resource/aws_autoscaling_group: Unexpected Removal of Target Group ARNs in version 2.21.0

[ghost]

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 "me too" comments, 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 Version

0.11.14

Affected Resource(s)

• aws_autoscaling_group

Terraform Configuration Files

3 resources similar to:

resource "aws_autoscaling_attachment" "application" {
  autoscaling_group_name = "${module.application.application_asg_name}"
  alb_target_group_arn   = "${aws_alb_target_group.application.arn}"
}

Debug Output

Panic Output

Expected Behavior

When planning using 2.20.0, No infrastructure changes are planned. Expectation is that this remains the case with 2.21.0.

Actual Behavior

Using 2.21.0 produced the following plan:

    ~ module.application1.module.application1.aws_autoscaling_group.main
        target_group_arns.#:          "1" => "0"
        target_group_arns.1012598726: "arn:aws:elasticloadbalancing:eu-west-1:xxxxxxxxxxx:targetgroup/application1/3c58ca2c0e49cb1b" => ""
  
    ~ module.application2.module.application2.aws_autoscaling_group.main
        target_group_arns.#:          "2" => "0"
        target_group_arns.3429299186: "arn:aws:elasticloadbalancing:eu-west-1:xxxxxxxxxxx:targetgroup/application2/63ed6bade1545859" => ""
        target_group_arns.703527089:  "arn:aws:elasticloadbalancing:eu-west-1:xxxxxxxxxxx:targetgroup/application2-private/d23ede06f42afcfb" => 

Steps to Reproduce

1. terraform apply

Important Factoids

References

[edmundcraske]

Not sure if this is caused by #9478 ?

[pabloajz]

This is happening also to me since today

[tcondeixa]

happening to me too, when you apply multiple times in a row it is alternating between destroying and creating

[darhon]

also faced with this problem

[bflad]

Hi folks 👋 Sorry for the unexpected new behavior here. 😖

The change within #9478 to start showing the difference of missing configurations of load_balancers and target_group_arns in the aws_autoscaling_group resource was good in isolation, but does not work in combination with the aws_autoscaling_attachment resource.

While we sort out the best course of action (potentially reverting the change and cutting a 2.21.1 bug fix release), you have a few options to deal with this situation:

• Update the affected aws_autoscaling_group resource configurations with the lifecycle configuration block ignore_changes argument available to all resources. This should be used where you are expectedly attaching load balancers outside the aws_autoscaling_group resource (e.g. with the aws_autoscaling_attachment resource):

# Compatible in all versions of Terraform

resource "aws_autoscaling_group" "example" {
  # ... other configuration ...

  lifecycle {
    ignore_changes = ["load_balancers", "target_group_arns"]
  }
}

# Terraform 0.12 and later syntax

resource "aws_autoscaling_group" "example" {
  # ... other configuration ...

  lifecycle {
    ignore_changes = [load_balancers, target_group_arns]
  }
}

• Hold back this change in behavior for now without any configuration changes by pinning the Terraform AWS Provider to the previous version, see also the Terraform documentation on provider versioning, e.g.

# Compatible in Terraform 0.10 and later

provider "aws" {
  # ... potentially other configuration ...

  version = "2.20.0"
}

# Terraform 0.12 and later only (applies provider version constraint globally)

terraform {
  # ... potentially other configuration ...

  required_providers {
    # ... potentially other configuration ...

    aws = "2.20.0"
  }
}

---

If we revert this change, it is likely to occur again with more proper notice in the next major version release of the Terraform AWS Provider, so if possible, we would recommend the ignore_changes configuration change for future proofing.

My apologies again for the unexpected and unannounced change in this behavior. We will update this issue in the next few hours with our plan of action.

[blalor]

If we revert this change, it is likely to occur again with more proper notice in the next major version release of the Terraform AWS Provider, so if possible, we would recommend the ignore_changes configuration change for future proofing.

This will be the suggested route forward when using aws_autoscaling_attachment? I understand there's no real coordination between resources, but that seems odd to me, lie aws_autoscaling_attachment is a 2nd-class resource.

[while1eq1]

This also affects terraform version: v0.11.11

[bflad]

Please note that we will be cutting a version 2.21.1 of the Terraform AWS Provider shortly with a fix that reverts the change. Associated pull request: #9518

As mentioned above, this change will likely be redone in a future major version of the Terraform AWS Provider to fix the behavior seen in: #5197

[bflad]

This has been released in version 2.21.1 of the Terraform AWS provider. Please see the Terraform documentation on provider versioning or reach out if you need any assistance upgrading.

For further feature requests, documentation updates, or bug reports with this functionality, please create a new GitHub issue following the template for triage. Thanks!

[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 feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. Thanks!