initial_node_count value not getting updated from terraform

[RITESHBRID]

TL;DR

Whenever we try to update initial_node_count variable, terrafrom plan and apply did not show any changes

Expected behavior

initial_node_count must be updated through terraform

Observed behavior

No response

Terraform Configuration

module "gke" {
  source = "terraform-google-modules/kubernetes-engine/google//modules/beta-private-cluster-update-variant"

  version = "13.1.0"

  project_id                  = var.project_id
  kubernetes_version          = var.kubernetes_version
  name                        = "${var.environment}-${var.location_acronym}-${var.cluster_name_suffix}"
  region                      = var.location
  zones                       = var.zones
  network                     = var.network_name
  subnetwork                  = var.subnet_name
  ip_range_services           = "${var.subnet_name}"
  ip_range_pods               = "${var.subnet_name}"
  http_load_balancing         = var.http_load_balancing
  horizontal_pod_autoscaling  = var.horizontal_pod_autoscaling
  network_policy              = var.network_policy
  enable_private_endpoint     = var.enable_private_endpoint
  enable_private_nodes        = var.enable_private_nodes
  master_ipv4_cidr_block      = var.master_ipv4_cidr_block
  enable_intranode_visibility = var.enable_intranode_visibility
  master_authorized_networks = var.master_authorized_networks
  create_service_account     = var.create_service_account
  service_account            = "gke-sa-${var.project_id}.iam.gserviceaccount.com"
  remove_default_node_pool   = var.remove_default_node_pool
  default_max_pods_per_node  = var.default_max_pods_per_node
  maintenance_start_time     = var.maintenance_start_time
  maintenance_end_time       = var.maintenance_end_time
  notification_config_topic  = "projects/${var.project_id}/topics/${var.notification_config_topic}"
  maintenance_exclusions     =  var.maintenance_exclusions
  cluster_autoscaling        = var.cluster_autoscaling
  
  node_pools = [
    {
      name               = "${var.network_name}-test"
      machine_type       = var.machine_type
      enable_secure_boot = true
      node_locations     = var.node_pools_locations
      min_count          = var.min_node_count_default_node
      max_count          = var.max_node_count_default_node
      local_ssd_count    = 0
      disk_size_gb       = var.disk_size_gb_default_node
      disk_type          = var.disk_type_defnp01_node
      image_type         = var.image_type_default_node
      auto_repair        = var.auto_repair_default_node
      auto_upgrade       = var.auto_upgrade_default_node
      service_account    = "gke-sa-${var.project_id}.iam.gserviceaccount.com"
      preemptible        = var.preemptible_default_node
      initial_node_count = var.initial_node_count_default_node
    }
]

  node_pools_oauth_scopes = {
    all = []

    "${var.network_name}-test" = [
      "https://www.googleapis.com/auth/cloud-platform",
    ]
  }

  node_pools_labels = {
    all = {}


    "${var.network_name}-test" = {"${var.network_name}-test" = true }
  }

  node_pools_metadata = {
    all = {}

   "${var.network_name}-test" = {
      node-pool-metadata-custom-value = "${var.network_name}-test"
    }

  }

  node_pools_taints = {
    all = []

   "${var.network_name}-test" = [
      {
        key    = "${var.network_name}-test"
        value  = true
        effect = "PREFER_NO_SCHEDULE"
      }
    ]
  }

  node_pools_tags = {
    all = []


    "${var.network_name}-test" = [
      "${var.network_name}-test",
    ]

  }
}

Terraform Version

Terraform v1.1.0

Additional information

No response

[morgante]

initial_node_count is only used for the default pool (that we delete anyways) and is therefore intentionally ignored.

We should probably remove it from the module entirely, but that will require additional refactoring.

If you want to change the number of nodes in your cluster, you should adjust the size of your node pools. There is actually no reason to adjust the number at any time.

[RITESHBRID]

@morgante I my case, I have down scaled my gke cluster to size 0 through GCP console, Now I have tried to upscale the cluster through terraform, I have changed min_count to 5 but still cluster is not spine up new nodes to match min node count. However GCP console show the Min count has been updated in autoscale

[morgante]

Please focus on modifying the node pool configuration.

If that is somehow not working for you, it has nothing to do with Terraform and you should open a support case with Google.