Merge pull request #327 from pfeifferj/feat/improved-multi-zone

feat(vpc): imrpoved multi zone

Author: pfeifferj

.github/workflows/e2e-tests-in-cluster.yaml

@@ -177,11 +177,14 @@ jobs:
         # Image selector tests from image_selector_test.go
         image_selector_tests="TestE2EImageSelector"
 
+        # Multi-zone tests from multizone_test.go
+        multizone_tests="TestE2EMultiZoneDistribution TestE2EZoneAntiAffinity TestE2ETopologySpreadConstraints TestE2EPlacementStrategyValidation TestE2EZoneFailover"
+
         # Cleanup tests from cleanup_test.go
         cleanup_tests="TestE2ECleanupNodePoolDeletion TestE2ECleanupNodeClassDeletion TestE2ECleanupOrphanedResources TestE2ECleanupIBMCloudResources"
 
         # Combine all tests
-        all_tests="$core_tests $validation_tests $block_device_tests $scheduling_tests $userdata_tests $image_selector_tests $cleanup_tests"
+        all_tests="$core_tests $validation_tests $block_device_tests $scheduling_tests $userdata_tests $image_selector_tests $multizone_tests $cleanup_tests"
 
         test_failed="false"
         passed_tests=0
@@ -195,6 +198,7 @@ jobs:
         echo "  Scheduling Tests: $(echo $scheduling_tests | wc -w)"
         echo "  UserData Tests: $(echo $userdata_tests | wc -w)"
         echo "  Image Selector Tests: $(echo $image_selector_tests | wc -w)"
+        echo "  Multi-Zone Tests: $(echo $multizone_tests | wc -w)"
         echo "  Cleanup Tests: $(echo $cleanup_tests | wc -w)"
         echo "  Total Tests: $total_tests"
         echo ""
@@ -212,6 +216,9 @@ jobs:
             "TestE2EDriftStability")
               timeout="30m"  # Drift test needs more time for monitoring
               ;;
+            "TestE2EMultiZone"*|"TestE2EZone"*|"TestE2ETopology"*|"TestE2EPlacementStrategy"*)
+              timeout="25m"  # Multi-zone tests need extra time for cross-zone provisioning
+              ;;
             "TestE2ECleanup"*)
               timeout="15m"  # Cleanup tests are typically faster
               ;;

.github/workflows/e2e-tests-pr.yml

@@ -336,11 +336,14 @@ jobs:
           # Image selector tests from image_selector_test.go
           image_selector_tests="TestE2EImageSelector"
 
+          # Multi-zone tests from multizone_test.go
+          multizone_tests="TestE2EMultiZoneDistribution TestE2EZoneAntiAffinity TestE2ETopologySpreadConstraints TestE2EPlacementStrategyValidation TestE2EZoneFailover"
+
           # Cleanup tests from cleanup_test.go
           cleanup_tests="TestE2ECleanupNodePoolDeletion TestE2ECleanupNodeClassDeletion TestE2ECleanupOrphanedResources TestE2ECleanupIBMCloudResources"
 
           # Combine all tests
-          all_tests="$core_tests $validation_tests $block_device_tests $scheduling_tests $userdata_tests $image_selector_tests $cleanup_tests"
+          all_tests="$core_tests $validation_tests $block_device_tests $scheduling_tests $userdata_tests $image_selector_tests $multizone_tests $cleanup_tests"
 
           test_failed="false"
           passed_tests=0
@@ -354,6 +357,7 @@ jobs:
           echo "  Scheduling Tests: $(echo $scheduling_tests | wc -w)"
           echo "  UserData Tests: $(echo $userdata_tests | wc -w)"
           echo "  Image Selector Tests: $(echo $image_selector_tests | wc -w)"
+          echo "  Multi-Zone Tests: $(echo $multizone_tests | wc -w)"
           echo "  Cleanup Tests: $(echo $cleanup_tests | wc -w)"
           echo "  Total Tests: $total_tests"
           echo ""
@@ -370,6 +374,9 @@ jobs:
               "TestE2EDriftStability")
                 timeout="30m"  # Drift test needs more time for monitoring
                 ;;
+              "TestE2EMultiZone"*|"TestE2EZone"*|"TestE2ETopology"*|"TestE2EPlacementStrategy"*)
+                timeout="25m"  # Multi-zone tests need extra time for cross-zone provisioning
+                ;;
               "TestE2ECleanup"*)
                 timeout="15m"  # Cleanup tests are typically faster
                 ;;

charts/crds/karpenter.sh_nodeoverlays.yaml

@@ -97,6 +97,10 @@ spec:
               value: {{ .Values.controller.metrics.port | quote }}
             - name: HEALTH_PROBE_PORT
               value: {{ .Values.controller.healthProbe.port | quote }}
+            {{- if .Values.controller.orphanCleanup.enabled }}
+            - name: KARPENTER_ENABLE_ORPHAN_CLEANUP
+              value: "true"
+            {{- end }}
             {{- if .Values.customResources.enabled }}
             {{- include "karpenter.cr.environmentConfig" . | nindent 12 }}
             {{- include "karpenter.cr.bootstrapConfig" . | nindent 12 }}

charts/templates/deployment.yaml

@@ -453,6 +453,12 @@ additionalCAs:
 
 # Controller configuration
 controller:
+  # Orphan cleanup configuration
+  orphanCleanup:
+    # Enable cleanup of orphaned instances (instances without NodeClaims)
+    # This prevents resource leaks when NodeClaims are deleted before instances are cleaned up
+    enabled: true
+
   metrics:
     port: 8080
   healthProbe:

charts/values.yaml

@@ -93,6 +93,7 @@ func main() {
 			op.UnavailableOfferings,
 			cloudProvider,
 			op.ProviderFactory.GetInstanceTypeProvider(),
+			op.ProviderFactory.GetSubnetProvider(),
 			op.ProviderFactory.GetClient(),
 		)...).
 		Start(ctx)

cmd/controller/main.go

@@ -85,6 +85,7 @@ ibmcloud is images --visibility public --status available | grep ubuntu
 - **API Server Endpoint**: Get your cluster's API endpoint (e.g., `https://10.240.0.1:6443`)
 - **Region**: `us-south` (or your preferred region)
 - **Zone**: `us-south-1` (subnet's availability zone)
+  - 💡 **Multi-Zone Tip**: For production deployments, consider using `placementStrategy` instead of explicit zones for automatic multi-zone distribution. See [Multi-Zone VPC Setup with Placement Constraints](vpc-integration.md#multi-zone-vpc-setup-with-placement-constraints)
 
 ## Installation