fix: update supported tf version to have < 0.14.0

[aidanmelen]

modules fails to destroy EKS cluster in 0.14.5 because destroys don't refresh.

PR o'clock

Description

In terraform 0.14+ destroy command no loner refreshes the state of resources before generating execution plan (like it did in 0.13.X). terraform apply still does the refresh, that's why this issue is more frequent the more time elapses after applying.

#1162 (comment)

As a result, the module will fail to destroy all of the resources it manages.

I should say that there is a workaround in 0.14.5. The user will be required to run terraform refresh then terraform destroy.

Checklist

• CI tests are passing
• README.md has been updated after any changes to variables and outputs. See https://github.com/terraform-aws-modules/terraform-aws-eks/#doc-generation

[daroga0002]

I dont think does limiting terraform version is good approach. I would rather try to solve a root cause than limiting terraform updates, especially that 0.14 is considered now as desired and stable version

[aidanmelen]

@daroga0002 I agree, let's fix it because this problem isn't going away. But this module claims to support 0.14.0 when that is not entirely true. Perhaps we should add a note to the README that says:

If you are using a terraform version greater than 0.14.0, you will need to run terraform refresh prior to running terraform destroy.

I just thought something like this would save people the 30 minutes of head scratching when the terraform destroy fails.

[daroga0002]

@aidanmelen can you describe a issue more (where it is failing, does those are only some module inputs required)?

[aidanmelen]

@daroga0002 The error can be reproduced by running the basic example on terraform 0.14.X.

The specific resource that the module fails to delete is:

Error: Unauthorized

module.kubernetes_cluster.module.eks.kubernetes_config_map.aws_auth[0] will be destroyed

There is a lot more context in this related issue #1162 than I can provide here on my own. This seems to be causing confusion for a number of people.

[dak1n1]

Prior to Terraform version 0.14, data sources would be refreshed before destroy. This functionality has been added back to Terraform 0.15. It's an issue with Terraform core. Unless they decide to backport that fix to 0.14, it makes sense to limit the versions that are used with this module.

[dak1n1]

It looks like the fix will be too difficult to backport to 0.14. hashicorp/terraform#28179 (comment)

[antonbabenko]

We don't have to restrict the max version of the Terraform in the module.

Btw, v15.1.0 has been just released where Terraform 0.15 is supported and the minimum version of Terraform is updated to be 0.13.1.

[github-actions]

I'm going to lock this pull request 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 related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.