Pop from the backoffQ when the activeQ is empty

pkg/scheduler/backend/queue/active_queue.go

@@ -113,14 +113,22 @@ func (uaq *unlockedActiveQueue) has(pInfo *framework.QueuedPodInfo) bool {
 	return uaq.queue.Has(pInfo)
 }
 
+// backoffQPopper defines method that is used to pop from the backoffQ when the activeQ is empty.
+type backoffQPopper interface {
+	// popBackoff pops the pInfo from the podBackoffQ.
+	popBackoff() (*framework.QueuedPodInfo, error)
+	// len returns length of the podBackoffQ queue.
+	lenBackoff() int
+}
+
 // activeQueue implements activeQueuer. All of the fields have to be protected using the lock.
 type activeQueue struct {
 	// lock synchronizes all operations related to activeQ.
 	// It protects activeQ, inFlightPods, inFlightEvents, schedulingCycle and closed fields.
 	// Caution: DO NOT take "SchedulingQueue.lock" after taking "lock".
 	// You should always take "SchedulingQueue.lock" first, otherwise the queue could end up in deadlock.
-	// "lock" should not be taken after taking "nLock".
-	// Correct locking order is: SchedulingQueue.lock > lock > nominator.nLock.
+	// "lock" should not be taken after taking "backoffQueue.lock" or "nominator.nLock".
+	// Correct locking order is: SchedulingQueue.lock > lock > backoffQueue.lock > nominator.nLock.
 	lock sync.RWMutex
 
 	// activeQ is heap structure that scheduler actively looks at to find pods to
@@ -132,6 +140,8 @@ type activeQueue struct {
 	unlockedQueue *unlockedActiveQueue
 
 	// cond is a condition that is notified when the pod is added to activeQ.
+	// When SchedulerPopFromBackoffQ feature is enabled,
+	// condition is also notified when the pod is added to backoffQ.
 	// It is used with lock.
 	cond sync.Cond
 
@@ -171,9 +181,13 @@ type activeQueue struct {
 	isSchedulingQueueHintEnabled bool
 
 	metricsRecorder metrics.MetricAsyncRecorder
+
+	// backoffQPopper is used to pop from backoffQ when activeQ is empty.
+	// It is non-nil only when SchedulerPopFromBackoffQ feature is enabled.
+	backoffQPopper backoffQPopper
 }
 
-func newActiveQueue(queue *heap.Heap[*framework.QueuedPodInfo], isSchedulingQueueHintEnabled bool, metricRecorder metrics.MetricAsyncRecorder) *activeQueue {
+func newActiveQueue(queue *heap.Heap[*framework.QueuedPodInfo], isSchedulingQueueHintEnabled bool, metricRecorder metrics.MetricAsyncRecorder, backoffQPopper backoffQPopper) *activeQueue {
 	aq := &activeQueue{
 		queue:                        queue,
 		inFlightPods:                 make(map[types.UID]*list.Element),
@@ -181,6 +195,7 @@ func newActiveQueue(queue *heap.Heap[*framework.QueuedPodInfo], isSchedulingQueu
 		isSchedulingQueueHintEnabled: isSchedulingQueueHintEnabled,
 		metricsRecorder:              metricRecorder,
 		unlockedQueue:                newUnlockedActiveQueue(queue),
+		backoffQPopper:               backoffQPopper,
 	}
 	aq.cond.L = &aq.lock
 
