AWS Application AutoScaling

[andskli]

Initial work on two new resource types in order to support AWS Application AutoScaling used by for example EC2 Container Service:

• aws_appautoscaling_target
• aws_appautoscaling_policy

In a sense, a lot of the code are similar or even the same as for the aws_autoscaling_* resources.

Example:

resource "aws_appautoscaling_target" "tgt" {
  service_namespace = "ecs"
  resource_id = "service/testcluster/testECSservice"
  scalable_dimension = "ecs:service:DesiredCount"
  role_arn = "arn:aws:iam::AWSAccountID:role/ecsAutoscaleRole"
  min_capacity = 1
  max_capacity = 4
}

resource "aws_appautoscaling_policy" "up" {
  name = "scale-up"
  service_namespace = "ecs"
  resource_id = "service/testcluster/testECSservice"
  scalable_dimension = "ecs:service:DesiredCount"
  adjustment_type = "ChangeInCapacity"
  cooldown = 60
  metric_aggregation_type = "Maximum"

  step_adjustment {
    metric_interval_lower_bound = 0
    scaling_adjustment = 1
  }

  depends_on = ["aws_appautoscaling_target.tgt"]
}

resource "aws_appautoscaling_policy" "down" {
  name = "scale-down"
  service_namespace = "ecs"
  resource_id = "service/testcluster/testECSservice"
  scalable_dimension = "ecs:service:DesiredCount"

  adjustment_type = "ChangeInCapacity"
  cooldown = 60
  metric_aggregation_type = "Maximum"

  step_adjustment {
    metric_interval_lower_bound = 0
    scaling_adjustment = -1
  }

  depends_on = ["aws_appautoscaling_target.tgt"]
}

resource "aws_cloudwatch_metric_alarm" "service_cpu_high" {
  alarm_name = "testhest-cpuutilization-high"
  comparison_operator = "GreaterThanOrEqualToThreshold"
  evaluation_periods = "2"
  metric_name = "CPUUtilization"
  namespace = "AWS/ECS"
  period = "60"
  statistic = "Maximum"
  threshold = "85"

  dimensions {
    ClusterName = "testcluster"
    ServiceName = "testECSservice"
  }

  alarm_actions = ["${aws_appautoscaling_policy.up.arn}"]
}

I've also included the applicationautoscaling part of the AWS SDK for Go using govendor as described in the contribution docs.

This is my first resource implementation for Terraform and I'm a novice in Go, so all feedback is welcome.

[andskli]

Just added a commit that bumps the aws-sdk-go version to v1.2.5.. I'm not sure how that's supposed to be handled, but as I state in the commit message, a version upgrade is necessary to support Application AutoScaling.

I upgraded all of the services and subsets of aws-sdk-go since a change in regards to where the V4 signer resides have been introduced, which many (if not all) service modules seems to depend on.

Maybe the aws-sdk-go update should be a separate PR?

[stack72]

Hi @andskli

I actually merged a PR that bumps the aws-sdk-go to 1.2.5 today :)

P.

[andskli]

Great! I've removed my commit doing that exact same thing and added a commit that adds just the applicationautoscaling part of the aws-sdk-go.
Acceptance tests works at my end (requires $AWS_ACCOUNT_ID to be set though).

The docs should be updated as well, anything else that I can or should do? :)

[stack72]

Hi @andskli

Please can you have a look at rebasing this? Then we can start reviewing it

Paul

[andskli]

@stack72 done!

[stack72]

Hi @andskli

Thanks so much for the work here. The code looks good but the tests don't pass I'm afraid

% make testacc TEST=./builtin/providers/aws TESTARGS='-run=TestAccAWSAppautoscalingPolicy_'
==> Checking that code complies with gofmt requirements...
go generate $(go list ./... | grep -v /terraform/vendor/)
TF_ACC=1 go test ./builtin/providers/aws -v -run=TestAccAWSAppautoscalingPolicy_ -timeout 120m
=== RUN   TestAccAWSAppautoscalingPolicy_basic
--- FAIL: TestAccAWSAppautoscalingPolicy_basic (112.53s)
    testing.go:264: Step 0 error: Error applying: 1 error(s) occurred:

        * aws_appautoscaling_target.tgt: Error creating application autoscaling target: ValidationException: Unable to assume IAM role: arn:aws:iam::xxxxxxx:role/ecsAutoscaleRole
            status code: 400, request id: 0c7dfe7b-4ffe-11e6-b1dc-798d7fb58f3e
FAIL
exit status 1
FAIL    github.com/hashicorp/terraform/builtin/providers/aws    112.550s
make: *** [testacc] Error 1

Can we make the ecsAutoscaleRole part of the test? Once we get this taken care of, this is ready for merge

Thanks

Paul

[stack72]

Ping @andskli :)

[andskli]

@stack72 sure, that makes sense. I think the fixes I just pushed takes care of what you're looking for.

I totally missed that the ecsAutoscaleRole is created only after fiddling around with AppAutoscaling + ECS in the GUI.

This also introduces a retry behaviour to creation of the scalable target, to avoid inconsistencies in IAM.

[stack72]

Hi @andskli

Just a FYI, I made a few tiny changes to the tests (76aea01) as i was getting a panic.

% make testacc TEST=./builtin/providers/aws TESTARGS='-run=TestAccAWSAppautoScaling'                             ✹
==> Checking that code complies with gofmt requirements...
go generate $(go list ./... | grep -v /terraform/vendor/)
TF_ACC=1 go test ./builtin/providers/aws -v -run=TestAccAWSAppautoScaling -timeout 120m
=== RUN   TestAccAWSAppautoScalingPolicy_basic
--- PASS: TestAccAWSAppautoScalingPolicy_basic (124.41s)
=== RUN   TestAccAWSAppautoScalingTarget_basic
--- PASS: TestAccAWSAppautoScalingTarget_basic (112.84s)
PASS
ok      github.com/hashicorp/terraform/builtin/providers/aws    237.270s

I just rebased and merged this :) Thanks for all the work here!

Paul

[andskli]

@stack72 thx! Happy to contribute.

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.