ecs instances not created with terraform

Question

So I'm trying to fire up a publicly hosted docker container into my ECS via terraform. The plan completes without any errors, and launches the service but the ECS instances are empty, and the service only complains that there are not ECS instances assigned to the cluster.

As part of my infrastructure I have defined:

• VPC
• security groups
• iam roles
• ecs cluster
• ecs service with ALB

Given that everything else launches and I can ssh to the ecs (and even manually pull and run the docker image) I'm pretty sure all the networking is ok, which makes me think it might be the connections between ECS services and ALB?

ECS service:

locals {
  application_name = "${var.environment}-kafka"
}

resource "aws_ecs_service" "kafka" {
  name                                = "${local.application_name}"
  iam_role                            = "${var.iam-role_arn}"
  cluster                             = "${var.ecs-cluster_id}"
  task_definition                     = "${aws_ecs_task_definition.kafka_definition.arn}"
  desired_count                       = "${var.count_kafka}"
  deployment_minimum_healthy_percent  = "${var.deployment_min_healthy_percent}"
  deployment_maximum_percent          = "${var.deployment_max_percent}"
  depends_on  = ["aws_alb_listener.kafka-alb-listener"]

  load_balancer {
    target_group_arn  = "${aws_alb_target_group.kafka-alb-target-group.arn}"
    container_port    = 80
    container_name    = "kafka"
  }

  lifecycle {
    create_before_destroy = true
  }
}

resource "aws_ecs_task_definition" "kafka_definition" {
  family                = "${var.environment}_kafka"
  container_definitions = "${data.template_file.kafka_task.rendered}"

  lifecycle {
    create_before_destroy = true
  }
}

data "template_file" "kafka_task" {
  template= "${file("${path.module}/kafka_task_definition.tpl")}"

  vars {
    kafka_docker_image = "${var.kafka-docker-image_name}:${var.kafka-docker-image_tag}"
  }
}

ALB:

resource "aws_alb" "kakfa-alb" {
  name                = "${local.application_name}-alb"
  security_groups     = ["${var.security_groups_ids}"]
  subnets             = ["${var.public_subnet_ids}"]

  tags {
    Name        = "${local.application_name}-alb"
    Environment = "${var.environment}"
  }
}

resource "aws_alb_target_group" "kafka-alb-target-group" {
  name                = "${aws_alb.kakfa-alb.name}-target-group"
  port                = "5000"
  protocol            = "HTTP"
  vpc_id              = "${var.vpc_id}"
  depends_on          = ["aws_alb.kakfa-alb"]

  health_check {
    healthy_threshold   = "5"
    unhealthy_threshold = "2"
    interval            = "30"
    matcher             = "200"
    path                = "/"
    port                = "traffic-port"
    protocol            = "HTTP"
    timeout             = "5"
  }

  tags {
    Name        = "${aws_alb.kakfa-alb.name}-target-group"
    Environment = "${var.environment}"
  }
}

resource "aws_alb_listener" "kafka-alb-listener" {
  load_balancer_arn = "${aws_alb.kakfa-alb.arn}"
  port              = "80"
  protocol          = "HTTP"

  default_action {
    target_group_arn = "${aws_alb_target_group.kafka-alb-target-group.arn}"
    type             = "forward"
  }
}

Answer 1

You probably haven't added the instances to the cluster.

user_data.sh:

#!/bin/bash

echo ECS_CLUSTER='${ecs_cluster_name}' > /etc/ecs/ecs.config

In your terraform config:

data "template_file" "user_data" {
  template = "${file("user_data.sh")}"

  vars {
    ecs_cluster_name = "${var.ecs_cluster_name}"
  }
}


resource "aws_launch_configuration" "current" {

  ...

  user_data = "${data.template_file.user_data.rendered}"

  ...

}