Creating GKE cluster without initial_node_count

[pbzdyl]

Terraform Version

Terraform v0.11.3

• provider.google v1.6.0

Affected Resource

• google_container_cluster

Terraform Configuration Files

variable "test_cluster_name" { default = "test2" }
variable "test_cluster_machine_type" { default = "n1-standard-1" }
variable "test_cluster_disk_size_gb" { default = 10 }
variable "test_cluster_min_nodes" { default = 1 }
variable "test_cluster_max_nodes" { default = 16 }
variable "test_cluster_auto_repair" { default = true }
variable "test_cluster_dashboard_disabled" { default = true }
variable "test_cluster_master_version" { default = "1.9.3-gke.0" }
variable "test_cluster_node_version" { default = "1.9.3-gke.0" }
variable "test_cluster_oauth_scopes" {
  type = "list"
  default = [
    "https://www.googleapis.com/auth/devstorage.read_only",
    "https://www.googleapis.com/auth/logging.write",
    "https://www.googleapis.com/auth/monitoring",
    "https://www.googleapis.com/auth/servicecontrol",
    "https://www.googleapis.com/auth/service.management.readonly",
    "https://www.googleapis.com/auth/trace.append"
  ]
}

resource "random_id" "test_gke_username" {
  byte_length = "${var.password_length}"
}

resource "random_id" "test_gke_password" {
  byte_length = "${var.password_length}"
}

resource "google_container_cluster" "test" {
  name               = "${var.test_cluster_name}"
  zone               = "${var.zone}"
  min_master_version = "${var.test_cluster_master_version}"
  node_version       = "${var.test_cluster_node_version}"

  master_auth {
    username = "${random_id.test_gke_username.b64}"
    password = "${random_id.test_gke_username.b64}"
  }

  addons_config {
    kubernetes_dashboard {
      disabled = "${var.test_cluster_dashboard_disabled}"
    }
  }

  node_pool = [
    {
      name = "default-pool"
      node_config {
        machine_type       = "${var.test_cluster_machine_type}"
        disk_size_gb       = "${var.test_cluster_disk_size_gb}"
        oauth_scopes = "${var.test_cluster_oauth_scopes}"
      }
      autoscaling {
        min_node_count = "${var.test_cluster_min_nodes}"
        max_node_count = "${var.test_cluster_max_nodes}"
      }
      management {
        auto_repair = "${var.test_cluster_auto_repair}"
      }
    }
  ]
}

Expected Behavior

With the configuration above - notice that it doesn't specify initial_node_count in the embedded node_pool - a new cluster with a node pool of size 1 is created.

Actual Behavior

A node pool with size 0 is created:

The cluster is not auto-resized above size 0 even if non kube-system pods are deployed:

Steps to Reproduce

terraform apply the config included above

Workaround

I can fix the issue by specifying initial_node_count = 1 explicitly.
But according to google_container_cluster docs initial_node_count is deprecated and node_count should be used instead. Unfortunately, I cannot use node_count with autoscaling configuration as node_count always sets the cluster size to node_count value when the actual size has been modified by autoscaler - thus usage of autoscaling and node_count is mutually exclusive.

I am concerned that support for initial_node_count will be dropped and it won't be possible to workaround the issue I have (cluster created with size 0) just by specifying initial_node_count.

[pbzdyl]

According to GKE API reference initialNodeCount is mandatory thus I think it shouldn't be deprecated in terraform-provider-google. It's value has to be within auto-scaling range.

If it's not present in tf configuration and auto-scaling is used then it could be defaulted to min_node_count.

[wmuizelaar]

I guess this is related to #844

[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 👉 hashibot-feedback@hashicorp.com. Thanks!