(aws-ecs-patterns): Enable autoscaling with ApplicationLoadBalancedEc2Service

[5nafu]

General Issue

I would like to use an ApplicationLoadBalancedEc2Service together with autoscaling of EC2 instances.

The Question

I would like to use an ApplicationLoadBalancedEc2Service together with an EC2 ASG. To enable autoscaling of the ECS instances, I would need to apply the capacity_provider_strategies parameter.
The L2 @aws-cdk/aws-ecs.Ec2Service already supports this option, but the L3 construct does not.

Unfortunately the EC2Service does not allow to add this strategy later to the finished service either.

CDK CLI Version

2.10.0

Framework Version

No response

Node.js Version

16.3.0

OS

OSX

Language

Typescript

Language Version

No response

Other information

No response

[peterwoodworth]

Hey @5nafu,

While you can't directly set this property on the L3 construct, you can use escape hatches to apply this property to the underlying service.

    const l3 = new ApplicationLoadBalancedFargateService(this, 'ALBFargateService', {
      memoryLimitMiB: 1024,
      taskImageOptions: {
        image: ContainerImage.fromRegistry('test'),
        environment: {
          TEST_ENVIRONMENT_VARIABLE1: "test environment variable 1 value",
          TEST_ENVIRONMENT_VARIABLE2: "test environment variable 2 value",
        },
      },
      desiredCount: 2,
    });

    const l1 = l3.service.node.defaultChild as CfnService;

    const cpsprops = {
      base: 2,
      weight: 2,
      capacityProvider: 'capacityProvider'
    }

    l1.addPropertyOverride('CapacityProviderStrategy', cpsprops);

[5nafu]

Hi @peterwoodworth ,

thanks for the quick response. Unfortunately the above code results in a Specifying both a launch type and capacity provider strategy is not supported. (Adding the required []around cpsprops on the last line).

I needed to add l1.addPropertyDeletionOverride('LaunchType') to make it work.

So the complete code would be (adding as reference for future google searches):

    const l3 = new ApplicationLoadBalancedEc2Service(this, 'ALBFargateService', {
      memoryLimitMiB: 1024,
      taskImageOptions: {
        image: ContainerImage.fromRegistry('test'),
        environment: {
          TEST_ENVIRONMENT_VARIABLE1: "test environment variable 1 value",
          TEST_ENVIRONMENT_VARIABLE2: "test environment variable 2 value",
        },
      },
      desiredCount: 2,
    });

    const l1 = l3.service.node.defaultChild as CfnService;

    const cpsprops = [{
      base: 2,
      weight: 2,
      capacityProvider: 'capacityProvider'
    }];

    l1.addPropertyOveride('CapacityProviderStrategy', cpsprops);
    l1.addPropertyDeletionOverride('LaunchType');

That said, I would like to request the addition of the capacity_provider_strategies parameter to the EC2-LB pattern ApplicationLoadBalancedEc2Service as using autoscaling groups with an EC2 based ECS cluster seems to be the standard use case, and instances-autoscaling does not work without this setting (at least in our tests).

Cheers and thanks again for the hint with the escape hatches.

[peterwoodworth]

Thanks for cleaning the code up, I only synthesized my code 🙂

I'll convert this to a feature request. Since this has a relatively easy workaround we don't have a timeline for it, but we happily accept community contributions. Check out our contributing guide if you're interested

[gshpychka]

If I may, I would suggest that if you're running into issues like this, it's time to switch from L3 to L2 constructs. L3 are very opinionated and limit the available options by design, to make it easy for newcomers to spin up complex solutions. Once you're comfortable with it and are in need of deeper configuration, I'd go with L2 to have full control.

[peterwoodworth]

@gshpychka thank you for the input, I tend to agree here, and directly implementing the prop just for the L2 construct wouldn't help with the abstraction the L3 is trying to offer. However, I believe there's a case for somehow abstracting the ability to autoscale the instances created 🙂

[github-actions]

⚠️COMMENT VISIBILITY WARNING⚠️

Comments on closed issues are hard for our team to see.
If you need more assistance, please either tag a team member or open a new issue that references this one.
If you wish to keep having a conversation with other community members under this issue feel free to do so.