Improve EndpointController's handling of headless services under dual-stack

EndpointController was accidentally requiring all headless services to
be IPv4-only in clusters with IPv6DualStack enabled.

This still leaves "legacy" (ie, IPFamily-less) headless services as
always IPv4-only because the controller doesn't currently have easy
access to the information that would allow it to fix that.
(EndpointSliceController had the same problem already, and still
does.) This can be fixed, if needed, by manually setting IPFamily,
and the proposed API for 1.20 will handle this situation better.

pkg/controller/endpoint/endpoints_controller.go

@@ -215,19 +215,13 @@ func podToEndpointAddressForService(svc *v1.Service, pod *v1.Pod) (*v1.EndpointA
 	var endpointIP string
 
 	if !utilfeature.DefaultFeatureGate.Enabled(features.IPv6DualStack) {
+		// In a legacy cluster, the pod IP is guaranteed to be usable
 		endpointIP = pod.Status.PodIP
 	} else {
-		// api-server service controller ensured that the service got the correct IP Family
-		// according to user setup, here we only need to match EndPoint IPs' family to service
-		// actual IP family. as in, we don't need to check service.IPFamily
-
-		ipv6ClusterIP := utilnet.IsIPv6String(svc.Spec.ClusterIP)
+		ipv6Service := endpointutil.IsIPv6Service(svc)
 		for _, podIP := range pod.Status.PodIPs {
 			ipv6PodIP := utilnet.IsIPv6String(podIP.IP)
-			// same family?
-			// TODO (khenidak) when we remove the max of 2 PodIP limit from pods
-			// we will have to return multiple endpoint addresses
-			if ipv6ClusterIP == ipv6PodIP {
+			if ipv6Service == ipv6PodIP {
 				endpointIP = podIP.IP
 				break
 			}

pkg/controller/endpoint/endpoints_controller_test.go

@@ -1249,21 +1249,21 @@ func TestPodToEndpointAddressForService(t *testing.T) {
 
 			expectedEndpointFamily: ipv6,
 		},
-		// {
-		// 	name: "v6 headless service, in a dual stack cluster",
-		//
-		// 	enableDualStack: true,
-		// 	ipFamilies:      ipv4ipv6,
-		//
-		// 	service: v1.Service{
-		// 		Spec: v1.ServiceSpec{
-		// 			ClusterIP: v1.ClusterIPNone,
-		// 			IPFamily:  &ipv6,
-		// 		},
-		// 	},
-		//
-		// 	expectedEndpointFamily: ipv6,
-		// },
+		{
+			name: "v6 headless service, in a dual stack cluster",
+
+			enableDualStack: true,
+			ipFamilies:      ipv4ipv6,
+
+			service: v1.Service{
+				Spec: v1.ServiceSpec{
+					ClusterIP: v1.ClusterIPNone,
+					IPFamily:  &ipv6,
+				},
+			},
+
+			expectedEndpointFamily: ipv6,
+		},
 		{
 			name: "v6 legacy headless service, in a dual stack cluster",
 

pkg/controller/endpointslice/reconciler.go

@@ -59,7 +59,7 @@ type endpointMeta struct {
 func (r *reconciler) reconcile(service *corev1.Service, pods []*corev1.Pod, existingSlices []*discovery.EndpointSlice, triggerTime time.Time) error {
 	addressType := discovery.AddressTypeIPv4
 
-	if isIPv6Service(service) {
+	if endpointutil.IsIPv6Service(service) {
 		addressType = discovery.AddressTypeIPv6
 	}
 

pkg/controller/endpointslice/utils.go

@@ -120,7 +120,7 @@ func getEndpointAddresses(podStatus corev1.PodStatus, service *corev1.Service) [
 
 	for _, podIP := range podStatus.PodIPs {
 		isIPv6PodIP := utilnet.IsIPv6String(podIP.IP)
-		if isIPv6PodIP == isIPv6Service(service) {
+		if isIPv6PodIP == endpointutil.IsIPv6Service(service) {
 			addresses = append(addresses, podIP.IP)
 		}
 	}
@@ -128,12 +128,6 @@ func getEndpointAddresses(podStatus corev1.PodStatus, service *corev1.Service) [
 	return addresses
 }
 
-// isIPv6Service returns true if the Service uses IPv6 addresses.
-func isIPv6Service(service *corev1.Service) bool {
-	// IPFamily is not guaranteed to be set, even in an IPv6 only cluster.
-	return (service.Spec.IPFamily != nil && *service.Spec.IPFamily == corev1.IPv6Protocol) || utilnet.IsIPv6String(service.Spec.ClusterIP)
-}
-
 // endpointsEqualBeyondHash returns true if endpoints have equal attributes
 // but excludes equality checks that would have already been covered with
 // endpoint hashing (see hashEndpoint func for more info).

pkg/controller/util/endpoint/BUILD

@@ -10,6 +10,7 @@ go_library(
     visibility = ["//visibility:public"],
     deps = [
         "//pkg/api/v1/pod:go_default_library",
+        "//pkg/apis/core/v1/helper:go_default_library",
         "//pkg/controller:go_default_library",
         "//pkg/util/hash:go_default_library",
         "//staging/src/k8s.io/api/core/v1:go_default_library",
@@ -19,6 +20,7 @@ go_library(
         "//staging/src/k8s.io/apimachinery/pkg/util/sets:go_default_library",
         "//staging/src/k8s.io/client-go/listers/core/v1:go_default_library",
         "//staging/src/k8s.io/client-go/tools/cache:go_default_library",
+        "//vendor/k8s.io/utils/net:go_default_library",
     ],
 )
 

pkg/controller/util/endpoint/controller_utils.go

@@ -32,8 +32,10 @@ import (
 	v1listers "k8s.io/client-go/listers/core/v1"
 	"k8s.io/client-go/tools/cache"
 	podutil "k8s.io/kubernetes/pkg/api/v1/pod"
+	"k8s.io/kubernetes/pkg/apis/core/v1/helper"
 	"k8s.io/kubernetes/pkg/controller"
 	"k8s.io/kubernetes/pkg/util/hash"
+	utilnet "k8s.io/utils/net"
 )
 
 // ServiceSelectorCache is a cache of service selectors to avoid high CPU consumption caused by frequent calls to AsSelectorPreValidated (see #73527)
@@ -275,3 +277,18 @@ func (sl portsInOrder) Less(i, j int) bool {
 	h2 := DeepHashObjectToString(sl[j])
 	return h1 < h2
 }
+
+// IsIPv6Service checks if svc should have IPv6 endpoints
+func IsIPv6Service(svc *v1.Service) bool {
+	if helper.IsServiceIPSet(svc) {
+		return utilnet.IsIPv6String(svc.Spec.ClusterIP)
+	} else if svc.Spec.IPFamily != nil {
+		return *svc.Spec.IPFamily == v1.IPv6Protocol
+	} else {
+		// FIXME: for legacy headless Services with no IPFamily, the current
+		// thinking is that we should use the cluster default. Unfortunately
+		// the endpoint controller doesn't know the cluster default. For now,
+		// assume it's IPv4.
+		return false
+	}
+}