Unnecessary ForceNew On Container Node Pool / Cluster gcfs_config

[tgoodsell-tempus]

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.
• If an issue is assigned to a user, that user is claiming responsibility for the issue.
• Customers working with a Google Technical Account Manager or Customer Engineer can ask them to reach out internally to expedite investigation and resolution of this issue.

Terraform Version & Provider Version(s)

Terraform v1.5.7
on darwin_arm64

• provider registry.terraform.io/hashicorp/google v5.33.0
• provider registry.terraform.io/hashicorp/google-beta v5.33.0

Affected Resource(s)

• google_container_cluster
• google_container_node_pool

Terraform Configuration

resource "google_container_node_pool" "main" {

  project     = "project"
  cluster     = "name"
  location    = "location"
  name_prefix = "${var.name}-"

  timeouts {
    update = "60m"
  }

  autoscaling {
    total_min_node_count = 0
    total_max_node_count = var.total_max_node_count
  }

  upgrade_settings {
    max_surge       = min(ceil(var.total_max_node_count / 2), 20)
    max_unavailable = 0
    strategy        = "SURGE"

  }

  node_config {
    disk_size_gb = var.disk_size_gb
    disk_type    = var.disk_type

    gcfs_config {
      enabled = false -> true (or vice versa)
    }
}

Debug Output

No response

Expected Behavior

The documentation for disabling the image streaming / gcfs_config option states that this can be performed in place on a standard cluster or specific node pool.

See: https://cloud.google.com/kubernetes-engine/docs/how-to/image-streaming#disable

This can be confirmed by running their gcloud command manually.

Actual Behavior

The Terraform provider considers this change as requiring the full destroy and re-creation of the resource.

          ~ gcfs_config {
              ~ enabled = true -> false # forces replacement
            }

Steps to reproduce

1. Create node pool with the gcfs_config set enabled true or false and apply
2. Change that config and apply

Important Factoids

No response

References

• https://github.com/hashicorp/terraform-provider-google/blob/main/google/services/container/node_config.go suggests a significant number of these config objects are treated as ForceNew, so I may be missing some context on why the Terraform provider leans that direction instead of following the GCP API more exactly.
• This field is included in the options for the update function for things like NodePool: https://cloud.google.com/kubernetes-engine/docs/reference/rest/v1/projects.locations.clusters.nodePools/update

b/347091584

[ggtisc]

This could be an enhancement changing the forces replacement for an update-in-place

[wyardley]

Looks like this can be closed now, though I believe updates in-place for some or all cases may not actually work properly until #19225 is resolved.