Exclude terminated Pods from group members (#6508) (#6519)

When calculating AddressGroups, terminated Pods should be excluded because their IPs can be recycled and reused by other active Pods. When calculating AppliedToGroups and EgressGroups, terminated Pods could be excluded as their network resources including network interfaces have been deleted. Signed-off-by: Quan Tian <[email protected]>
antrea-io · Jul 10, 2024 · be2e919 · be2e919
1 parent 67a1ce9
commit be2e919
Show file tree

Hide file tree

Showing 5 changed files with 102 additions and 8 deletions.
diff --git a/pkg/controller/egress/controller.go b/pkg/controller/egress/controller.go
@@ -44,6 +44,7 @@ import (
 	"antrea.io/antrea/pkg/controller/externalippool"
 	"antrea.io/antrea/pkg/controller/grouping"
 	antreatypes "antrea.io/antrea/pkg/controller/types"
+	"antrea.io/antrea/pkg/util/k8s"
 )
 
 const (
@@ -371,9 +372,9 @@ func (c *EgressController) syncEgress(key string) error {
 	egressGroup := egressGroupObj.(*antreatypes.EgressGroup)
 	pods, _ := c.groupingInterface.GetEntities(egressGroupType, key)
 	for _, pod := range pods {
-		// Ignore Pod if it's not scheduled or not running. And Egress does not support HostNetwork Pods, so also ignore
+		// Ignore Pod if it's not scheduled or is already terminated. And Egress does not support HostNetwork Pods, so also ignore
 		// Pod if it's HostNetwork Pod.
-		if pod.Spec.NodeName == "" || pod.Spec.HostNetwork {
+		if pod.Spec.NodeName == "" || pod.Spec.HostNetwork || k8s.IsPodTerminated(pod) {
 			continue
 		}
 		podNum++

diff --git a/pkg/controller/egress/controller_test.go b/pkg/controller/egress/controller_test.go
@@ -203,6 +203,11 @@ func newController(objects, crdObjects []runtime.Object) *egressController {
 }
 
 func TestAddEgress(t *testing.T) {
+	podSucceeded := newPod("default", "succeeded-pod", map[string]string{"app": "foo"}, node1, "1.1.5.1", false)
+	podSucceeded.Status.Phase = v1.PodSucceeded
+	podFailed := newPod("default", "failed-pod", map[string]string{"app": "foo"}, node1, "1.1.5.2", false)
+	podFailed.Status.Phase = v1.PodFailed
+
 	tests := []struct {
 		name                 string
 		inputEgress          *v1beta1.Egress
@@ -347,7 +352,7 @@ func TestAddEgress(t *testing.T) {
 			defer close(stopCh)
 			var fakeObjects []runtime.Object
 			fakeObjects = append(fakeObjects, nsDefault, nsOther)
-			fakeObjects = append(fakeObjects, podFoo1, podFoo2, podBar1, podFoo1InOtherNamespace, podUnscheduled, podNonIP, podWithHostNetwork)
+			fakeObjects = append(fakeObjects, podFoo1, podFoo2, podBar1, podFoo1InOtherNamespace, podUnscheduled, podNonIP, podWithHostNetwork, podSucceeded, podFailed)
 			var fakeCRDObjects []runtime.Object
 			fakeCRDObjects = append(fakeCRDObjects, eipFoo1)
 			controller := newController(fakeObjects, fakeCRDObjects)

