Add --cluster-status-update-frequency flag for karmada-agent (#272)

Signed-off-by: RainbowMango <renhongcai@huawei.com>

artifacts/agent/karmada-agent.yaml

@@ -27,6 +27,8 @@ spec:
             - /bin/karmada-agent
             - --karmada-kubeconfig=/etc/kubeconfig/karmada-kubeconfig
             - --cluster-name={{member_cluster_name}}
+            - --cluster-status-update-frequency=10s
+            - --v=4
           volumeMounts:
             - name: kubeconfig
               mountPath: /etc/kubeconfig

cmd/agent/app/agent.go

@@ -74,7 +74,7 @@ func run(opts *options.Options, stopChan <-chan struct{}) error {
 		return err
 	}
 
-	setupControllers(controllerManager, controlPlaneRestConfig, opts.ClusterName, stopChan)
+	setupControllers(controllerManager, opts, stopChan)
 
 	// blocks until the stop channel is closed.
 	if err := controllerManager.Start(stopChan); err != nil {
@@ -85,16 +85,16 @@ func run(opts *options.Options, stopChan <-chan struct{}) error {
 	return nil
 }
 
-func setupControllers(mgr controllerruntime.Manager, controlPlaneRestConfig *restclient.Config, clusterName string, stopChan <-chan struct{}) {
+func setupControllers(mgr controllerruntime.Manager, opts *options.Options, stopChan <-chan struct{}) {
 	predicateFun := predicate.Funcs{
 		CreateFunc: func(createEvent event.CreateEvent) bool {
-			return createEvent.Meta.GetName() == clusterName
+			return createEvent.Meta.GetName() == opts.ClusterName
 		},
 		UpdateFunc: func(updateEvent event.UpdateEvent) bool {
-			return updateEvent.MetaOld.GetName() == clusterName
+			return updateEvent.MetaOld.GetName() == opts.ClusterName
 		},
 		DeleteFunc: func(deleteEvent event.DeleteEvent) bool {
-			return deleteEvent.Meta.GetName() == clusterName
+			return deleteEvent.Meta.GetName() == opts.ClusterName
 		},
 		GenericFunc: func(genericEvent event.GenericEvent) bool {
 			return false
@@ -102,10 +102,11 @@ func setupControllers(mgr controllerruntime.Manager, controlPlaneRestConfig *res
 	}
 
 	clusterStatusController := &status.ClusterStatusController{
-		Client:               mgr.GetClient(),
-		EventRecorder:        mgr.GetEventRecorderFor(status.ControllerName),
-		PredicateFunc:        predicateFun,
-		ClusterClientSetFunc: util.NewClusterClientSetForAgent,
+		Client:                       mgr.GetClient(),
+		EventRecorder:                mgr.GetEventRecorderFor(status.ControllerName),
+		PredicateFunc:                predicateFun,
+		ClusterClientSetFunc:         util.NewClusterClientSetForAgent,
+		ClusterStatusUpdateFrequency: opts.ClusterStatusUpdateFrequency,
 	}
 	if err := clusterStatusController.SetupWithManager(mgr); err != nil {
 		klog.Fatalf("Failed to setup cluster status controller: %v", err)

cmd/agent/app/options/options.go

@@ -1,13 +1,23 @@
 package options
 
 import (
+	"time"
+
 	"github.com/spf13/pflag"
+
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 )
 
 // Options contains everything necessary to create and run controller-manager.
 type Options struct {
 	KarmadaKubeConfig string
 	ClusterName       string
+
+	// ClusterStatusUpdateFrequency is the frequency that karmada-agent computes cluster status.
+	// If cluster lease feature is not enabled, it is also the frequency that karmada-agent posts cluster status
+	// to karmada-apiserver. In that case, be cautious when changing the constant, it must work with
+	// ClusterMonitorGracePeriod(--cluster-monitor-grace-period) in karmada-controller-manager.
+	ClusterStatusUpdateFrequency metav1.Duration
 }
 
 // NewOptions builds an default scheduler options.
@@ -23,4 +33,5 @@ func (o *Options) AddFlags(fs *pflag.FlagSet) {
 
 	fs.StringVar(&o.KarmadaKubeConfig, "karmada-kubeconfig", o.KarmadaKubeConfig, "Path to karmada kubeconfig.")
 	fs.StringVar(&o.ClusterName, "cluster-name", o.ClusterName, "Name of member cluster that the agent serves for.")
+	fs.DurationVar(&o.ClusterStatusUpdateFrequency.Duration, "cluster-status-update-frequency", 10*time.Second, "Specifies how often karmada-agent posts cluster status to karmada-apiserver. Note: be cautious when changing the constant, it must work with ClusterMonitorGracePeriod in karmada-controller-manager.")
 }

cmd/controller-manager/app/controllermanager.go

@@ -7,6 +7,7 @@ import (
 	"time"
 
