Updating hosting from count: 1 to a load balanced configuration doesn't work with http.alb

[joelypolly]

We have an existing ALB which is populated in the http.alb field. When used in conjunction with a load balanced hosting setup as specified below we get an error in deploying the service. Note: Deploying a single host works (count: 1), Target Group shows IP in the Target Type field for single host deployments.

deploy service service to environment beta: deploy service: check if changeset is empty: create change set copilot-ID for stack app-beta-service: ValidationError: Template error: instance of Fn::GetAtt references undefined resource TargetGroup
	status code: 400, request id: ID: describe change set copilot-dd029c74-a3e8-49fe-9802-4f1d06593d22 for stack app-beta-service: ValidationError: Stack [app-beta-service] does not exist
	status code: 400, request id: ID: describe stack: describe stack events for stack app-beta-service: ValidationError: Stack [app-beta-service] does not exist
	status code: 400, request id: ID

Load Balancing Config

    http:
      # Import the existing ownit-shared-lb
      alb: arn:aws:elasticloadbalancing:us-east-1:account:loadbalancer/app/shared-beta-lb/restofarn
    count: 
      range: 1-2
      cooldown:
        in: 60s
        out: 30s
      cpu_percentage: 70
      memory_percentage: 80
      response_time: 2s

[iamhopaul123]

Looks like it is related to this bug #5485. We are working on a patch release which shall fix this issue.

[iamhopaul123]

Hello! The fix is now released in v1.32.1: https://github.com/aws/copilot-cli/releases/tag/v1.32.1

[joelypolly]

@iamhopaul123
Using the v1.32.1 doesn't seem to resolve the issue.
Same error as the original comment.

copilot --version
copilot version: v1.32.1

I delete the stack complete and ran it with the following yaml

  beta:
    cpu: 512 # Number of CPU units for the task.
    memory: 1024 # Amount of memory in MiB used by the task.
    platform: linux/arm64 # See https://aws.github.io/copilot-cli/docs/manifest/lb-web-service/#platform
    count: # Number of tasks that should be running in your service. https://aws.github.io/copilot-cli/docs/manifest/lb-web-service/#count-range
      range: 1-2
      cooldown:
        in: 60s
        out: 30s
      cpu_percentage: 70
      memory_percentage: 80
      response_time: 2s
    exec: true # Enable running commands in your container.
    network:
    connect: true # Enable Service Connect for intra-environment traffic between services.
    variables:
      IS_AWS: true
      ASPNETCORE_ENVIRONMENT: Beta
    http:
      # Import the existing shared-lb
      alb: arn:aws:elasticloadbalancing:us-east-1:accountid:loadbalancer/app/shared-beta-lb/id
      path: "/"
      # Healthcheck endpoint from the SystemModule
      healthcheck:
        path: "/api/system/heartbeat"
        healthy_threshold: 2
        unhealthy_threshold: 2
        grace_period: 10s
        interval: 6s
        timeout: 5s
      deregistration_delay: 5s # Speeds up deploys
      redirect_to_https: true
      alias: "example.com"
      hosted_zone: "ZONEID"

[iamhopaul123]

Hello @joelypolly. You are right. I am terribly sorry about that. It is because we didn't fix the part for target group referencing the ALB metric (response_time) when users import ALB. Would you mind using yaml patch for a temporary workaround?

- op: replace
  path: /Resources/AutoScalingPolicyALBAverageResponseTime/Properties/TargetTrackingScalingPolicyConfiguration/CustomizedMetricSpecification/Dimensions/1/Value
  value: !GetAtt TargetGroupForImportedALB.TargetGroupFullName

[joelypolly]

That fixed the problem. Thanks @iamhopaul123

[d-flood]

I'm experiencing a related problem. When trying to import an existing ALB into a Copilot service manifest that includes autoscaling, I get the following error:

✘ Proposing infrastructure changes for stack amendments-prod-amendments
✘ deploy service amendments to environment prod: deploy service: check if changeset is empty: create change set copilot-f99e579c-f5af-477e-b59c-fb58561aa26b for stack amendments-prod-amendments: ValidationError: Template error: instance of Fn::GetAtt references undefined resource TargetGroup
        status code: 400, request id: c2feacc3-42a0-464d-b45e-029c8b006127: describe change set copilot-f99e579c-f5af-477e-b59c-fb58561aa26b for stack amendments-prod-amendments: ChangeSetNotFound: ChangeSet [copilot-f99e579c-f5af-477e-b59c-fb58561aa26b] does not exist

The relevant portion of the service manifest:

    count:
      cooldown:
        in: 30s
        out: 60s
      cpu_percentage: 70
      memory_percentage:
        value: 80
        cooldown:
          in: 60s
          out: 160s
      requests: 1000 # problem is here
      response_time: 3s
      range:
        min: 1
        max: 3

When I remove requests, the service deploys deploys and uses the imported ALB.

[iamhopaul123]

Hello @d-flood. Does the yaml patch workaround work for you?