Configuring provider using another resource's outputs does not work on second run on Terraform v0.14.x

[mcanevet]

Terraform, Provider, Kubernetes and Helm Versions

Terraform version: 0.14.4
Provider version: 2.0.1
Kubernetes version: 1.18.13
Helm version: 3.4.2

Affected Resource(s)

• helm_release

Terraform Configuration Files

Provider configuration:

provider "helm" {
  kubernetes {
    insecure = true
    host     = local.kubernetes_host
    username = local.kubernetes_username
    password = local.kubernetes_password
  }
}

Expected Behavior

Subsequent terraform apply should work.

Actual Behavior

First apply completes:

Apply complete! Resources: 26 added, 0 changed, 0 destroyed.

But subsequent applies fails when refreshing the file helm_release resource:

...
module.cluster.module.argocd.helm_release.argocd: Refreshing state... [id=argocd]

Error: provider not configured: you must configure a path to your kubeconfig
or explicitly supply credentials via the provider block or environment variables.

See our authentication documentation at: https://registry.terraform.io/providers/hashicorp/helm/latest/docs#authentication

Important Factoids

It works well with Terraform v0.13.x.
Helm provider is configured using another resource's outputs, but somehow it tries to read kubeconfig.

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment

[redeux]

Thanks for opening this issue @mcanevet. Do you mind sharing your debug log as a gist?

[jrhouston]

@mcanevet Is this still an issue for you in v2.0.2 of the provider?

[mcanevet]

@redeux @jrhouston I still have the issue with Terraform 0.14.6 and provider 2.0.2. Note that I also had the issue with Terraform v0.14.x and provider v1.x, so it's probably related to a change in Terraform behavior.
See gist : https://gist.github.com/mcanevet/898d6093e06df6528ec3a8b7b7af02dd
1rst run is working fine, but 2nd run fails with Error: Kubernetes cluster unreachable: invalid configuration: no configuration has been provided, try setting KUBERNETES_MASTER environment variable

[gabeio]

So I had an extremely similar issue to this one, exactly the same 2 errors. After a ton of unnecessary debugging (had I realized the problem 🤦), I found I had a module which was out of place (which had a helm_release) and apparently was causing it to run first, prior to it's dependencies and in this case the module that provided the kubernetes variables. I had refactored a bit of my terraform code and forgot to move the helm release to the correct location. Again, not sure if this is the problem in this case but I thought I would share in hopes it would help others.

tip: you can look through your terraform state (terraform state pull) and see if you have a helm release with incorrect dependencies.

Hope this helps 🌮

[mcanevet]

OK, I found my issue. I was running terraform init -upgrade before the second run and as I was storing my kubeconfig file inside the .terraform directory, it was purged. It looks like terraform 0.13 does not purge the .terraform directory when running terraform init -upgrade.

[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. If you feel I made an error 🤖 🙉 , please reach out to my human friends 👉 [email protected]. Thanks!