terraform, azure and utoronto: fixes and aligning files with terraform state

docs/topic/infrastructure/cluster-design.md

@@ -122,9 +122,9 @@ The three machine types based on the cloud provider are the following:
   - r5.4xlarge
   - r5.16xlarge
 - [AKS](https://learn.microsoft.com/en-us/azure/virtual-machines/eav4-easv4-series)
-  - Standard_E4a_v4
-  - Standard_E16_v4
-  - Standard_E64_v4
+  - Standard_E4s_v5
+  - Standard_E16s_v5
+  - Standard_E64s_v5
 
 ## Network Policy
 

terraform/azure/main.tf

@@ -8,20 +8,23 @@ terraform {
       # FIXME: v3 has been released and we are still at v2, see release notes:
       #        https://github.com/hashicorp/terraform-provider-azurerm/releases/tag/v3.0.0
       #
+      #        We may need to remove old state and then then import it according to
+      #        https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs/guides/3.0-upgrade-guide#migrating-to-new--renamed-resources.
+      #
       source  = "hashicorp/azurerm"
       version = "~> 2.99"
     }
 
     azuread = {
       # ref: https://registry.terraform.io/providers/hashicorp/azuread/latest
       source  = "hashicorp/azuread"
-      version = "~> 2.35"
+      version = "~> 2.47"
     }
 
     kubernetes = {
       # ref: https://registry.terraform.io/providers/hashicorp/kubernetes/latest
       source  = "hashicorp/kubernetes"
-      version = "~> 2.18"
+      version = "~> 2.25"
     }
 
   }
@@ -92,9 +95,8 @@ resource "azurerm_kubernetes_cluster" "jupyterhub" {
 
   # Core node-pool
   default_node_pool {
-    name       = "core"
-    node_count = 1
     # Unfortunately, changing anything about VM type / size recreates *whole cluster
+    name                = "core"
     vm_size             = var.core_node_vm_size
     os_disk_size_gb     = 40
     enable_auto_scaling = true
@@ -197,7 +199,7 @@ resource "azurerm_container_registry" "container_registry" {
   name                = var.global_container_registry_name
   resource_group_name = azurerm_resource_group.jupyterhub.name
   location            = azurerm_resource_group.jupyterhub.location
-  sku                 = "premium"
+  sku                 = "Premium"
   admin_enabled       = true
 }
 

terraform/azure/projects/utoronto.tfvars

@@ -1,23 +1,28 @@
-tenant_id          = "78aac226-2f03-4b4d-9037-b46d56c55210"
-subscription_id    = "ead3521a-d994-4a44-a68d-b16e35642d5b"
-resourcegroup_name = "2i2c-utoronto-cluster"
-
-
-kubernetes_version = "1.26.3"
+tenant_id                      = "78aac226-2f03-4b4d-9037-b46d56c55210"
+subscription_id                = "ead3521a-d994-4a44-a68d-b16e35642d5b"
+resourcegroup_name             = "2i2c-utoronto-cluster"
+global_container_registry_name = "2i2cutorontohubregistry"
+global_storage_account_name    = "2i2cutorontohubstorage"
+location                       = "canadacentral"
 
 storage_size = 8192
+ssh_pub_key  = "ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDQJ4h39UYNi1wybxAH+jCFkNK2aqRcuhDkQSMx0Hak5xkbt3KnT3cOwAgUP1Vt/SjhltSTuxpOHxiAKCRnjwRk60SxKhUNzPHih2nkfYTmBBjmLfdepDPSke/E0VWvTDIEXz/L8vW8aI0QGPXnXyqzEDO9+U1buheBlxB0diFAD3vEp2SqBOw+z7UgrGxXPdP+2b3AV+X6sOtd6uSzpV8Qvdh+QAkd4r7h9JrkFvkrUzNFAGMjlTb0Lz7qAlo4ynjEwzVN2I1i7cVDKgsGz9ZG/8yZfXXx+INr9jYtYogNZ63ajKR/dfjNPovydhuz5zQvQyxpokJNsTqt1CiWEUNj georgiana@georgiana"
 
-ssh_pub_key = "ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDQJ4h39UYNi1wybxAH+jCFkNK2aqRcuhDkQSMx0Hak5xkbt3KnT3cOwAgUP1Vt/SjhltSTuxpOHxiAKCRnjwRk60SxKhUNzPHih2nkfYTmBBjmLfdepDPSke/E0VWvTDIEXz/L8vW8aI0QGPXnXyqzEDO9+U1buheBlxB0diFAD3vEp2SqBOw+z7UgrGxXPdP+2b3AV+X6sOtd6uSzpV8Qvdh+QAkd4r7h9JrkFvkrUzNFAGMjlTb0Lz7qAlo4ynjEwzVN2I1i7cVDKgsGz9ZG/8yZfXXx+INr9jYtYogNZ63ajKR/dfjNPovydhuz5zQvQyxpokJNsTqt1CiWEUNj georgiana@georgiana"
-
-global_container_registry_name = "2i2cutorontohubregistry"
-global_storage_account_name    = "2i2cutorontohubstorage"
+# FIXME: upgrade to 1.27.7, and then 1.28.3, based on the latest versions
+#        available via: az aks get-versions --location westus2 -o table
+#
+kubernetes_version = "1.26.3"
 
-location          = "canadacentral"
+# FIXME: upgrade core_node_vm_size to Standard_E4s_v5
 core_node_vm_size = "Standard_E4s_v3"
+
 notebook_nodes = {
   "default" : {
-    min : 1,
-    max : 100,
+    # NOTE: min-max below was set to 0-86 retroactively to align with
+    #       observed state without understanding on why 0-86 was picked.
+    min : 0,
+    max : 86,
+    # FIXME: upgrade user nodes vm_size to Standard_E8s_v5
     vm_size : "Standard_E8s_v3",
   }
 }

terraform/azure/proxycommand.py

@@ -1,4 +1,11 @@
 #!/usr/bin/env python3
+"""
+This script can be used to migrate Azure Files storage from one cluster to
+another.
+
+Learn more at https://infrastructure.2i2c.org/hub-deployment-guide/hubs/other-hub-ops/move-hubs/across-clusters/#azure-files.
+"""
+
 import subprocess
 import sys
 import time

terraform/azure/storage.tf

@@ -12,6 +12,12 @@ resource "azurerm_storage_account" "homes" {
 
   network_rules {
     # Allow NFS access only from our nodes, deny access from all other networks
+    #
+    # Use of terraform plan or apply can run into issues due to this, but they
+    # can be handled by temporarily adding your public IP to a firewall
+    # exception like described in
+    # https://github.com/2i2c-org/infrastructure/issues/890#issuecomment-1879072422.
+    #
     default_action = "Deny"
     virtual_network_subnet_ids = [
       azurerm_subnet.node_subnet.id
@@ -23,7 +29,7 @@ resource "azurerm_storage_share" "homes" {
   name                 = "homes"
   storage_account_name = azurerm_storage_account.homes.name
   quota                = var.storage_size
-  enabled_protocol     = var.storage_protocol
+  enabled_protocol     = "NFS"
   lifecycle {
     # Additional safeguard against deleting the share
     # as this causes irreversible data loss!
@@ -34,9 +40,3 @@ resource "azurerm_storage_share" "homes" {
 output "azure_fileshare_url" {
   value = azurerm_storage_share.homes.url
 }
-
-resource "kubernetes_namespace" "homes" {
-  metadata {
-    name = "azure-file"
-  }
-}