/* Copyright 2017 Jetstack Ltd. 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 framework import ( "context" "crypto/tls" "fmt" "net" "net/http" "strings" "time" "k8s.io/ingress-nginx/test/e2e/framework/httpexpect" "github.com/onsi/ginkgo/v2" "github.com/stretchr/testify/assert" appsv1 "k8s.io/api/apps/v1" corev1 "k8s.io/api/core/v1" v1 "k8s.io/api/core/v1" networking "k8s.io/api/networking/v1" apiextcs "k8s.io/apiextensions-apiserver/pkg/client/clientset/clientset" apierrors "k8s.io/apimachinery/pkg/api/errors" metav1 "k8s.io/apimachinery/pkg/apis/meta/v1" "k8s.io/apimachinery/pkg/fields" "k8s.io/apimachinery/pkg/util/wait" "k8s.io/client-go/kubernetes" "k8s.io/client-go/kubernetes/scheme" "k8s.io/client-go/rest" restclient "k8s.io/client-go/rest" "k8s.io/klog/v2" ) // RequestScheme define a scheme used in a test request. type RequestScheme string // These are valid test request schemes. const ( HTTP RequestScheme = "http" HTTPS RequestScheme = "https" ) var ( // KubectlPath defines the full path of the kubectl binary KubectlPath = "/usr/local/bin/kubectl" ) // Framework supports common operations used by e2e tests; it will keep a client & a namespace for you. type Framework struct { BaseName string // A Kubernetes and Service Catalog client KubeClientSet kubernetes.Interface KubeConfig *restclient.Config APIExtensionsClientSet apiextcs.Interface Namespace string IngressClass string pod *corev1.Pod } // NewDefaultFramework makes a new framework and sets up a BeforeEach/AfterEach for // you (you can write additional before/after each functions). func NewDefaultFramework(baseName string) *Framework { defer ginkgo.GinkgoRecover() f := &Framework{ BaseName: baseName, } ginkgo.BeforeEach(f.BeforeEach) ginkgo.AfterEach(f.AfterEach) return f } // NewSimpleFramework makes a new framework that allows the usage of a namespace // for arbitraty tests. func NewSimpleFramework(baseName string) *Framework { defer ginkgo.GinkgoRecover() f := &Framework{ BaseName: baseName, } ginkgo.BeforeEach(f.CreateEnvironment) ginkgo.AfterEach(f.DestroyEnvironment) return f } func (f *Framework) CreateEnvironment() { var err error if f.KubeClientSet == nil { f.KubeConfig, err = loadConfig() assert.Nil(ginkgo.GinkgoT(), err, "loading a kubernetes client configuration") // TODO: remove after k8s v1.22 f.KubeConfig.WarningHandler = rest.NoWarnings{} f.KubeClientSet, err = kubernetes.NewForConfig(f.KubeConfig) assert.Nil(ginkgo.GinkgoT(), err, "creating a kubernetes client") } f.Namespace, err = CreateKubeNamespace(f.BaseName, f.KubeClientSet) assert.Nil(ginkgo.GinkgoT(), err, "creating namespace") } func (f *Framework) DestroyEnvironment() { defer ginkgo.GinkgoRecover() err := DeleteKubeNamespace(f.KubeClientSet, f.Namespace) assert.Nil(ginkgo.GinkgoT(), err, "deleting namespace %v", f.Namespace) } // BeforeEach gets a client and makes a namespace. func (f *Framework) BeforeEach() { var err error f.CreateEnvironment() f.IngressClass, err = CreateIngressClass(f.Namespace, f.KubeClientSet) assert.Nil(ginkgo.GinkgoT(), err, "creating IngressClass") err = f.newIngressController(f.Namespace, f.BaseName) assert.Nil(ginkgo.GinkgoT(), err, "deploying the ingress controller") err = f.updateIngressNGINXPod() assert.Nil(ginkgo.GinkgoT(), err, "updating ingress controller pod information") f.WaitForNginxListening(80) } // AfterEach deletes the namespace, after reading its events. func (f *Framework) AfterEach() { defer f.DestroyEnvironment() defer func(kubeClient kubernetes.Interface, ingressclass string) { defer ginkgo.GinkgoRecover() err := deleteIngressClass(kubeClient, ingressclass) assert.Nil(ginkgo.GinkgoT(), err, "deleting IngressClass") }(f.KubeClientSet, f.IngressClass) if !ginkgo.CurrentSpecReport().Failed() { return } cmd := fmt.Sprintf("cat /etc/nginx/nginx.conf") o, err := f.ExecCommand(f.pod, cmd) if err != nil { Logf("Unexpected error obtaining nginx.conf file: %v", err) return } ginkgo.By("Dumping NGINX configuration after failure") Logf("%v", o) log, err := f.NginxLogs() if err != nil { Logf("Unexpected error obtaining NGINX logs: %v", err) return } ginkgo.By("Dumping NGINX logs") Logf("%v", log) o, err = f.NamespaceContent() if err != nil { Logf("Unexpected error obtaining namespace information: %v", err) return } ginkgo.By("Dumping namespace content") Logf("%v", o) } // IngressNginxDescribe wrapper function for ginkgo describe. Adds namespacing. func IngressNginxDescribe(text string, body func()) bool { return ginkgo.Describe(text, body) } // DescribeAnnotation wrapper function for ginkgo describe. Adds namespacing. func DescribeAnnotation(text string, body func()) bool { return ginkgo.Describe("[Annotations] "+text, body) } // DescribeSetting wrapper function for ginkgo describe. Adds namespacing. func DescribeSetting(text string, body func()) bool { return ginkgo.Describe("[Setting] "+text, body) } // MemoryLeakIt is wrapper function for ginkgo It. Adds "[MemoryLeak]" tag and makes static analysis easier. func MemoryLeakIt(text string, body interface{}) bool { return ginkgo.It(text+" [MemoryLeak]", body) } // GetNginxIP returns the number of TCP port where NGINX is running func (f *Framework) GetNginxIP() string { s, err := f.KubeClientSet. CoreV1(). Services(f.Namespace). Get(context.TODO(), "nginx-ingress-controller", metav1.GetOptions{}) assert.Nil(ginkgo.GinkgoT(), err, "obtaining NGINX IP address") return s.Spec.ClusterIP } // GetNginxPodIP returns the IP addresses of the running pods func (f *Framework) GetNginxPodIP() string { return f.pod.Status.PodIP } // GetURL returns the URL should be used to make a request to NGINX func (f *Framework) GetURL(scheme RequestScheme) string { ip := f.GetNginxIP() return fmt.Sprintf("%v://%v", scheme, ip) } // GetIngressNGINXPod returns the ingress controller running pod func (f *Framework) GetIngressNGINXPod() *corev1.Pod { return f.pod } // UpdateIngressNGINXPod search and updates the ingress controller running pod func (f *Framework) updateIngressNGINXPod() error { var err error f.pod, err = getIngressNGINXPod(f.Namespace, f.KubeClientSet) return err } // WaitForNginxServer waits until the nginx configuration contains a particular server section. // `cfg` passed to matcher is normalized by replacing all tabs and spaces with single space. func (f *Framework) WaitForNginxServer(name string, matcher func(cfg string) bool) { err := wait.Poll(Poll, DefaultTimeout, f.matchNginxConditions(name, matcher)) assert.Nil(ginkgo.GinkgoT(), err, "waiting for nginx server condition/s") Sleep(1 * time.Second) } // WaitForNginxConfiguration waits until the nginx configuration contains a particular configuration // `cfg` passed to matcher is normalized by replacing all tabs and spaces with single space. func (f *Framework) WaitForNginxConfiguration(matcher func(cfg string) bool) { err := wait.Poll(Poll, DefaultTimeout, f.matchNginxConditions("", matcher)) assert.Nil(ginkgo.GinkgoT(), err, "waiting for nginx server condition/s") Sleep(1 * time.Second) } // WaitForNginxCustomConfiguration waits until the nginx configuration given part (from, to) contains a particular configuration func (f *Framework) WaitForNginxCustomConfiguration(from string, to string, matcher func(cfg string) bool) { err := wait.Poll(Poll, DefaultTimeout, f.matchNginxCustomConditions(from, to, matcher)) assert.Nil(ginkgo.GinkgoT(), err, "waiting for nginx server condition/s") } // NginxLogs returns the logs of the nginx ingress controller pod running func (f *Framework) NginxLogs() (string, error) { if isRunning, err := podRunningReady(f.pod); err == nil && isRunning { return Logs(f.KubeClientSet, f.Namespace, f.pod.Name) } return "", fmt.Errorf("no nginx ingress controller pod is running (logs)") } func (f *Framework) matchNginxConditions(name string, matcher func(cfg string) bool) wait.ConditionFunc { return func() (bool, error) { var cmd string if name == "" { cmd = fmt.Sprintf("cat /etc/nginx/nginx.conf") } else { cmd = fmt.Sprintf("cat /etc/nginx/nginx.conf | awk '/## start server %v/,/## end server %v/'", name, name) } o, err := f.ExecCommand(f.pod, cmd) if err != nil { return false, nil } if klog.V(10).Enabled() && len(o) > 0 { klog.InfoS("NGINX", "configuration", o) } // passes the nginx config to the passed function if matcher(strings.Join(strings.Fields(o), " ")) { return true, nil } return false, nil } } func (f *Framework) matchNginxCustomConditions(from string, to string, matcher func(cfg string) bool) wait.ConditionFunc { return func() (bool, error) { cmd := fmt.Sprintf("cat /etc/nginx/nginx.conf| awk '/%v/,/%v/'", from, to) o, err := f.ExecCommand(f.pod, cmd) if err != nil { return false, nil } if klog.V(10).Enabled() && len(o) > 0 { klog.InfoS("NGINX", "configuration", o) } // passes the nginx config to the passed function if matcher(strings.Join(strings.Fields(o), " ")) { return true, nil } return false, nil } } func (f *Framework) getConfigMap(name string) (*v1.ConfigMap, error) { if f.KubeClientSet == nil { return nil, fmt.Errorf("KubeClientSet not initialized") } config, err := f.KubeClientSet. CoreV1(). ConfigMaps(f.Namespace). Get(context.TODO(), name, metav1.GetOptions{}) if err != nil { return nil, err } return config, err } // SetNginxConfigMapData sets ingress-nginx's nginx-ingress-controller configMap data func (f *Framework) SetNginxConfigMapData(cmData map[string]string) { cfgMap, err := f.getConfigMap("nginx-ingress-controller") assert.Nil(ginkgo.GinkgoT(), err) assert.NotNil(ginkgo.GinkgoT(), cfgMap, "expected a configmap but none returned") cfgMap.Data = cmData fn := func() { _, err = f.KubeClientSet. CoreV1(). ConfigMaps(f.Namespace). Update(context.TODO(), cfgMap, metav1.UpdateOptions{}) assert.Nil(ginkgo.GinkgoT(), err, "updating configuration configmap") } f.WaitForReload(fn) } // CreateConfigMap creates a new configmap in the current namespace func (f *Framework) CreateConfigMap(name string, data map[string]string) { _, err := f.KubeClientSet.CoreV1().ConfigMaps(f.Namespace).Create(context.TODO(), &v1.ConfigMap{ ObjectMeta: metav1.ObjectMeta{ Name: name, Namespace: f.Namespace, }, Data: data, }, metav1.CreateOptions{}) assert.Nil(ginkgo.GinkgoT(), err, "creating configMap") } // UpdateNginxConfigMapData updates single field in ingress-nginx's nginx-ingress-controller map data func (f *Framework) UpdateNginxConfigMapData(key string, value string) { config, err := f.getConfigMap("nginx-ingress-controller") assert.Nil(ginkgo.GinkgoT(), err) assert.NotNil(ginkgo.GinkgoT(), config, "expected a configmap but none returned") config.Data[key] = value fn := func() { _, err = f.KubeClientSet. CoreV1(). ConfigMaps(f.Namespace). Update(context.TODO(), config, metav1.UpdateOptions{}) assert.Nil(ginkgo.GinkgoT(), err, "updating configuration configmap") } f.WaitForReload(fn) } // WaitForReload calls the passed function and // asser it has caused at least 1 reload. func (f *Framework) WaitForReload(fn func()) { initialReloadCount := getReloadCount(f.pod, f.Namespace, f.KubeClientSet) fn() count := 0 err := wait.Poll(1*time.Second, DefaultTimeout, func() (bool, error) { reloads := getReloadCount(f.pod, f.Namespace, f.KubeClientSet) // most of the cases reload the ingress controller // in cases where the value is not modified we could wait forever if count > 5 && reloads == initialReloadCount { return true, nil } count++ return (reloads > initialReloadCount), nil }) assert.Nil(ginkgo.GinkgoT(), err, "while waiting for ingress controller reload") } func getReloadCount(pod *corev1.Pod, namespace string, client kubernetes.Interface) int { events, err := client.CoreV1().Events(namespace).Search(scheme.Scheme, pod) assert.Nil(ginkgo.GinkgoT(), err, "obtaining NGINX Pod") reloadCount := 0 for _, e := range events.Items { if e.Reason == "RELOAD" && e.Type == corev1.EventTypeNormal { reloadCount++ } } return reloadCount } // DeleteNGINXPod deletes the currently running pod. It waits for the replacement pod to be up. // Grace period to wait for pod shutdown is in seconds. func (f *Framework) DeleteNGINXPod(grace int64) { ns := f.Namespace err := f.KubeClientSet.CoreV1().Pods(ns).Delete(context.TODO(), f.pod.GetName(), *metav1.NewDeleteOptions(grace)) assert.Nil(ginkgo.GinkgoT(), err, "deleting ingress nginx pod") err = wait.Poll(Poll, DefaultTimeout, func() (bool, error) { err := f.updateIngressNGINXPod() if err != nil || f.pod == nil { return false, nil } return f.pod.GetName() != "", nil }) assert.Nil(ginkgo.GinkgoT(), err, "while waiting for ingress nginx pod to come up again") } // HTTPDumbTestClient returns a new httpexpect client without BaseURL. func (f *Framework) HTTPDumbTestClient() *httpexpect.HTTPRequest { return f.newHTTPTestClient(nil, false) } // HTTPTestClient returns a new HTTPRequest client for end-to-end HTTP testing. func (f *Framework) HTTPTestClient() *httpexpect.HTTPRequest { return f.newHTTPTestClient(nil, true) } // HTTPTestClientWithTLSConfig returns a new httpexpect client for end-to-end // HTTP testing with a custom TLS configuration. func (f *Framework) HTTPTestClientWithTLSConfig(config *tls.Config) *httpexpect.HTTPRequest { return f.newHTTPTestClient(config, true) } func (f *Framework) newHTTPTestClient(config *tls.Config, setIngressURL bool) *httpexpect.HTTPRequest { if config == nil { config = &tls.Config{ InsecureSkipVerify: true, } } var baseURL string if setIngressURL { baseURL = f.GetURL(HTTP) } return httpexpect.NewRequest(baseURL, &http.Client{ Transport: &http.Transport{ TLSClientConfig: config, }, CheckRedirect: func(req *http.Request, via []*http.Request) error { return http.ErrUseLastResponse }, }, httpexpect.NewAssertReporter()) } // WaitForNginxListening waits until NGINX starts accepting connections on a port func (f *Framework) WaitForNginxListening(port int) { err := waitForPodsReady(f.KubeClientSet, DefaultTimeout, 1, f.Namespace, metav1.ListOptions{ LabelSelector: "app.kubernetes.io/name=ingress-nginx", }) assert.Nil(ginkgo.GinkgoT(), err, "waiting for ingress pods to be ready") podIP := f.GetNginxIP() err = wait.Poll(500*time.Millisecond, DefaultTimeout, func() (bool, error) { hostPort := net.JoinHostPort(podIP, fmt.Sprintf("%v", port)) conn, err := net.Dial("tcp", hostPort) if err != nil { return false, nil } defer conn.Close() return true, nil }) assert.Nil(ginkgo.GinkgoT(), err, "waiting for ingress controller pod listening on port 80") } // WaitForPod waits for a specific Pod to be ready, using a label selector func (f *Framework) WaitForPod(selector string, timeout time.Duration, shouldFail bool) { err := waitForPodsReady(f.KubeClientSet, timeout, 1, f.Namespace, metav1.ListOptions{ LabelSelector: selector, }) if shouldFail { assert.NotNil(ginkgo.GinkgoT(), err, "waiting for pods to be ready") } else { assert.Nil(ginkgo.GinkgoT(), err, "waiting for pods to be ready") } } // UpdateDeployment runs the given updateFunc on the deployment and waits for it to be updated func UpdateDeployment(kubeClientSet kubernetes.Interface, namespace string, name string, replicas int, updateFunc func(d *appsv1.Deployment) error) error { deployment, err := kubeClientSet.AppsV1().Deployments(namespace).Get(context.TODO(), name, metav1.GetOptions{}) if err != nil { return err } if updateFunc != nil { if err := updateFunc(deployment); err != nil { return err } err = waitForDeploymentRollout(kubeClientSet, deployment) if err != nil { return err } } if *deployment.Spec.Replicas != int32(replicas) { deployment.Spec.Replicas = NewInt32(int32(replicas)) _, err = kubeClientSet.AppsV1().Deployments(namespace).Update(context.TODO(), deployment, metav1.UpdateOptions{}) if err != nil { return fmt.Errorf("scaling the number of replicas to %d: %w", replicas, err) } err = waitForDeploymentRollout(kubeClientSet, deployment) if err != nil { return err } } err = waitForPodsReady(kubeClientSet, DefaultTimeout, replicas, namespace, metav1.ListOptions{ LabelSelector: fields.SelectorFromSet(fields.Set(deployment.Spec.Template.ObjectMeta.Labels)).String(), }) if err != nil { return fmt.Errorf("waiting for nginx-ingress-controller replica count to be %d: %w", replicas, err) } return nil } func waitForDeploymentRollout(kubeClientSet kubernetes.Interface, resource *appsv1.Deployment) error { return wait.Poll(Poll, 5*time.Minute, func() (bool, error) { d, err := kubeClientSet.AppsV1().Deployments(resource.Namespace).Get(context.TODO(), resource.Name, metav1.GetOptions{}) if apierrors.IsNotFound(err) { return false, nil } if err != nil { return false, nil } if d.DeletionTimestamp != nil { return false, fmt.Errorf("deployment %q is being deleted", resource.Name) } if d.Generation <= d.Status.ObservedGeneration && d.Status.UpdatedReplicas == d.Status.Replicas && d.Status.UnavailableReplicas == 0 { return true, nil } return false, nil }) } // UpdateIngress runs the given updateFunc on the ingress func UpdateIngress(kubeClientSet kubernetes.Interface, namespace string, name string, updateFunc func(d *networking.Ingress) error) error { ingress, err := kubeClientSet.NetworkingV1().Ingresses(namespace).Get(context.TODO(), name, metav1.GetOptions{}) if err != nil { return err } if ingress == nil { return fmt.Errorf("there is no ingress with name %v in namespace %v", name, namespace) } if ingress.ObjectMeta.Annotations == nil { ingress.ObjectMeta.Annotations = map[string]string{} } if err := updateFunc(ingress); err != nil { return err } _, err = kubeClientSet.NetworkingV1().Ingresses(namespace).Update(context.TODO(), ingress, metav1.UpdateOptions{}) if err != nil { return err } Sleep(1 * time.Second) return nil } // NewSingleIngressWithTLS creates a simple ingress rule with TLS spec included func NewSingleIngressWithTLS(name, path, host string, tlsHosts []string, ns, service string, port int, annotations map[string]string) *networking.Ingress { return newSingleIngressWithRules(name, path, host, ns, service, port, annotations, tlsHosts) } // NewSingleIngress creates a simple ingress rule func NewSingleIngress(name, path, host, ns, service string, port int, annotations map[string]string) *networking.Ingress { return newSingleIngressWithRules(name, path, host, ns, service, port, annotations, nil) } func NewSingleIngressWithIngressClass(name, path, host, ns, service, ingressClass string, port int, annotations map[string]string) *networking.Ingress { ing := newSingleIngressWithRules(name, path, host, ns, service, port, annotations, nil) ing.Spec.IngressClassName = &ingressClass return ing } // NewSingleIngressWithMultiplePaths creates a simple ingress rule with multiple paths func NewSingleIngressWithMultiplePaths(name string, paths []string, host, ns, service string, port int, annotations map[string]string) *networking.Ingress { pathtype := networking.PathTypePrefix spec := networking.IngressSpec{ IngressClassName: GetIngressClassName(ns), Rules: []networking.IngressRule{ { Host: host, IngressRuleValue: networking.IngressRuleValue{ HTTP: &networking.HTTPIngressRuleValue{}, }, }, }, } for _, path := range paths { spec.Rules[0].IngressRuleValue.HTTP.Paths = append(spec.Rules[0].IngressRuleValue.HTTP.Paths, networking.HTTPIngressPath{ Path: path, PathType: &pathtype, Backend: networking.IngressBackend{ Service: &networking.IngressServiceBackend{ Name: service, Port: networking.ServiceBackendPort{ Number: int32(port), }, }, }, }) } return newSingleIngress(name, ns, annotations, spec) } func newSingleIngressWithRules(name, path, host, ns, service string, port int, annotations map[string]string, tlsHosts []string) *networking.Ingress { pathtype := networking.PathTypePrefix spec := networking.IngressSpec{ IngressClassName: GetIngressClassName(ns), Rules: []networking.IngressRule{ { IngressRuleValue: networking.IngressRuleValue{ HTTP: &networking.HTTPIngressRuleValue{ Paths: []networking.HTTPIngressPath{ { Path: path, PathType: &pathtype, Backend: networking.IngressBackend{ Service: &networking.IngressServiceBackend{ Name: service, Port: networking.ServiceBackendPort{ Number: int32(port), }, }, }, }, }, }, }, }, }, } // allow ingresses without host field if host != "" { spec.Rules[0].Host = host } if len(tlsHosts) > 0 { spec.TLS = []networking.IngressTLS{ { Hosts: tlsHosts, SecretName: host, }, } } return newSingleIngress(name, ns, annotations, spec) } // NewSingleIngressWithBackendAndRules creates an ingress with both a default backend and a rule func NewSingleIngressWithBackendAndRules(name, path, host, ns, defaultService string, defaultPort int, service string, port int, annotations map[string]string) *networking.Ingress { pathtype := networking.PathTypePrefix spec := networking.IngressSpec{ IngressClassName: GetIngressClassName(ns), DefaultBackend: &networking.IngressBackend{ Service: &networking.IngressServiceBackend{ Name: defaultService, Port: networking.ServiceBackendPort{ Number: int32(defaultPort), }, }, }, Rules: []networking.IngressRule{ { Host: host, IngressRuleValue: networking.IngressRuleValue{ HTTP: &networking.HTTPIngressRuleValue{ Paths: []networking.HTTPIngressPath{ { Path: path, PathType: &pathtype, Backend: networking.IngressBackend{ Service: &networking.IngressServiceBackend{ Name: service, Port: networking.ServiceBackendPort{ Number: int32(port), }, }, }, }, }, }, }, }, }, } return newSingleIngress(name, ns, annotations, spec) } // NewSingleCatchAllIngress creates a simple ingress with a catch-all backend func NewSingleCatchAllIngress(name, ns, service string, port int, annotations map[string]string) *networking.Ingress { spec := networking.IngressSpec{ IngressClassName: GetIngressClassName(ns), DefaultBackend: &networking.IngressBackend{ Service: &networking.IngressServiceBackend{ Name: service, Port: networking.ServiceBackendPort{ Number: int32(port), }, }, }, } return newSingleIngress(name, ns, annotations, spec) } func newSingleIngress(name, ns string, annotations map[string]string, spec networking.IngressSpec) *networking.Ingress { ing := &networking.Ingress{ ObjectMeta: metav1.ObjectMeta{ Name: name, Namespace: ns, }, Spec: spec, } if annotations == nil { annotations = make(map[string]string) } ing.SetAnnotations(annotations) return ing } // defaultWaitDuration default sleep time for operations related // to the API server and NGINX reloads. var defaultWaitDuration = 5 * time.Second // Sleep pauses the current goroutine for at least the duration d. // If no duration is defined, it uses a default func Sleep(duration ...time.Duration) { sleepFor := defaultWaitDuration if len(duration) != 0 { sleepFor = duration[0] } time.Sleep(sleepFor) } func loadConfig() (*restclient.Config, error) { config, err := rest.InClusterConfig() if err != nil { return nil, err } config.UserAgent = "ingress-nginx-e2e" return config, nil }