@@ -238,7 +253,13 @@ func (aq *activeQueue) pop(logger klog.Logger) (*framework.QueuedPodInfo, error)
 }
 
 func (aq *activeQueue) unlockedPop(logger klog.Logger) (*framework.QueuedPodInfo, error) {
+	var pInfo *framework.QueuedPodInfo
 	for aq.queue.Len() == 0 {
+		// backoffQPopper is non-nil only if SchedulerPopFromBackoffQ feature is enabled.
+		// In case of non-empty backoffQ, try popping from there.
+		if aq.backoffQPopper != nil && aq.backoffQPopper.lenBackoff() != 0 {
+			break
+		}
 		// When the queue is empty, invocation of Pop() is blocked until new item is enqueued.
 		// When Close() is called, the p.closed is set and the condition is broadcast,
 		// which causes this loop to continue and return from the Pop().
@@ -250,7 +271,15 @@ func (aq *activeQueue) unlockedPop(logger klog.Logger) (*framework.QueuedPodInfo
 	}
 	pInfo, err := aq.queue.Pop()
 	if err != nil {
-		return nil, err
+		if aq.backoffQPopper == nil {
+			return nil, err
+		}
+		// Try to pop from backoffQ when activeQ is empty.
+		pInfo, err = aq.backoffQPopper.popBackoff()
+		if err != nil {
+			return nil, err
+		}
+		metrics.SchedulerQueueIncomingPods.WithLabelValues("active", framework.PopFromBackoffQ).Inc()
 	}
 	pInfo.Attempts++
 	pInfo.BackoffExpiration = time.Time{}

pkg/scheduler/backend/queue/active_queue_test.go

@@ -30,7 +30,7 @@ import (
 func TestClose(t *testing.T) {
 	logger, ctx := ktesting.NewTestContext(t)
 	rr := metrics.NewMetricsAsyncRecorder(10, time.Second, ctx.Done())
-	aq := newActiveQueue(heap.NewWithRecorder(podInfoKeyFunc, heap.LessFunc[*framework.QueuedPodInfo](newDefaultQueueSort()), metrics.NewActivePodsRecorder()), true, *rr)
+	aq := newActiveQueue(heap.NewWithRecorder(podInfoKeyFunc, heap.LessFunc[*framework.QueuedPodInfo](newDefaultQueueSort()), metrics.NewActivePodsRecorder()), true, *rr, nil)
 
 	aq.underLock(func(unlockedActiveQ unlockedActiveQueuer) {
 		unlockedActiveQ.add(&framework.QueuedPodInfo{PodInfo: &framework.PodInfo{Pod: st.MakePod().Namespace("foo").Name("p1").UID("p1").Obj()}}, framework.EventUnscheduledPodAdd.Label())

pkg/scheduler/backend/queue/backoff_queue.go

@@ -17,6 +17,7 @@ limitations under the License.
 package queue
 
 import (
+	"sync"
 	"time"
 
 	v1 "k8s.io/api/core/v1"
@@ -35,13 +36,14 @@ import (
 const backoffQOrderingWindowDuration = time.Second
 
 // backoffQueuer is a wrapper for backoffQ related operations.
+// Its methods that relies on the queues, take the lock inside.
 type backoffQueuer interface {
 	// isPodBackingoff returns true if a pod is still waiting for its backoff timer.
 	// If this returns true, the pod should not be re-tried.
 	// If the pod backoff time is in the actual ordering window, it should still be backing off.
 	isPodBackingoff(podInfo *framework.QueuedPodInfo) bool
-	// popEachBackoffCompleted run fn for all pods from podBackoffQ and podErrorBackoffQ that completed backoff while popping them.
-	popEachBackoffCompleted(logger klog.Logger, fn func(pInfo *framework.QueuedPodInfo))
+	// popAllBackoffCompleted pops all pods from podBackoffQ and podErrorBackoffQ that completed backoff.
+	popAllBackoffCompleted(logger klog.Logger) []*framework.QueuedPodInfo
 
