Fix Graphd Autofailover When Autoscaler Scale Up and Then Down (#552)

* Fix stacktrace and nebula autoscalar stuck during failover

* remove unnecessary comment

Author: kevinliu24

apis/apps/v1alpha1/nebulacluster_common.go

@@ -27,6 +27,7 @@ import (
 	kruisev1beta1 "github.com/openkruise/kruise-api/apps/v1beta1"
 	appsv1 "k8s.io/api/apps/v1"
 	corev1 "k8s.io/api/core/v1"
+	apiresource "k8s.io/apimachinery/pkg/api/resource"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/apis/meta/v1/unstructured"
 	"k8s.io/apimachinery/pkg/runtime"

pkg/controller/component/graphd_cluster.go

@@ -268,9 +268,7 @@ func (c *graphdCluster) syncNebulaClusterStatus(
 		nc.Status.Graphd.Phase = v1alpha1.RunningPhase
 	}
 
-	workloadReplicas := getWorkloadReplicas(nc.Status.Graphd.Workload)
-	if !nc.IsAutoFailoverEnabled() ||
-		pointer.Int32Deref(nc.Spec.Graphd.Replicas, 0) != workloadReplicas {
+	if !nc.IsAutoFailoverEnabled() {
 		return syncComponentStatus(nc.GraphdComponent(), &nc.Status.Graphd, oldWorkload)
 	}
 

pkg/controller/component/graphd_failover.go

@@ -105,11 +105,16 @@ func (g *graphdFailover) toleratePods(nc *v1alpha1.NebulaCluster) ([]string, err
 		if fh.PodRebuilt {
 			continue
 		}
+
 		pod, err := g.clientSet.Pod().GetPod(nc.Namespace, podName)
 		if err != nil && !apierrors.IsNotFound(err) {
 			return nil, err
 		}
-		if pod != nil && isPodTerminating(pod) {
+		if pod == nil {
+			continue
+		}
+
+		if isPodTerminating(pod) {
 			return nil, utilerrors.ReconcileErrorf("failure graphd pod [%s/%s] is deleting", nc.Namespace, podName)
 		}
 		if isPodHealthy(pod) {