentation about moving from old modules!

FAQ

• Missing required arguments?

  • Newer versions of this module requires hashicorp/aws version >= 5.0.0