Merge pull request #4476 from shraddhabang/agaautodiscovery

[feat aga] Implement auto-discovery feature for supported endpoints

Author: k8s-ci-robot

controllers/aga/globalaccelerator_controller.go

@@ -107,6 +107,7 @@ func NewGlobalAcceleratorReconciler(k8sClient client.Client, eventRecorder recor
 		config.ExternalManagedTags,
 		logger.WithName("aga-model-builder"),
 		metricsCollector,
+		cloud.ELBV2(),
 	)
 
 	// Create stack marshaller

controllers/ingress/group_controller.go

@@ -153,7 +153,7 @@ func (r *groupReconciler) reconcile(ctx context.Context, req reconcile.Request)
 		return ctrlerrors.NewErrorWithMetrics(controllerName, "add_group_finalizer_error", err, r.metricsCollector)
 	}
 
-	_, lb, frontendNlb, err := r.buildAndDeployModel(ctx, ingGroup)
+	_, lb, frontendNlb, listenerPorts, err := r.buildAndDeployModel(ctx, ingGroup)
 	if err != nil {
 		return err
 	}
@@ -173,7 +173,7 @@ func (r *groupReconciler) reconcile(ctx context.Context, req reconcile.Request)
 					return
 				}
 			}
-			statusErr = r.updateIngressGroupStatus(ctx, ingGroup, lbDNS, frontendNlbDNS)
+			statusErr = r.updateIngressGroupStatus(ctx, ingGroup, lbDNS, frontendNlbDNS, listenerPorts)
 			if statusErr != nil {
 				r.recordIngressGroupEvent(ctx, ingGroup, corev1.EventTypeWarning, k8s.IngressEventReasonFailedUpdateStatus,
 					fmt.Sprintf("Failed update status due to %v", statusErr))
@@ -200,25 +200,26 @@ func (r *groupReconciler) reconcile(ctx context.Context, req reconcile.Request)
 	return nil
 }
 
-func (r *groupReconciler) buildAndDeployModel(ctx context.Context, ingGroup ingress.Group) (core.Stack, *elbv2model.LoadBalancer, *elbv2model.LoadBalancer, error) {
+func (r *groupReconciler) buildAndDeployModel(ctx context.Context, ingGroup ingress.Group) (core.Stack, *elbv2model.LoadBalancer, *elbv2model.LoadBalancer, []int32, error) {
 	var stack core.Stack
 	var lb *elbv2model.LoadBalancer
 	var secrets []types.NamespacedName
 	var backendSGRequired bool
 	var err error
 	var frontendNlb *elbv2model.LoadBalancer
+	var listenerPorts []int32
 	buildModelFn := func() {
-		stack, lb, secrets, backendSGRequired, frontendNlb, err = r.modelBuilder.Build(ctx, ingGroup, r.metricsCollector)
+		stack, lb, secrets, backendSGRequired, frontendNlb, listenerPorts, err = r.modelBuilder.Build(ctx, ingGroup, r.metricsCollector)
 	}
 	r.metricsCollector.ObserveControllerReconcileLatency(controllerName, "build_model", buildModelFn)
 	if err != nil {
 		r.recordIngressGroupEvent(ctx, ingGroup, corev1.EventTypeWarning, k8s.IngressEventReasonFailedBuildModel, fmt.Sprintf("Failed build model due to %v", err))
-		return nil, nil, nil, ctrlerrors.NewErrorWithMetrics(controllerName, "build_model_error", err, r.metricsCollector)
+		return nil, nil, nil, nil, ctrlerrors.NewErrorWithMetrics(controllerName, "build_model_error", err, r.metricsCollector)
 	}
 	stackJSON, err := r.stackMarshaller.Marshal(stack)
 	if err != nil {
 		r.recordIngressGroupEvent(ctx, ingGroup, corev1.EventTypeWarning, k8s.IngressEventReasonFailedBuildModel, fmt.Sprintf("Failed build model due to %v", err))
-		return nil, nil, nil, err
+		return nil, nil, nil, nil, err
 	}
 	r.logger.Info("successfully built model", "model", stackJSON)
 
@@ -229,10 +230,10 @@ func (r *groupReconciler) buildAndDeployModel(ctx context.Context, ingGroup ingr
 	if err != nil {
 		var requeueNeededAfter *ctrlerrors.RequeueNeededAfter
 		if errors.As(err, &requeueNeededAfter) {
-			return nil, nil, nil, err
+			return nil, nil, nil, nil, err
 		}
 		r.recordIngressGroupEvent(ctx, ingGroup, corev1.EventTypeWarning, k8s.IngressEventReasonFailedDeployModel, fmt.Sprintf("Failed deploy model due to %v", err))
-		return nil, nil, nil, ctrlerrors.NewErrorWithMetrics(controllerName, "deploy_model_error", err, r.metricsCollector)
+		return nil, nil, nil, nil, ctrlerrors.NewErrorWithMetrics(controllerName, "deploy_model_error", err, r.metricsCollector)
 	}
 	r.logger.Info("successfully deployed model", "ingressGroup", ingGroup.ID)
 	r.secretsManager.MonitorSecrets(ingGroup.ID.String(), secrets)
@@ -242,9 +243,9 @@ func (r *groupReconciler) buildAndDeployModel(ctx context.Context, ingGroup ingr
 		inactiveResources = append(inactiveResources, k8s.ToSliceOfNamespacedNames(ingGroup.Members)...)
 	}
 	if err := r.backendSGProvider.Release(ctx, networkingpkg.ResourceTypeIngress, inactiveResources); err != nil {
-		return nil, nil, nil, ctrlerrors.NewErrorWithMetrics(controllerName, "release_auto_generated_backend_sg_error", err, r.metricsCollector)
+		return nil, nil, nil, nil, ctrlerrors.NewErrorWithMetrics(controllerName, "release_auto_generated_backend_sg_error", err, r.metricsCollector)
 	}
-	return stack, lb, frontendNlb, nil
+	return stack, lb, frontendNlb, listenerPorts, nil
 }
 
