This message was deleted.

Hi @calm-oxygen-85457, this looks like it might be a bug with that specific Azure resources. Do you have a snippet of the Terraform code you’re using to create that resource? That would help us debug it.

Sure, below the main.tf :

resource "azurerm_kubernetes_cluster" "k8s" {

_name_                 = data.azurerm_resource_group.aks_rg.name

_location_             = data.azurerm_resource_group.aks_rg.location

_resource_group_name_  = data.azurerm_resource_group.aks_rg.name

_dns_prefix_           = data.azurerm_resource_group.aks_rg.name

_kubernetes_version_   = data.azurerm_kubernetes_service_versions.current.latest_version

_node_resource_group_  = format("%s-%s", data.azurerm_resource_group.aks_rg.name, "node")

_azure_policy_enabled_ = var.enable_azure_policy

_disk_encryption_set_id_ = local.os_disk_encryption_set_id

network_profile {

_network_plugin_ = "azure"

_network_policy_ = "azure"

} linux_profile {

_admin_username_ = "ubuntu"

ssh_key {

_key_data_ = data.azurerm_key_vault_secret.client_instance_name_cluster_key_public.value

} } auto_scaler_profile {

_balance_similar_node_groups_      = false

_max_graceful_termination_sec_     = "600"

_scale_down_delay_after_add_       = var.aks_scale_down_delay_after_add

_scale_down_delay_after_delete_    = "10s"

_scale_down_delay_after_failure_   = "3m"

_scale_down_unneeded_              = "10m"

_scale_down_unready_               = "20m"

_scale_down_utilization_threshold_ = var.aks_scale_down_utilization_threshold

_scan_interval_                    = "10s"

_skip_nodes_with_local_storage_    = false

_skip_nodes_with_system_pods_      = false

} default_node_pool {

_name_                 = var.node_pool_name

_vm_size_              = var.vm_size

_orchestrator_version_ = data.azurerm_kubernetes_service_versions.current.latest_version

_os_disk_type_         = var.os_disk_type

_os_disk_size_gb_      = var.os_disk_type == "Ephemeral" && var.os_disk_size_gb == 30 ? null : var.os_disk_size_gb

# The type of Default Node Pool for the Kubernetes Cluster must be VirtualMachineScaleSets to attach multiple node pools.

_type_                = "VirtualMachineScaleSets"

_enable_auto_scaling_ = var.enable_auto_scaling

_availability_zones_  = var.availability_zones

# Allow smooth autoscaling enabling from node count setting and agent_count default value

_node_count_     = var.enable_auto_scaling ? null : var.agent_count

_min_count_      = var.min_count

_max_count_      = var.max_count

_vnet_subnet_id_ = data.azurerm_subnet.aks_vnet_subnet.id

_max_pods_       = var.max_pods

_node_labels_ = {

nodetype = "default" }

_tags_ = local.aggregated_tags

dynamic "linux_os_config" { # If override_vm_max_count is true, inject linux_os_config configuration otherwise nothing is injected

_for_each_ = var.override_vm_max_count ? [""] : []

content { sysctl_config {

_vm_max_map_count_ = var.vm_max_map_count

} } } } lifecycle {

_ignore_changes_ = [role_based_access_control, windows_profile, default_node_pool[0].node_labels, default_node_pool[0].max_pods, default_node_pool[0].tags, default_node_pool[0].orchestrator_version, kubernetes_version]

} identity {

_type_ = "SystemAssigned"

} role_based_access_control {

_enabled_ = true

azure_active_directory {

_managed_                = true

_admin_group_object_ids_ = var.admin_group_object_ids

} }

_tags_ = local.aggregated_tags

}

Thanks, I’ll see if I can replicate.

I have the same behavior with another smaller module : main.tf

locals {

_aggregated_tags_ = merge(var.tags, tomap({ "InstanceName" : format("k8saas-%s", var.client_instance_name) }))

} resource "azurerm_log_analytics_workspace" "dedicated" {

_name_                = format("k8saas-%s", var.client_instance_name)

_location_            = data.azurerm_resource_group.aks_rg.location

_resource_group_name_ = data.azurerm_resource_group.aks_rg.name

_sku_                 = var.sku

_retention_in_days_   = var.retention_in_days

_tags_                = local.aggregated_tags

_daily_quota_gb_      = var.daily_quota_gb

} variables.tf

variable "client_instance_name" {

_type_        = string

_description_ = "Name of the cluster"

} variable "retention_in_days" {

_type_        = number

_default_     = 30

_description_ = "The workspace data retention in days. Possible values are either 7 (Free Tier only) or range between 30 and 730."

} variable "sku" {

_type_        = string

_default_     = "PerGB2018"

_description_ = "Optional) Specifies the Sku of the Log Analytics Workspace. Possible values are Free, PerNode, Premium, Standard, Standalone, Unlimited, and PerGB2018 (new Sku as of 2018-04-03)."

} variable "tags" {

_type_        = map(any)

_default_     = {}

_description_ = "tags following k8saas convention"

} variable "daily_quota_gb" {

_type_        = string

_default_     = null

_description_ = "The workspace daily quota for ingestion in GB. Defaults to null (unlimited) if omitted."

} data.tf

data "azurerm_resource_group" "aks_rg" {

_name_ = format("k8saas-%s", var.client_instance_name)

}

Can you run with

--log-level=debug

and see if that gives any clues?

Hi @little-author-61621, Yes, we use terragrunt in addition to terraform and some variables are in hcl file. It works If I change the location by eastus, thanks !

Great, yeah our HCL parsing method won’t be able to pull the data to get

data.azurerm_resource_group.aks_rg.location

, so I wonder how we can find this otherwise. Maybe we could allow a configuration to set the default region. The other option is to generate a Terraform plan JSON and run that through infracost, since that will pull the correct region.