 	// podInitialBackoffDuration returns initial backoff duration that pod can get.
 	podInitialBackoffDuration() time.Duration
@@ -61,7 +63,8 @@ type backoffQueuer interface {
 	// It returns new pod info if updated, nil otherwise.
 	update(newPod *v1.Pod, oldPodInfo *framework.QueuedPodInfo) *framework.QueuedPodInfo
 	// delete deletes the pInfo from backoffQueue.
-	delete(pInfo *framework.QueuedPodInfo)
+	// It returns true if the pod was deleted.
+	delete(pInfo *framework.QueuedPodInfo) bool
 	// get returns the pInfo matching given pInfoLookup, if exists.
 	get(pInfoLookup *framework.QueuedPodInfo) (*framework.QueuedPodInfo, bool)
 	// has inform if pInfo exists in the queue.
@@ -75,6 +78,14 @@ type backoffQueuer interface {
 // backoffQueue implements backoffQueuer and wraps two queues inside,
 // providing seamless access as if it were one queue.
 type backoffQueue struct {
+	// lock synchronizes all operations related to backoffQ.
+	// It protects both podBackoffQ and podErrorBackoffQ.
+	// Caution: DO NOT take "SchedulingQueue.lock" or "activeQueue.lock" after taking "lock".
+	// You should always take "SchedulingQueue.lock" and "activeQueue.lock" first, otherwise the queue could end up in deadlock.
+	// "lock" should not be taken after taking "nominator.nLock".
+	// Correct locking order is: SchedulingQueue.lock > activeQueue.lock > lock > nominator.nLock.
+	lock sync.RWMutex
+
 	clock clock.WithTicker
 
 	// podBackoffQ is a heap ordered by backoff expiry. Pods which have completed backoff
@@ -239,7 +250,8 @@ func (bq *backoffQueue) calculateBackoffDuration(podInfo *framework.QueuedPodInf
 	return duration
 }
 
-func (bq *backoffQueue) popEachBackoffCompletedWithQueue(logger klog.Logger, fn func(pInfo *framework.QueuedPodInfo), queue *heap.Heap[*framework.QueuedPodInfo]) {
+func (bq *backoffQueue) popAllBackoffCompletedWithQueue(logger klog.Logger, queue *heap.Heap[*framework.QueuedPodInfo]) []*framework.QueuedPodInfo {
+	var poppedPods []*framework.QueuedPodInfo
 	for {
 		pInfo, ok := queue.Peek()
 		if !ok || pInfo == nil {
@@ -254,23 +266,27 @@ func (bq *backoffQueue) popEachBackoffCompletedWithQueue(logger klog.Logger, fn
 			logger.Error(err, "Unable to pop pod from backoff queue despite backoff completion", "pod", klog.KObj(pod))
 			break
 		}
-		if fn != nil {
-			fn(pInfo)
-		}
+		poppedPods = append(poppedPods, pInfo)
 	}
+	return poppedPods
 }
 
-// popEachBackoffCompleted run fn for all pods from podBackoffQ and podErrorBackoffQ that completed backoff while popping them.
-func (bq *backoffQueue) popEachBackoffCompleted(logger klog.Logger, fn func(pInfo *framework.QueuedPodInfo)) {
+// popAllBackoffCompleted pops all pods from podBackoffQ and podErrorBackoffQ that completed backoff.
+func (bq *backoffQueue) popAllBackoffCompleted(logger klog.Logger) []*framework.QueuedPodInfo {
+	bq.lock.Lock()
+	defer bq.lock.Unlock()
+
 	// Ensure both queues are called
-	bq.popEachBackoffCompletedWithQueue(logger, fn, bq.podBackoffQ)
-	bq.popEachBackoffCompletedWithQueue(logger, fn, bq.podErrorBackoffQ)
+	return append(bq.popAllBackoffCompletedWithQueue(logger, bq.podBackoffQ), bq.popAllBackoffCompletedWithQueue(logger, bq.podErrorBackoffQ)...)
 }
 
