Fixed node being taken repeatedly causing queue block

pkg/kwok/controllers/node_lease_controller.go

@@ -182,12 +182,15 @@ func (c *NodeLeaseController) sync(ctx context.Context, nodeName string) (lease
 			return nil, nil
 		}
 		logger.Info("Syncing lease")
-		lease, err := c.renewLease(ctx, lease)
+		lease, transitions, err := c.renewLease(ctx, lease)
 		if err != nil {
 			return nil, fmt.Errorf("failed to update lease using lease: %w", err)
 		}
 
-		c.onNodeManaged(nodeName)
+		// Only the lease have been transitions and then retry to manage the node.
+		if transitions {
+			c.onNodeManaged(nodeName)
+		}
 		return lease, nil
 	}
 
@@ -249,7 +252,7 @@ func (c *NodeLeaseController) ensureLease(ctx context.Context, leaseName string)
 }
 
 // renewLease attempts to update the lease for maxUpdateRetries, call this once you're sure the lease has been created
-func (c *NodeLeaseController) renewLease(ctx context.Context, base *coordinationv1.Lease) (*coordinationv1.Lease, error) {
+func (c *NodeLeaseController) renewLease(ctx context.Context, base *coordinationv1.Lease) (*coordinationv1.Lease, bool, error) {
 	lease := base.DeepCopy()
 
 	transitions := format.ElemOrDefault(lease.Spec.HolderIdentity) != c.holderIdentity
@@ -263,15 +266,15 @@ func (c *NodeLeaseController) renewLease(ctx context.Context, base *coordination
 	if c.mutateLeaseFunc != nil {
 		err := c.mutateLeaseFunc(lease)
 		if err != nil {
-			return nil, err
+			return nil, false, err
 		}
 	}
 
 	lease, err := c.typedClient.CoordinationV1().Leases(lease.Namespace).Update(ctx, lease, metav1.UpdateOptions{})
 	if err != nil {
-		return nil, err
+		return nil, false, err
 	}
-	return lease, nil
+	return lease, transitions, nil
 }
 
 // setNodeOwnerFunc helps construct a mutateLeaseFunc which sets a node OwnerReference to the given lease object