Unable to Create DAEMON Service In ECS Cluster with default_capacity_provider_strategy

pilchkinstein-perspectum commented 3 years ago

Community Note

Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
Please do not leave "+1" or other comments that do not add relevant new information or questions, they generate extra noise for issue followers and do not help prioritize the request
If you are interested in working on this issue or have submitted a pull request, please leave a comment

Terraform CLI and Terraform AWS Provider Version

Terraform v0.13.5

provider registry.terraform.io/hashicorp/aws v3.21.0

Affected Resource(s)

aws_ecs_service
aws_ecs_cluster

Terraform Configuration Files

Please include all Terraform configurations required to reproduce the bug. Bug reports without a functional reproduction may be closed without investigation.

resource "aws_ecs_service" "service" {
  name            = "service_16772"
  task_definition = aws_ecs_task_definition.task.family
  cluster         = "cluster_16772"

  scheduling_strategy = "DAEMON"

  placement_constraints {
    type = "distinctInstance"
  }
}

resource "aws_ecs_task_definition" "task" {
  container_definitions    = file("service.json")
  family                   = "task_16772"
  requires_compatibilities = ["EC2"]
}

resource "aws_ecs_cluster" "cluster" {
  name               = "cluster_16772"
  capacity_providers = ["cps_16772"]
  default_capacity_provider_strategy {
    capacity_provider = "cps_16772"
    weight            = "100"
    base              = "1"
  }
}

resource "aws_ecs_capacity_provider" "ecs_capacity_provider" {
  name = "cps_16772"
  auto_scaling_group_provider {
    auto_scaling_group_arn = aws_autoscaling_group.asg.arn
  }
}

resource "aws_autoscaling_group" "asg" {
  name             = "asg_16772"
  max_size         = 1
  min_size         = 1
  desired_capacity = 1

  launch_template {
    id      = aws_launch_template.lc.id
    version = aws_launch_template.lc.latest_version
  }

  availability_zones = ["us-east-1a"]
}

resource "aws_launch_template" "lc" {
  image_id = "ami-00ddb0e5626798373"
  instance_type = "t2.small"
}

// service.json

[
  {
    "name": "main",
    "image": "nginx",
    "cpu": 10,
    "memory": 512,
    "essential": true,
    "portMappings": [
      {
        "containerPort": 80,
        "hostPort": 80
      }
    ]
  }
]

Debug Output

bug-16772-debug.log

Panic Output

N/A

Expected Behavior

New DAEMON service should be created on each instance in the ECS cluster

Actual Behavior

No service is created. Terraform returns the following:-

Error: InvalidParameterException: Specifying a capacity provider strategy is not supported when you create a service using the DAEMON scheduling strategy. "cadvisor"

Steps to Reproduce

terraform apply -target=aws_launch_template.lc -target=aws_autoscaling_group.asg -target=aws_ecs_capacity_provider.ecs_capacity_provider
terraform apply -target=aws_ecs_cluster.cluster -target=aws_ecs_task_definition.task
terraform apply -target=aws_ecs_service.service

Important Factoids

None

References

0000

pilchkinstein-perspectum commented 3 years ago

This is now reproducible behaviour. An ECS cluster with a default capacity provider cannot create a DAEMON task. The workaround is:-

update the ECS cluster and remove the default capacity provider
create the DAEMON task
update the ECS cluster and add back the default capacity provider

wendtek commented 3 years ago

I was able to get around this by setting launch_type = "EC2" on the service. I got this tip from https://github.com/aws/containers-roadmap/issues/692 The strange thing is the AWS provider docs indicate EC2 is the default value for that argument, but it did not work until I explicitly set it on the service.

Edit: I see now, while my method successfully applied, the service was not deployed to the nodes in the default capacity provider. I had to change from using launch_type to manually specifying the capacity_provider_strategy, which results in the original error again.

pilchkinstein-perspectum commented 3 years ago

Any news on this? I'd hate to see it suddenly be archived due to inaction

infracaninophile commented 3 years ago

Contrary to @wendtek 's experience, I tried the launch_type = "EC2" explicit setting, and it worked fine for me: DAEMON processes successfully running on every instance in the cluster, whether using the default capacity provider or not.

terraform 0.15.3 aws-provider 3.45.0

dspv commented 2 years ago

Contrary to @wendtek 's experience, I tried the launch_type = "EC2" explicit setting, and it worked fine for me: DAEMON processes successfully running on every instance in the cluster, whether using the default capacity provider or not.

terraform 0.15.3 aws-provider 3.45.0

Worked for me, thanks

eyal-solomon1 commented 2 years ago

Contrary to @wendtek 's experience, I tried the launch_type = "EC2" explicit setting, and it worked fine for me: DAEMON processes successfully running on every instance in the cluster, whether using the default capacity provider or not.

terraform 0.15.3 aws-provider 3.45.0

worked for me too, thanks @infracaninophile

github-actions[bot] commented 1 week ago

Marking this issue as stale due to inactivity. This helps our maintainers find and focus on the active issues. If this issue receives no comments in the next 30 days it will automatically be closed. Maintainers can also remove the stale label.

If this issue was automatically closed and you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. Thank you!

hashicorp / terraform-provider-aws