 // add adds the pInfo to backoffQueue.
 // The event should show which event triggered this addition and is used for the metric recording.
 // It also ensures that pInfo is not in both queues.
 func (bq *backoffQueue) add(logger klog.Logger, pInfo *framework.QueuedPodInfo, event string) {
+	bq.lock.Lock()
+	defer bq.lock.Unlock()
+
 	// If pod has empty both unschedulable plugins and pending plugins,
 	// it means that it failed because of error and should be moved to podErrorBackoffQ.
 	if pInfo.UnschedulablePlugins.Len() == 0 && pInfo.PendingPlugins.Len() == 0 {
@@ -297,6 +313,9 @@ func (bq *backoffQueue) add(logger klog.Logger, pInfo *framework.QueuedPodInfo,
 // update updates the pod in backoffQueue if oldPodInfo is already in the queue.
 // It returns new pod info if updated, nil otherwise.
 func (bq *backoffQueue) update(newPod *v1.Pod, oldPodInfo *framework.QueuedPodInfo) *framework.QueuedPodInfo {
+	bq.lock.Lock()
+	defer bq.lock.Unlock()
+
 	// If the pod is in the backoff queue, update it there.
 	if pInfo, exists := bq.podBackoffQ.Get(oldPodInfo); exists {
 		_ = pInfo.Update(newPod)
@@ -313,13 +332,32 @@ func (bq *backoffQueue) update(newPod *v1.Pod, oldPodInfo *framework.QueuedPodIn
 }
 
 // delete deletes the pInfo from backoffQueue.
-func (bq *backoffQueue) delete(pInfo *framework.QueuedPodInfo) {
-	_ = bq.podBackoffQ.Delete(pInfo)
-	_ = bq.podErrorBackoffQ.Delete(pInfo)
+// It returns true if the pod was deleted.
+func (bq *backoffQueue) delete(pInfo *framework.QueuedPodInfo) bool {
+	bq.lock.Lock()
+	defer bq.lock.Unlock()
+
+	if bq.podBackoffQ.Delete(pInfo) == nil {
+		return true
+	}
+	return bq.podErrorBackoffQ.Delete(pInfo) == nil
+}
+
+// popBackoff pops the pInfo from the podBackoffQ.
+// It returns error if the queue is empty.
+// This doesn't pop the pods from the podErrorBackoffQ.
+func (bq *backoffQueue) popBackoff() (*framework.QueuedPodInfo, error) {
+	bq.lock.Lock()
+	defer bq.lock.Unlock()
+
+	return bq.podBackoffQ.Pop()
 }
 
 // get returns the pInfo matching given pInfoLookup, if exists.
 func (bq *backoffQueue) get(pInfoLookup *framework.QueuedPodInfo) (*framework.QueuedPodInfo, bool) {
+	bq.lock.RLock()
+	defer bq.lock.RUnlock()
+
 	pInfo, exists := bq.podBackoffQ.Get(pInfoLookup)
 	if exists {
 		return pInfo, true
@@ -329,11 +367,17 @@ func (bq *backoffQueue) get(pInfoLookup *framework.QueuedPodInfo) (*framework.Qu
 
 // has inform if pInfo exists in the queue.
 func (bq *backoffQueue) has(pInfo *framework.QueuedPodInfo) bool {
+	bq.lock.RLock()
+	defer bq.lock.RUnlock()
+
 	return bq.podBackoffQ.Has(pInfo) || bq.podErrorBackoffQ.Has(pInfo)
 }
 
 // list returns all pods that are in the queue.
 func (bq *backoffQueue) list() []*v1.Pod {
+	bq.lock.RLock()
+	defer bq.lock.RUnlock()
+
 	var result []*v1.Pod
 	for _, pInfo := range bq.podBackoffQ.List() {
 		result = append(result, pInfo.Pod)
@@ -346,5 +390,16 @@ func (bq *backoffQueue) list() []*v1.Pod {
 
 // len returns length of the queue.
 func (bq *backoffQueue) len() int {
+	bq.lock.RLock()
+	defer bq.lock.RUnlock()
+
 	return bq.podBackoffQ.Len() + bq.podErrorBackoffQ.Len()
 }
+
+// lenBackoff returns length of the podBackoffQ.
+func (bq *backoffQueue) lenBackoff() int {
+	bq.lock.RLock()
+	defer bq.lock.RUnlock()
+
+	return bq.podBackoffQ.Len()
+}

pkg/scheduler/backend/queue/backoff_queue_test.go

@@ -78,7 +78,7 @@ func TestBackoffQueue_calculateBackoffDuration(t *testing.T) {
 	}
 }
 