 	"github.com/spf13/cobra"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/client-go/discovery"
 	"k8s.io/client-go/dynamic"
 	"k8s.io/component-base/logs"
@@ -146,6 +147,8 @@ func setupControllers(mgr controllerruntime.Manager, stopChan <-chan struct{}) {
 		EventRecorder:        mgr.GetEventRecorderFor(status.ControllerName),
 		PredicateFunc:        clusterPredicateFunc,
 		ClusterClientSetFunc: util.NewClusterClientSet,
+		// TODO(RainbowMango): Temporarily hard code the duration until we add flags for this.
+		ClusterStatusUpdateFrequency: metav1.Duration{Duration: 10 * time.Second},
 	}
 	if err := clusterStatusController.SetupWithManager(mgr); err != nil {
 		klog.Fatalf("Failed to setup cluster status controller: %v", err)

pkg/controllers/status/cluster_status_controller.go

@@ -3,13 +3,12 @@ package status
 import (
 	"context"
 	"net/http"
-	"time"
 
 	corev1 "k8s.io/api/core/v1"
 	"k8s.io/apimachinery/pkg/api/equality"
 	"k8s.io/apimachinery/pkg/api/errors"
 	"k8s.io/apimachinery/pkg/api/resource"
-	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/tools/record"
 	"k8s.io/klog/v2"
@@ -39,6 +38,11 @@ type ClusterStatusController struct {
 	EventRecorder        record.EventRecorder
 	PredicateFunc        predicate.Predicate
 	ClusterClientSetFunc func(c *v1alpha1.Cluster, client client.Client) (*util.ClusterClient, error)
+
+	// ClusterStatusUpdateFrequency is the frequency that controller computes cluster status.
+	// If cluster lease feature is not enabled, it is also the frequency that controller posts cluster status
+	// to karmada-apiserver.
+	ClusterStatusUpdateFrequency metav1.Duration
 }
 
 // Reconcile syncs status of the given member cluster.
@@ -137,7 +141,7 @@ func (c *ClusterStatusController) updateStatusIfNeeded(cluster *v1alpha1.Cluster
 		}
 	}
 
-	return controllerruntime.Result{RequeueAfter: 10 * time.Second}, nil
+	return controllerruntime.Result{RequeueAfter: c.ClusterStatusUpdateFrequency.Duration}, nil
 }
 
 func getClusterHealthStatus(clusterClient *util.ClusterClient) (online, healthy bool) {
@@ -166,29 +170,29 @@ func healthEndpointCheck(client *kubernetes.Clientset, path string) (int, error)
 	return healthStatus, resp.Error()
 }
 
-func generateReadyCondition(online, healthy bool) []v1.Condition {
-	var conditions []v1.Condition
-	currentTime := v1.Now()
+func generateReadyCondition(online, healthy bool) []metav1.Condition {
+	var conditions []metav1.Condition
+	currentTime := metav1.Now()
 
-	newClusterOfflineCondition := v1.Condition{
+	newClusterOfflineCondition := metav1.Condition{
 		Type:               v1alpha1.ClusterConditionReady,
-		Status:             v1.ConditionFalse,
+		Status:             metav1.ConditionFalse,
 		Reason:             clusterNotReachableReason,
 		Message:            clusterNotReachableMsg,
 		LastTransitionTime: currentTime,
 	}
 
-	newClusterReadyCondition := v1.Condition{
+	newClusterReadyCondition := metav1.Condition{
 		Type:               v1alpha1.ClusterConditionReady,
-		Status:             v1.ConditionTrue,
+		Status:             metav1.ConditionTrue,
 		Reason:             clusterReady,
 		Message:            clusterHealthy,
 		LastTransitionTime: currentTime,
 	}
 
-	newClusterNotReadyCondition := v1.Condition{
+	newClusterNotReadyCondition := metav1.Condition{
 		Type:               v1alpha1.ClusterConditionReady,
-		Status:             v1.ConditionFalse,
+		Status:             metav1.ConditionFalse,
 		Reason:             clusterNotReady,
 		Message:            clusterUnhealthy,
 		LastTransitionTime: currentTime,
@@ -248,7 +252,7 @@ func getAPIEnablements(clusterClient *util.ClusterClient) ([]v1alpha1.APIEnablem
 
 func getNodeSummary(clusterClient *util.ClusterClient) (v1alpha1.NodeSummary, error) {
 	var nodeSummary = v1alpha1.NodeSummary{}
-	nodeList, err := clusterClient.KubeClient.CoreV1().Nodes().List(context.TODO(), v1.ListOptions{})
+	nodeList, err := clusterClient.KubeClient.CoreV1().Nodes().List(context.TODO(), metav1.ListOptions{})
 	if err != nil {
 		return nodeSummary, err
 	}
@@ -264,7 +268,7 @@ func getNodeSummary(clusterClient *util.ClusterClient) (v1alpha1.NodeSummary, er
 
 	allocatable := getClusterAllocatable(nodeList)
 
-	podList, err := clusterClient.KubeClient.CoreV1().Pods("").List(context.TODO(), v1.ListOptions{})
+	podList, err := clusterClient.KubeClient.CoreV1().Pods("").List(context.TODO(), metav1.ListOptions{})
 	if err != nil {
 		return nodeSummary, err
 	}