Terraform-provider-azurerm: "http_application_routing_zone_name" of resource azurerm_kubernetes_cluster is empty

[luc1f4]

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment

Terraform (and AzureRM Provider) Version

Terraform v0.12.25
AzureRM 2.23.0

Affected Resource(s)

• azurerm_kubernetes_cluster

Terraform Configuration Files

resource "azurerm_kubernetes_cluster" "this" {
  count   = var.common_dev_rc ? 0 : 1
  name                = local.component_names["name"]
  location            = var.region
  resource_group_name = azurerm_resource_group.this.0.name
  tags                = local.tags

  dns_prefix         = "dns-${local.component_names["name"]}"
  enable_pod_security_policy = false
  kubernetes_version = var.k8s_version

   default_node_pool {
    name                = "default"
    node_count      = var.vm_count
    vm_size             = var.vm_size
    enable_auto_scaling = var.enable_auto_scaling

    min_count = var.vm_count
    max_count = var.vm_maxcount

    os_disk_size_gb = 30
    vnet_subnet_id  = azurerm_subnet.this.0.id
    type                   = "VirtualMachineScaleSets"
  }

  addon_profile {
    http_application_routing {
      enabled = true
    }
  }

  network_profile {
    network_plugin = "azure"
    load_balancer_sku = "Basic"
  }

  service_principal {
    client_id     = var.aks_sp_client_id
    client_secret = var.aks_sp_client_secret
  }

  linux_profile {
    admin_username = "linux"

    ssh_key {
      key_data = var.aks_ssh_public_key
    }
  }

  role_based_access_control {
    enabled = false
  }

  lifecycle {
    ignore_changes = [
      windows_profile,
    ]
  }
}


output "dnszone" {
  value = azurerm_kubernetes_cluster.this.0.addon_profile[0].http_application_routing.0.http_application_routing_zone_name
}

Debug Output

Tried to solve the data ressource in "terraform console". The output is also empty

azurerm_kubernetes_cluster.this.0.addon_profile[0].http_application_routing.0.http_application_routing_zone_name
{
"enabled" = true
"http_application_routing_zone_name" = ""
}

Expected Behavior

The correct zone name should be printed out

Actual Behavior

http_application_routing_zone_name is empty. Yesterday it worked as usual. This issue started beginning today.

[tombuildsstuff]

hi @luc1f4

Thanks for opening this issue.

To be able to look into this we're going to need a Terraform Configuration showing how the AKS Cluster is provisioned - would you mind updating the description above to include that?

Thanks!

[luc1f4]

hi @tombuildsstuff

I have pasted the config of the aks resource itself. Hopefully that is understandble.

[wandrew26]

I assume, that the naming of the properties have changed on azure side. 😞
The microsoft documentation states that you can get the value with the query:

az aks show --resource-group myResourceGroup --name myAKSCluster --query addonProfiles.httpApplicationRouting.config.HTTPApplicationRoutingZoneName -o table

But it doesn't seem to be valid since a few days.
Callling: az aks show --resource-group xxxx --name xxxx --query addonProfiles.httpapplicationrouting will output:

{ "config": { "httpapplicationroutingzonename": "xxxxxxxxxx.yyyyyyyyy.aksapp.io" }, "enabled": true, "identity": null }

They using lowercase now instead of camel case. 😭 Which leads to a breaking change on tf side.

[luc1f4]

I created a Pull Request with the changes to catch this issue:
https://github.com/terraform-providers/terraform-provider-azurerm/pull/8387

Can somebody review this?
Thanks.

[robbiezhang]

Hi all, this is Robbie from the AKS team. Sorry about the breaking change. We're working on a hotfix to revert it back to the previous behavior - preserve the casing from the input. Please kindly stay tuned.

[wandrew26]

Microsoft deployed a hotfix, issue can be closed. @luc1f4

[robbiezhang]

The hotfix is rolled out to all the regions last week. For the new clusters, the casing of the addon name and addon config names are preserved. for the existing cluster, you need to re-enable the addon with the desired casing to fix the issue.