 func (r *groupReconciler) recordIngressGroupEvent(_ context.Context, ingGroup ingress.Group, eventType string, reason string, message string) {
@@ -253,25 +254,35 @@ func (r *groupReconciler) recordIngressGroupEvent(_ context.Context, ingGroup in
 	}
 }
 
-func (r *groupReconciler) updateIngressGroupStatus(ctx context.Context, ingGroup ingress.Group, lbDNS string, frontendNLBDNS string) error {
+func (r *groupReconciler) updateIngressGroupStatus(ctx context.Context, ingGroup ingress.Group, lbDNS string, frontendNLBDNS string, listenerPorts []int32) error {
 	for _, member := range ingGroup.Members {
-		if err := r.updateIngressStatus(ctx, lbDNS, frontendNLBDNS, member.Ing); err != nil {
+		if err := r.updateIngressStatus(ctx, lbDNS, frontendNLBDNS, member.Ing, listenerPorts); err != nil {
 			return err
 		}
 	}
 	return nil
 }
 
-func (r *groupReconciler) updateIngressStatus(ctx context.Context, lbDNS string, frontendNlbDNS string, ing *networking.Ingress) error {
+func (r *groupReconciler) updateIngressStatus(ctx context.Context, lbDNS string, frontendNlbDNS string, ing *networking.Ingress, ports []int32) error {
+	ingressPorts := make([]networking.IngressPortStatus, len(ports))
+	for i, port := range ports {
+		ingressPorts[i] = networking.IngressPortStatus{
+			Port: port,
+		}
+	}
+
 	ingOld := ing.DeepCopy()
 	if len(ing.Status.LoadBalancer.Ingress) != 1 ||
 		ing.Status.LoadBalancer.Ingress[0].IP != "" ||
 		ing.Status.LoadBalancer.Ingress[0].Hostname != lbDNS {
 		ing.Status.LoadBalancer.Ingress = []networking.IngressLoadBalancerIngress{
 			{
 				Hostname: lbDNS,
+				Ports:    ingressPorts,
 			},
 		}
+	} else if len(ports) > 0 {
+		ing.Status.LoadBalancer.Ingress[0].Ports = ingressPorts
 	}
 
 	// Ensure frontendNLBDNS is appended if it is not already added
@@ -405,21 +416,56 @@ func isIngressStatusEqual(a, b []networking.IngressLoadBalancerIngress) bool {
 		return false
 	}
 
-	setA := make(map[string]struct{}, len(a))
-	setB := make(map[string]struct{}, len(b))
+	hostnameToPortsA := make(map[string]map[int32]struct{})
+	hostnameToPortsB := make(map[string]map[int32]struct{})
 
 	for _, ingress := range a {
-		setA[ingress.Hostname] = struct{}{}
+		if ingress.Hostname == "" {
+			continue
+		}
+
+		portSet := make(map[int32]struct{})
+		for _, portStatus := range ingress.Ports {
+			portSet[portStatus.Port] = struct{}{}
+		}
+		hostnameToPortsA[ingress.Hostname] = portSet
 	}
 
 	for _, ingress := range b {
-		setB[ingress.Hostname] = struct{}{}
+		if ingress.Hostname == "" {
+			continue
+		}
+
+		portSet := make(map[int32]struct{})
+		for _, portStatus := range ingress.Ports {
+			portSet[portStatus.Port] = struct{}{}
+		}
+		hostnameToPortsB[ingress.Hostname] = portSet
 	}
 
-	for key := range setA {
-		if _, exists := setB[key]; !exists {
+	// Check if the maps are equal (same hostnames with same ports)
+	if len(hostnameToPortsA) != len(hostnameToPortsB) {
+		return false
+	}
+
+	// Check if all hostnames in A exist in B with the same ports
+	for hostname, portsA := range hostnameToPortsA {
+		portsB, exists := hostnameToPortsB[hostname]
+		if !exists {
+			return false // Hostname in A doesn't exist in B
+		}
+
+		// Check if port sets are equal (same length and same values)
+		if len(portsA) != len(portsB) {
 			return false
 		}
+
+		// Check if all ports in A exist in B
+		for port := range portsA {
+			if _, exists := portsB[port]; !exists {
+				return false
+			}
+		}
 	}
 	return true
 }