briancaffey/terraform-aws-django

Terraform modules for deploying Django applications on AWS using ECS Fargate

Terraform AWS Django

About

This repo is a collection of Terraform modules for deploying Django applications on AWS using EC2, ECS (EC2 + Fargate), and EKS.

Currently this repo contains one high-level module for deploying a Django application on AWS using ECS (Fargate launch type).

See the examples/ directory for examples of how to use this module.

Here is a companion repo that shows how to use this module with a Django application: https://github.com/briancaffey/django-step-by-step.

Also see https://github.com/briancaffey/terraform-aws-ad-hoc-environments for a repo that shows how to use this module with an ad-hoc environment.

terraform-aws-django's Issues

change var name log_stream_name to log_stream_prefix

split IAM out of ecs module into another module

add outputs for frontend service

Keep RDS from automatically upgrading minor version

When an RDS instance is created, it will automatically upgrade the minor version which causes state file issues. This happens because AWS automatically upgrades the RDS engine version to the latest version for security and performance reasons.

For example, this repo currently has engine_version set to “13.4”, and a new deployment will work as expected (at 13.4). However, AWS will eventually upgrade this to “13.7” (I think during a maintenance window, likely hours after you were last looking at the base deployment).

This can cause upleasant surprises when you go to update your base stack and it thinks you want to downgrade (which ultimately fails):

  # module.main.module.rds.aws_db_instance.this will be updated in-place
  ~ resource "aws_db_instance" "this" {
  	~ engine_version                    	= "13.7" -> "13.4"
    	id                                	= "ocpython-rds"
    	name                              	= "postgres"
    	tags                              	= {}
    	# (49 unchanged attributes hidden)
	}

To prevent this issue, we should be able to set the auto_minor_version_upgrade parameter to false when creating the RDS instance, which will prevent automatic minor version upgrades.

Add autoscaling module for horizontal scaling on high CPU

Add a module that can be used to do autoscaling on a service. Here are some references:

• https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/appautoscaling_policy#ecs-service-autoscaling

• https://github.com/victorgomezg93/terraform-aws-go/blob/af449339fb061482fc169f628c752102d353662a/infraestructure/auto_scaling.tf

• Add autoscaling target

Use Fargate instead of EC2 Launch type for ECS cluster

change service settings to speed up deployment of new services

Implement the ideas from this article: https://nathanpeck.com/speeding-up-amazon-ecs-container-deployments/ in order to speed up the update of services

Avoiding issues with lists in Terraform

Problem

In Terraform, when I have a list of 2 items (e.g. A, B) it doesn’t always preserve the order and sometimes thinks there is a change (e.g. B, A).

I’m trying to remember where exactly I saw it in this project, but I’ve been occasionally running into this for years and finally dug around for a potential solution.

Potential Solution

In Terraform, when using lists, the order of the elements is not guaranteed to be preserved. To make Terraform aware of the order of the elements in the list and avoid changes, we can use the tuple data type instead of a list. A tuple is an ordered collection of elements that can be of different types, and Terraform will treat each element of the tuple as a separate resource. This way, Terraform will be aware of the order of the elements and will not consider changes when the order is changed.

Alternatively, we might be able use the jsonencode function in some cases to convert the list to a json string and use that as input. This will preserve the order of the elements in the list and Terraform will not consider changes when the order is changed.

locals {
  ordered_list = jsonencode([ "A", "B"])
}

and use the ordered_list variable in your resource configuration.