tests: update e2e test for endpointslice compatibility

Signed-off-by: Jayapriya Pai <[email protected]>

Author: slashpai

test/e2e/delete_test.go

@@ -183,9 +183,20 @@ func (tc *testContext) checkAWSMetadataEndpointRouteIsRestored(t *testing.T, add
 	if tc.CloudProvider.GetType() != config.AWSPlatformType {
 		t.Skipf("Skipping for %s", tc.CloudProvider.GetType())
 	}
-	out, err := tc.runPowerShellSSHJob("check-routes", "Get-NetRoute", address)
-	require.NoError(t, err, "error checking routes")
-	assert.True(t, strings.Contains(out, "169.254.169.254"), "metadata endpoint route is not restored")
+
+	// To allow time for route restoration during node cleanup
+	// since EndpointSlice discovery may have
+	// different cleanup timing compared to Endpoints.
+	err := wait.PollUntilContextTimeout(context.Background(), tc.retryInterval, 2*time.Minute, true,
+		func(ctx context.Context) (bool, error) {
+			out, err := tc.runPowerShellSSHJob("check-routes", "Get-NetRoute", address)
+			if err != nil {
+				t.Logf("error checking routes (will retry): %v", err)
+				return false, nil
+			}
+			return strings.Contains(out, "169.254.169.254"), nil
+		})
+	require.NoError(t, err, "metadata endpoint route was not restored within timeout")
 }
 
 // waitForWindowsNodeRemoval returns when there are zero Windows nodes of the given type, machine or byoh, in the cluster

test/e2e/metrics_test.go

@@ -239,9 +239,12 @@ func (tc *testContext) testPodMetrics(t *testing.T, podName string) {
 		fmt.Sprintf("pod_interface_network:container_network_receive_bytes:irate5m{pod='%s',namespace='%s'}", podName, tc.workloadNamespace),
 		fmt.Sprintf("pod_interface_network:container_network_transmit_bytes_total:irate5m{pod='%s',namespace='%s'}", podName, tc.workloadNamespace),
 	}
+	// Use extended timeout to account for EndpointSlice propagation, metric scraping,
+	// and recording rule evaluation when using EndpointSlice-based service discovery
+	podMetricsTimeout := 5 * time.Minute
 	for i, query := range queries {
 		t.Run("query "+strconv.Itoa(i), func(t *testing.T) {
-			err := wait.PollUntilContextTimeout(context.TODO(), retry.Interval, retry.ResourceChangeTimeout, true,
+			err := wait.PollUntilContextTimeout(context.TODO(), retry.Interval, podMetricsTimeout, true,
 				func(ctx context.Context) (done bool, err error) {
 					results, err := makePrometheusQuery(prometheusRoute.Spec.Host, query, prometheusToken)
 					if err != nil {