fix: ignore changes when deployment controller is codedeploy

[inhumantsar]

what

• adding yet another ecs_service block to work around this issue: ECS Service can't update desired replicas when Blue Green deployment is enabled  hashicorp/terraform-provider-aws#13658

why

• when deployment_controller is CODE_DEPLOY, network_configuration, task_definition, and desired_count must all be ignored.

references

• ECS Service can't update desired replicas when Blue Green deployment is enabled  hashicorp/terraform-provider-aws#13658
• Less weird than Add possible disable created service and propose new approach about customize ignore_changes #119 and fixes the same issue

[nitrocode]

I don't think adding additional aws_ecs_service resources are a good idea. It's a lot to support and we were already pushing it with the last 2 additional resources.

See this PR #119 where the attempt is to output the information so you can construct your own aws_ecs_service outside of this repository.

[Nuru]

I agree with @nitrocode , this module is already too repetitive, and it is a workaround for an upstream bug that has an approved PR fix. Let's see if we can wait it out.

We can reopen this in October 2021 for reconsideration if the upstream bug has still not been fixed.

[akramfstg]

Hi,
I am encountering the same issue with my ECS service using Terraform 1.0.4. Here is a brief declaration of my resources:

resource "aws_ecs_service" "transcoder" {
  name            = local.prefix == "production" ? "transcoder" : "${local.prefix}-api"
  cluster         = aws_ecs_cluster.main.name
  task_definition = aws_ecs_task_definition.transcoder.family
  desired_count   = 1
  launch_type     = "FARGATE"
  #wait_for_steady_state = true

  # Type of deployment controller
  deployment_controller {
    type = "CODE_DEPLOY"
  }

And the code deployment group configuration is the following:

resource "aws_codedeploy_deployment_group" "transcoder_blue_green" {
  app_name               = aws_codedeploy_app.transcoder.name
  deployment_group_name  = "${local.prefix}-dg-transcoder"
  service_role_arn       = data.aws_iam_role.codedeploy.arn
  deployment_config_name = "CodeDeployDefault.ECSAllAtOnce"

  auto_rollback_configuration {
    enabled = true
    events  = ["DEPLOYMENT_FAILURE"]
  }

  blue_green_deployment_config {
    deployment_ready_option {
      action_on_timeout = "CONTINUE_DEPLOYMENT"
    }

    terminate_blue_instances_on_deployment_success {
      action                           = "TERMINATE"
      termination_wait_time_in_minutes = 5
    }
  }

  deployment_style {
    deployment_option = "WITH_TRAFFIC_CONTROL"
    deployment_type   = "BLUE_GREEN"
  }

  ecs_service {
    cluster_name = aws_ecs_cluster.main.name
    service_name = aws_ecs_service.transcoder.name
  }

  load_balancer_info {
    target_group_pair_info {

      prod_traffic_route {
        listener_arns = [aws_lb_listener.api.arn]
      }

      test_traffic_route {
        listener_arns = [aws_lb_listener.api.arn]
      }

      target_group {
        name = aws_lb_target_group.transcoder.name
      }

      target_group {
        name = aws_lb_target_group.green_transcoder.name
      }
    }
  }
}

I am unfortunately getting this exception that I could not understand or find a way out:

╷
│ Error: error updating ECS Service (arn:aws:ecs:eu-central-1:999618886922:service/1536-cluster/1536-api): InvalidParameterException: Unable to update task definition on services with a CODE_DEPLOY deployment controller. Use AWS CodeDeploy to trigger a new deployment.
│ 
│   with aws_ecs_service.transcoder,
│   on transcoder.tf line 59, in resource "aws_ecs_service" "transcoder":
│   59: resource "aws_ecs_service" "transcoder" {
````

Any suggestions please?

Thanks

[nitrocode]

@akramdhiabii please see the messages above. We're working on it by allowing you to disable the ecs service so you can take advantage of only the iam roles and task definition. See #119

[akramfstg]

@nitrocode Waiting for that would it be useful to taint the ECS service in addition to the above code deploy configuration in order perform the Blue/Green deployment?
Or is there any workaround available? that is actually impacting our deployments.
Thanks in advance.