Terraform always replaces GKE autopilot even though there’s no code change

Terraform Providers Google
1

Terraform always replaces GKE autopilot even though there’s no change in the code.
Is this bug? Does anyone know how to avoid it?

Terraform Version

v1.0.8

Prerequisites

I made a gke cluster by calling terraform apply with the following code.

resource "google_container_cluster" "default" {
  name     = "test"
  location = "asia-northeast1"

  network    = {id of existing vpc}
  subnetwork = {id of existing subnetwork}

  enable_autopilot = true

  node_config {
    oauth_scopes = [
      "https://www.googleapis.com/auth/cloud-platform",
      "https://www.googleapis.com/auth/monitoring",
    ]
  }
}

Expected Behavior

When I exec terrafrom plan with the above same code, terraform shows there’s no change and terraform won’t do anything.

Actual Behavior

When I exec terrafrom plan with the above same code, terraform says it will replace (destroy & create) the gke cluster even though there’s no change.
If the code doens’t have node_config , it won’t happen.

Steps to Reproduce

  1. write the above code
  2. call terraform init
  3. call terraform apply
  4. call terraform apply again
2

Did you ever resolve this? Nearly 2 years later and I’m experiencing the same thing, even though I don’t have the node_config options set. My config is very simple and is as follows:

resource “google_container_cluster” “cluster” {
name = var.service_name
location = var.region
node_locations = var.node_locations

network = var.network
subnetwork = var.subnet

Enabling Autopilot for this cluster

enable_autopilot = true
}

The service_name is “cluster” the same as the cluster.

While it takes over 8 minutes to apply this simple config and create the cluster, if I do a “terraform plan” and “terraform apply” immediately afterward, without changing anything, it replaces the cluster by destroying it and creating it again.