-func TestBackoffQueue_popEachBackoffCompleted(t *testing.T) {
+func TestBackoffQueue_popAllBackoffCompleted(t *testing.T) {
 	fakeClock := testingclock.NewFakeClock(time.Now())
 	podInfos := map[string]*framework.QueuedPodInfo{
 		"pod0": {
@@ -156,10 +156,11 @@ func TestBackoffQueue_popEachBackoffCompleted(t *testing.T) {
 				for _, podName := range tt.podsInBackoff {
 					bq.add(logger, podInfos[podName], framework.EventUnscheduledPodAdd.Label())
 				}
+				gotPodInfos := bq.popAllBackoffCompleted(logger)
 				var gotPods []string
-				bq.popEachBackoffCompleted(logger, func(pInfo *framework.QueuedPodInfo) {
+				for _, pInfo := range gotPodInfos {
 					gotPods = append(gotPods, pInfo.Pod.Name)
-				})
+				}
 				if diff := cmp.Diff(tt.wantPods, gotPods); diff != "" {
 					t.Errorf("Unexpected pods moved (-want, +got):\n%s", diff)
 				}
@@ -248,10 +249,11 @@ func TestBackoffQueueOrdering(t *testing.T) {
 			for _, podInfo := range podInfos {
 				bq.add(logger, podInfo, framework.EventUnscheduledPodAdd.Label())
 			}
+			gotPodInfos := bq.popAllBackoffCompleted(logger)
 			var gotPods []string
-			bq.popEachBackoffCompleted(logger, func(pInfo *framework.QueuedPodInfo) {
+			for _, pInfo := range gotPodInfos {
 				gotPods = append(gotPods, pInfo.Pod.Name)
-			})
+			}
 			if diff := cmp.Diff(tt.wantPods, gotPods); diff != "" {
 				t.Errorf("Unexpected pods moved (-want, +got):\n%s", diff)
 			}

pkg/scheduler/backend/queue/nominator.go

@@ -35,10 +35,10 @@ import (
 type nominator struct {
 	// nLock synchronizes all operations related to nominator.
 	// It should not be used anywhere else.
-	// Caution: DO NOT take ("SchedulingQueue.lock" or "activeQueue.lock") after taking "nLock".
-	// You should always take "SchedulingQueue.lock" and "activeQueue.lock" first,
+	// Caution: DO NOT take ("SchedulingQueue.lock" or "activeQueue.lock" or "backoffQueue.lock") after taking "nLock".
+	// You should always take "SchedulingQueue.lock" and "activeQueue.lock" and "backoffQueue.lock" first,
 	// otherwise the nominator could end up in deadlock.
-	// Correct locking order is: SchedulingQueue.lock > activeQueue.lock > nLock.
+	// Correct locking order is: SchedulingQueue.lock > activeQueue.lock = backoffQueue.lock > nLock.
 	nLock sync.RWMutex
 
 	// podLister is used to verify if the given pod is alive.

pkg/scheduler/backend/queue/scheduling_queue.go

@@ -160,8 +160,8 @@ type PriorityQueue struct {
 	clock clock.WithTicker
 
 	// lock takes precedence and should be taken first,
-	// before any other locks in the queue (activeQueue.lock or nominator.nLock).
-	// Correct locking order is: lock > activeQueue.lock > nominator.nLock.
+	// before any other locks in the queue (activeQueue.lock or backoffQueue.lock or nominator.nLock).
+	// Correct locking order is: lock > activeQueue.lock > backoffQueue.lock > nominator.nLock.
 	lock sync.RWMutex
 
 	// the maximum time a pod can stay in the unschedulablePods.
@@ -331,12 +331,12 @@ func NewPriorityQueue(
 	isSchedulingQueueHintEnabled := utilfeature.DefaultFeatureGate.Enabled(features.SchedulerQueueingHints)
 	isPopFromBackoffQEnabled := utilfeature.DefaultFeatureGate.Enabled(features.SchedulerPopFromBackoffQ)
 
