Another approach to fixing the thing

Author: nullchimp

.github/workflows/deploy-azure.yml

@@ -100,6 +100,39 @@ jobs:
           echo "🆕 AKS cluster doesn't exist, will be created by Terraform"
         fi
 
+    ################################################################
+    # 2.5 · Reconcile AKS cluster if it exists
+    ################################################################
+    - name: Reconcile AKS cluster
+      run: |
+        set -euo pipefail
+        RG="rg-aks-memgraph-${{ env.ENVIRONMENT }}"
+        AKS="aks-memgraph-${{ env.ENVIRONMENT }}"
+        
+        if az aks show --resource-group "$RG" --name "$AKS" &>/dev/null; then
+          echo "🔄 Reconciling AKS cluster to fix node resource group..."
+          az aks update --resource-group "$RG" --name "$AKS" --no-wait
+          
+          echo "⏳ Waiting for reconciliation to complete..."
+          max_retries=10
+          retry_count=0
+          while [ $retry_count -lt $max_retries ]; do
+            if az aks show --resource-group "$RG" --name "$AKS" --query "provisioningState" -o tsv | grep -q "Succeeded"; then
+              echo "✅ AKS reconciliation completed successfully"
+              break
+            fi
+            echo "⏳ AKS still reconciling, waiting 30s... ($(($retry_count + 1))/$max_retries)"
+            sleep 30
+            retry_count=$((retry_count + 1))
+          done
+          
+          if [ $retry_count -eq $max_retries ]; then
+            echo "⚠️ AKS reconciliation timed out but continuing with deployment"
+          fi
+        else
+          echo "⏭️ No existing AKS cluster to reconcile"
+        fi
+
     ################################################################
     # 3 · Terraform apply
     ################################################################

infra/azure/main.tf

@@ -51,6 +51,11 @@ resource "azurerm_kubernetes_cluster" "this" {
     name       = "default"
     node_count = 1
     vm_size    = var.node_vm_size
+
+    # Explicitly define upgrade settings to match Azure defaults
+    upgrade_settings {
+      max_surge = "33%"
+    }
   }
 
   identity { type = "SystemAssigned" }