test: fix race condition in config daemon pod restart test

getConfigDaemonPod had a hard Gomega Expect assertion on pod count,
causing it to panic when called inside Eventually immediately after
pod deletion — before the DaemonSet replacement was created.

Change getConfigDaemonPod to return (*corev1.Pod, error) so callers
inside Eventually can retry gracefully on transient zero-pod states.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: Zhiqiang Fang

test/conformance/tests/test_aws_platform.go

@@ -182,7 +182,8 @@ var _ = Describe("[sriov] aws platform", Ordered, func() {
 			savedStatus := nodeState.Status
 
 			By("delete config daemon pod on the node")
-			configDaemonPod := getConfigDaemonPod(node)
+			configDaemonPod, err := getConfigDaemonPod(node)
+			Expect(err).ToNot(HaveOccurred())
 			oldPodName := configDaemonPod.Name
 			err = clients.Pods(operatorNamespace).Delete(context.Background(), configDaemonPod.Name, metav1.DeleteOptions{
 				GracePeriodSeconds: ptr.To(int64(0)),
@@ -191,9 +192,9 @@ var _ = Describe("[sriov] aws platform", Ordered, func() {
 
 			By("waiting for the new config daemon pod to be running")
 			Eventually(func() bool {
-				newPod := getConfigDaemonPod(node)
-				// Make sure we got a different pod (new one)
-				if newPod.Name == oldPodName {
+				newPod, err := getConfigDaemonPod(node)
+				// Make sure the replacement pod exists and is different from the deleted one
+				if err != nil || newPod.Name == oldPodName {
 					return false
 				}
 				return newPod.Status.Phase == corev1.PodRunning

test/conformance/tests/test_sriov_operator.go

@@ -1964,20 +1964,24 @@ func createVanillaNetworkPolicy(node string, sriovInfos *cluster.EnabledNodes, n
 		})))
 }
 
-func getConfigDaemonPod(nodeName string) *corev1.Pod {
+func getConfigDaemonPod(nodeName string) (*corev1.Pod, error) {
 	pods := &corev1.PodList{}
 	label, err := labels.Parse("app=sriov-network-config-daemon")
 	Expect(err).ToNot(HaveOccurred())
 	field, err := fields.ParseSelector(fmt.Sprintf("spec.nodeName=%s", nodeName))
 	Expect(err).ToNot(HaveOccurred())
 	err = clients.List(context.Background(), pods, &runtimeclient.ListOptions{Namespace: operatorNamespace, LabelSelector: label, FieldSelector: field})
 	Expect(err).ToNot(HaveOccurred())
-	Expect(len(pods.Items)).To(Equal(1))
-	return &pods.Items[0]
+	if len(pods.Items) != 1 {
+		return nil, fmt.Errorf("expected 1 config daemon pod on node %s, got %d", nodeName, len(pods.Items))
+	}
+	return &pods.Items[0], nil
 }
 
 func runCommandOnConfigDaemon(nodeName string, command ...string) (string, string, error) {
-	output, errOutput, err := pod.ExecCommand(clients, getConfigDaemonPod(nodeName), command...)
+	daemonPod, err := getConfigDaemonPod(nodeName)
+	Expect(err).ToNot(HaveOccurred())
+	output, errOutput, err := pod.ExecCommand(clients, daemonPod, command...)
 	return output, errOutput, err
 }