+	backoffQ := newBackoffQueue(options.clock, options.podInitialBackoffDuration, options.podMaxBackoffDuration, lessFn, isPopFromBackoffQEnabled)
 	pq := &PriorityQueue{
 		clock:                             options.clock,
 		stop:                              make(chan struct{}),
 		podMaxInUnschedulablePodsDuration: options.podMaxInUnschedulablePodsDuration,
-		activeQ:                           newActiveQueue(heap.NewWithRecorder(podInfoKeyFunc, heap.LessFunc[*framework.QueuedPodInfo](lessFn), metrics.NewActivePodsRecorder()), isSchedulingQueueHintEnabled, options.metricsRecorder),
-		backoffQ:                          newBackoffQueue(options.clock, options.podInitialBackoffDuration, options.podMaxBackoffDuration, lessFn, isPopFromBackoffQEnabled),
+		backoffQ:                          backoffQ,
 		unschedulablePods:                 newUnschedulablePods(metrics.NewUnschedulablePodsRecorder(), metrics.NewGatedPodsRecorder()),
 		preEnqueuePluginMap:               options.preEnqueuePluginMap,
 		queueingHintMap:                   options.queueingHintMap,
@@ -346,6 +346,11 @@ func NewPriorityQueue(
 		isSchedulingQueueHintEnabled:      isSchedulingQueueHintEnabled,
 		isPopFromBackoffQEnabled:          isPopFromBackoffQEnabled,
 	}
+	var backoffQPopper backoffQPopper
+	if isPopFromBackoffQEnabled {
+		backoffQPopper = backoffQ
+	}
+	pq.activeQ = newActiveQueue(heap.NewWithRecorder(podInfoKeyFunc, heap.LessFunc[*framework.QueuedPodInfo](lessFn), metrics.NewActivePodsRecorder()), isSchedulingQueueHintEnabled, options.metricsRecorder, backoffQPopper)
 	pq.nsLister = informerFactory.Core().V1().Namespaces().Lister()
 	pq.nominator = newPodNominator(options.podLister)
 
@@ -672,6 +677,12 @@ func (p *PriorityQueue) activate(logger klog.Logger, pod *v1.Pod) bool {
 		if !exists {
 			return false
 		}
+		// Delete pod from the backoffQ now to make sure it won't be popped from the backoffQ
+		// just before moving it to the activeQ
+		if deleted := p.backoffQ.delete(pInfo); !deleted {
+			// Pod was popped from the backoffQ in the meantime. Don't activate it.
+			return false
+		}
 	}
 