diff --git a/pkg/controller/networkpolicy/networkpolicy_controller.go b/pkg/controller/networkpolicy/networkpolicy_controller.go
@@ -1184,9 +1184,9 @@ func (n *NetworkPolicyController) getMemberSetForGroupType(groupType grouping.Gr
 	groupMemberSet := controlplane.GroupMemberSet{}
 	pods, externalEntities := n.groupingInterface.GetEntities(groupType, name)
 	for _, pod := range pods {
-		// HostNetwork Pods should be excluded from group members
-		// https://github.com/antrea-io/antrea/issues/3078
-		if pod.Spec.HostNetwork == true || len(pod.Status.PodIPs) == 0 {
+		// HostNetwork Pods should be excluded from group members: https://github.com/antrea-io/antrea/issues/3078.
+		// Terminated Pods should be excluded as their IPs can be recycled and used by other Pods.
+		if pod.Spec.HostNetwork || k8s.IsPodTerminated(pod) || len(pod.Status.PodIPs) == 0 {
 			continue
 		}
 		groupMemberSet.Insert(podToGroupMember(pod, true))
@@ -1329,8 +1329,8 @@ func (n *NetworkPolicyController) syncAppliedToGroup(key string) error {
 		} else {
 			scheduledPodNum, scheduledExtEntityNum := 0, 0
 			for _, pod := range pods {
-				if pod.Spec.NodeName == "" || pod.Spec.HostNetwork == true {
-					// No need to process Pod when it's not scheduled.
+				if pod.Spec.NodeName == "" || pod.Spec.HostNetwork || k8s.IsPodTerminated(pod) {
+					// No need to process Pod when it's not scheduled or is already terminated.
 					// HostNetwork Pods will not be applied to by policies.
 					continue
 				}

diff --git a/pkg/controller/networkpolicy/networkpolicy_controller_test.go b/pkg/controller/networkpolicy/networkpolicy_controller_test.go
@@ -614,6 +614,72 @@ func TestAddPod(t *testing.T) {
 			outAddressGroupMatch: false,
 			groupMatch:           false,
 		},
+		{
+			name: "match-all-selectors-succeeded",
+			addedPod: &corev1.Pod{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "podA",
+					Namespace: "nsA",
+					Labels: map[string]string{
+						"role":         "app",
+						"group":        "appliedTo",
+						"inGroup":      "inAddress",
+						"outGroup":     "outAddress",
+						"clustergroup": "yes",
+					},
+				},
+				Spec: corev1.PodSpec{
+					Containers: []corev1.Container{{
+						Name: "container-1",
+					}},
+					NodeName: "nodeA",
+				},
+				Status: corev1.PodStatus{
+					Phase: corev1.PodSucceeded,
+					PodIP: "1.2.3.4",
+					PodIPs: []corev1.PodIP{
+						{IP: "1.2.3.4"},
+					},
+				},
+			},
+			appGroupMatch:        false,
+			inAddressGroupMatch:  false,
+			outAddressGroupMatch: false,
+			groupMatch:           false,
+		},
+		{
+			name: "match-all-selectors-failed",
+			addedPod: &corev1.Pod{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "podA",
+					Namespace: "nsA",
+					Labels: map[string]string{
+						"role":         "app",
+						"group":        "appliedTo",
+						"inGroup":      "inAddress",
+						"outGroup":     "outAddress",
+						"clustergroup": "yes",
+					},
+				},
+				Spec: corev1.PodSpec{
+					Containers: []corev1.Container{{
+						Name: "container-1",
+					}},
+					NodeName: "nodeA",
+				},
+				Status: corev1.PodStatus{
+					Phase: corev1.PodFailed,
+					PodIP: "1.2.3.4",
+					PodIPs: []corev1.PodIP{
+						{IP: "1.2.3.4"},
+					},
+				},
+			},
+			appGroupMatch:        false,
+			inAddressGroupMatch:  false,
+			outAddressGroupMatch: false,
+			groupMatch:           false,
+		},
 		{
 			name: "match-spec-podselector-no-podip",
 			addedPod: &corev1.Pod{

diff --git a/pkg/util/k8s/pod.go b/pkg/util/k8s/pod.go
@@ -0,0 +1,22 @@
+// Copyright 2024 Antrea Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package k8s
+
+import v1 "k8s.io/api/core/v1"
+
+// IsPodTerminated returns true if a pod is terminated, all containers are stopped and cannot ever regress.
+func IsPodTerminated(pod *v1.Pod) bool {
+	return pod.Status.Phase == v1.PodFailed || pod.Status.Phase == v1.PodSucceeded
+}