Invalid plan when re-applying terraform module

[ananthrav]

When creating a gke cluster using the beta-private-cluster module, I'm running into issues when I apply the configuration again. It works fine for the first apply, subsequent plans and applies fail for the same code.

Terraform version

Terraform v0.12.8
+ provider.google v2.14.0
+ provider.google-beta v2.14.0
+ provider.kubernetes v1.9.0
+ provider.null v2.1.2
+ provider.random v2.2.0

Getting the following error

------------------------------------------------------------------------

Error: Provider produced invalid plan

Provider "google-beta" has indicated "requires replacement" on
module.gke.google_container_cluster.primary for a non-existent attribute path
cty.Path{cty.GetAttrStep{Name:"private_cluster_config"},
cty.IndexStep{Key:cty.NumberIntVal(0)},
cty.GetAttrStep{Name:"master_ipv4_cidr_block"}}.

The module looks something like this

module "gke" {

  source = "git::https://github.com/terraform-google-modules/terraform-google-kubernetes-engine.git//modules/beta-private-cluster?ref=master"

  name                       = local.kubernetes_cluster_name
  project_id                 = "${var.project_name}"
  region                     = "${var.region}"
  network                    = "${data.terraform_remote_state.network.outputs.network_name}"
  subnetwork                 = data.terraform_remote_state.network.outputs.subnets_names[0]
  ip_range_pods              = "${data.terraform_remote_state.network.outputs.subnets_names[0]}-1"
  ip_range_services          = "${data.terraform_remote_state.network.outputs.subnets_names[0]}-2"
  http_load_balancing        = "${var.http_load_balancing}"
  horizontal_pod_autoscaling = "${var.horizontal_pod_autoscaling}"
  kubernetes_dashboard       = "${var.kubernetes_dashboard}"
  network_policy             = "${var.network_policy}"
  monitoring_service         = "${var.kubernetes_monitoring_service}"
  logging_service            = "${var.kubernetes_logging_service}"
  create_service_account     = false
  service_account            = "${var.service_account}"
  enable_private_endpoint    = "${var.enable_private_endpoint}"
  enable_private_nodes       = "${var.enable_private_nodes}"
  remove_default_node_pool   = "${var.remove_default_node_pool}"

  master_authorized_networks_config = [
    {
      cidr_blocks = [
        {
          cidr_block   = "xx.xx.xx.xx/32"
          display_name = "Whitelist"
        },
      ]
    },
  ]

  node_pools = [
    {
      name               = "default-node-pool"
      machine_type       = "${var.kubernetes_machine_type}"
      disk_size_gb       = "${var.kubernetes_node_pool_disk_size}"
      disk_type          = "${var.kubernetes_node_pool_disk_type}"
      image_type         = "${var.kubernetes_node_pool_image_type}"
      auto_repair        = true
      auto_upgrade       = true
      service_account    = "${var.service_account}"
      preemptible        = false
      initial_node_count = "${var.initial_node_count}"
    },
  ]

  node_pools_oauth_scopes = {
    all = []

    default-node-pool = [
      "https://www.googleapis.com/auth/cloud-platform",
    ]
  }
}

Using master because a PR that was causing issues for me was merged and a new release hasn't been tagged. Please let me know if I can provide more info

[cilindrox]

Happened to me twice with both the beta and standard private cluster modules. Think the issue's related to the deletion of the default cluster node pool

[nick4fake]

Related PRs:
hashicorp/terraform-provider-google#4577
hashicorp/terraform-provider-google-beta#1211

[nick4fake]

Looks like it works in 2.17.0

[morgante]

@ananthrav Can you upgrade to the v2.17.0 provider and confirm it fixed this?