Add dependency service deployment on AWS with terraform

Question

I want to find a solution to deploy one service called service2 only if the service service service3 is started and healthy. I use terraform to manage my infrastructure and deploy my different tasks.

What I tried to do is to add depends_on condition to my service:

resource "aws_ecs_service" "service2" {
  name                               = "service2"
  cluster                            = module.ecs_cluster_id
  task_definition                    = aws_ecs_task_definition.service2.arn
  desired_count                      = 1
  launch_type                        = "FARGATE"
  deployment_minimum_healthy_percent = 100
  deployment_maximum_percent         = 200
  force_new_deployment               = true
  depends_on                         = [aws_ecs_service.service3]
  
  propagate_tags = "SERVICE"

I also added for_new_deployment to force to deploy again even if there is no changes. When I deploy my service2, it seems wait until service3 stops deploying, but I have no guarantee that my service 3 is healthy and I'm not sure that the depends_on is the solution of my problem.

Do you have some recommendations?

Answer 1

You need to add wait_for_steady_state = true to your service3 ECS service definition. Then Terraform will wait for service3 to be fully deployed and healthy before it proceeds with the rest of the deployment.