gcfs_config variable forces replacement

[dominykasn]

TL;DR

gcfs_config variable forces replacement to node pools when importing

Expected behavior

Importing without issues when gcfs is disabled in GUI and in gcloud CLI

Observed behavior

When importing plan shows forced replacement to node pool, image streaming is disabled in GCP GUI and in gcloud CLI

- gcfs_config { # forces replacement
              - enabled = false -> null
            }

Terraform Configuration

module "gke" {
  source  = "terraform-google-modules/kubernetes-engine/google//modules/private-cluster"
  version = "~> 32.0.0"

  ....
  node_pools = [
    {
      name = "n2d-c8-m32"
      image_type = "COS_CONTAINERD"
      enable_secure_boot = true
      enable_gcfs = false
    },

Terraform Version

Terraform v1.8.4

Additional information

No response

[AlvaroGG0]

Working on it already: #1976

[dominykasn]

Nice, I also have a pull request open for similar issue GoogleCloudPlatform/magic-modules#11553

[dominykasn]

@apeabody can I open a new pull request in this repository to change this ? From what I can see the PR is being only rebased 🤔

[apeabody]

This should be fixed by https://github.com/hashicorp/terraform-provider-google/releases/tag/v6.2.0

[wyardley]

I saw the fix with 6.2.0 supporting updating in place (will test with 6.2.0 shortly), but the thing that's odd to me is that I already explicitly had enable_gcfs = false in my config with version 32 of the module, and I seem to get a planned diff force replacement by all of removing that (allowing default), setting it to true, or setting it to false.

[apeabody]

I saw the fix with 6.2.0 supporting updating in place (will test with 6.2.0 shortly), but the thing that's odd to me is that I already explicitly had enable_gcfs = false in my config with version 32 of the module, and I seem to get a planned diff force replacement by all of removing that (allowing default), setting it to true, or setting it to false.

Hi @wyardley - I believe that particular issue with enable_gcfs = false should be resolved by #1976 in v33.

[wyardley]

@apeabody thanks!

fwiw, I was seeing this with 33.0.1. I'm having some issues with OpenTofu registry not having 6.2.0, but I'll try again once that's available.

% tf version
OpenTofu v1.8.2
on darwin_arm64
+ provider registry.opentofu.org/hashicorp/external v2.3.4
+ provider registry.opentofu.org/hashicorp/google v6.1.0
+ provider registry.opentofu.org/hashicorp/kubernetes v2.32.0
+ provider registry.opentofu.org/hashicorp/null v3.2.2
+ provider registry.opentofu.org/hashicorp/random v3.6.2

and then

  source                = "terraform-google-modules/kubernetes-engine/google//modules/private-cluster"
  version               = "33.0.1"

[wyardley]

@apeabody thanks for the additional quick fix 🙏