Fix race in setting nominated node

pkg/scheduler/core/generic_scheduler.go

@@ -351,7 +351,7 @@ func (g *genericScheduler) processPreemptionWithExtenders(
 // worth the complexity, especially because we generally expect to have a very
 // small number of nominated pods per node.
 func (g *genericScheduler) getLowerPriorityNominatedPods(pod *v1.Pod, nodeName string) []*v1.Pod {
-	pods := g.schedulingQueue.WaitingPodsForNode(nodeName)
+	pods := g.schedulingQueue.NominatedPodsForNode(nodeName)
 
 	if len(pods) == 0 {
 		return nil
@@ -501,7 +501,7 @@ func addNominatedPods(pod *v1.Pod, meta algorithm.PredicateMetadata,
 		// This may happen only in tests.
 		return false, meta, nodeInfo
 	}
-	nominatedPods := queue.WaitingPodsForNode(nodeInfo.Node().Name)
+	nominatedPods := queue.NominatedPodsForNode(nodeInfo.Node().Name)
 	if nominatedPods == nil || len(nominatedPods) == 0 {
 		return false, meta, nodeInfo
 	}

pkg/scheduler/scheduler.go

@@ -328,11 +328,19 @@ func (sched *Scheduler) preempt(preemptor *v1.Pod, scheduleErr error) (string, e
 	var nodeName = ""
 	if node != nil {
 		nodeName = node.Name
+		// Update the scheduling queue with the nominated pod information. Without
+		// this, there would be a race condition between the next scheduling cycle
+		// and the time the scheduler receives a Pod Update for the nominated pod.
+		sched.config.SchedulingQueue.UpdateNominatedPodForNode(preemptor, nodeName)
+
+		// Make a call to update nominated node name of the pod on the API server.
 		err = sched.config.PodPreemptor.SetNominatedNodeName(preemptor, nodeName)
 		if err != nil {
 			klog.Errorf("Error in preemption process. Cannot update pod %v/%v annotations: %v", preemptor.Namespace, preemptor.Name, err)
+			sched.config.SchedulingQueue.DeleteNominatedPodIfExists(preemptor)
 			return "", err
 		}
+
 		for _, victim := range victims {
 			if err := sched.config.PodPreemptor.DeletePod(victim); err != nil {
 				klog.Errorf("Error preempting pod %v/%v: %v", victim.Namespace, victim.Name, err)