 	if pInfo == nil {
@@ -756,7 +767,11 @@ func (p *PriorityQueue) addUnschedulableWithoutQueueingHint(logger klog.Logger,
 		// - No unschedulable plugins are associated with this Pod,
 		//   meaning something unusual (a temporal failure on kube-apiserver, etc) happened and this Pod gets moved back to the queue.
 		//   In this case, we should retry scheduling it because this Pod may not be retried until the next flush.
-		_ = p.moveToBackoffQ(logger, pInfo, framework.ScheduleAttemptFailure)
+		if added := p.moveToBackoffQ(logger, pInfo, framework.ScheduleAttemptFailure); added {
+			if p.isPopFromBackoffQEnabled {
+				p.activeQ.broadcast()
+			}
+		}
 	} else {
 		p.unschedulablePods.addOrUpdate(pInfo, framework.ScheduleAttemptFailure)
 		logger.V(5).Info("Pod moved to an internal scheduling queue", "pod", klog.KObj(pod), "event", framework.ScheduleAttemptFailure, "queue", unschedulablePods)
@@ -809,7 +824,7 @@ func (p *PriorityQueue) AddUnschedulableIfNotPresent(logger klog.Logger, pInfo *
 	// In this case, we try to requeue this Pod to activeQ/backoffQ.
 	queue := p.requeuePodViaQueueingHint(logger, pInfo, schedulingHint, framework.ScheduleAttemptFailure)
 	logger.V(3).Info("Pod moved to an internal scheduling queue", "pod", klog.KObj(pod), "event", framework.ScheduleAttemptFailure, "queue", queue, "schedulingCycle", podSchedulingCycle, "hint", schedulingHint, "unschedulable plugins", rejectorPlugins)
-	if queue == activeQ {
+	if queue == activeQ || (p.isPopFromBackoffQEnabled && queue == backoffQ) {
 		// When the Pod is moved to activeQ, need to let p.cond know so that the Pod will be pop()ed out.
 		p.activeQ.broadcast()
 	}
@@ -822,11 +837,12 @@ func (p *PriorityQueue) flushBackoffQCompleted(logger klog.Logger) {
 	p.lock.Lock()
 	defer p.lock.Unlock()
 	activated := false
-	p.backoffQ.popEachBackoffCompleted(logger, func(pInfo *framework.QueuedPodInfo) {
+	podsCompletedBackoff := p.backoffQ.popAllBackoffCompleted(logger)
+	for _, pInfo := range podsCompletedBackoff {
 		if added := p.moveToActiveQ(logger, pInfo, framework.BackoffComplete); added {
 			activated = true
 		}
-	})
+	}
 	if activated {
 		p.activeQ.broadcast()
 	}
@@ -954,7 +970,7 @@ func (p *PriorityQueue) Update(logger klog.Logger, oldPod, newPod *v1.Pod) {
 					logger.V(5).Info("Pod moved to an internal scheduling queue because the Pod is updated", "pod", klog.KObj(newPod), "event", evt.Label(), "queue", queue)
 					p.unschedulablePods.delete(pInfo.Pod, gated)
 				}
-				if queue == activeQ {
+				if queue == activeQ || (p.isPopFromBackoffQEnabled && queue == backoffQ) {
 					p.activeQ.broadcast()
 					break
 				}
@@ -967,6 +983,9 @@ func (p *PriorityQueue) Update(logger klog.Logger, oldPod, newPod *v1.Pod) {
 			if p.backoffQ.isPodBackingoff(pInfo) {
 				if added := p.moveToBackoffQ(logger, pInfo, framework.EventUnscheduledPodUpdate.Label()); added {
 					p.unschedulablePods.delete(pInfo.Pod, gated)
+					if p.isPopFromBackoffQEnabled {
+						p.activeQ.broadcast()
+					}
 				}
 				return
 			}
@@ -995,12 +1014,14 @@ func (p *PriorityQueue) Delete(pod *v1.Pod) {
 	defer p.lock.Unlock()
 	p.DeleteNominatedPodIfExists(pod)
 	pInfo := newQueuedPodInfoForLookup(pod)
-	if err := p.activeQ.delete(pInfo); err != nil {
-		// The item was probably not found in the activeQ.
-		p.backoffQ.delete(pInfo)
-		if pInfo = p.unschedulablePods.get(pod); pInfo != nil {
-			p.unschedulablePods.delete(pod, pInfo.Gated)
-		}
+	if err := p.activeQ.delete(pInfo); err == nil {
+		return
+	}
+	if deleted := p.backoffQ.delete(pInfo); deleted {
+		return
+	}
+	if pInfo = p.unschedulablePods.get(pod); pInfo != nil {
+		p.unschedulablePods.delete(pod, pInfo.Gated)
 	}
 }
 
@@ -1127,7 +1148,7 @@ func (p *PriorityQueue) movePodsToActiveOrBackoffQueue(logger klog.Logger, podIn
 		p.unschedulablePods.delete(pInfo.Pod, pInfo.Gated)
 		queue := p.requeuePodViaQueueingHint(logger, pInfo, schedulingHint, event.Label())
 		logger.V(4).Info("Pod moved to an internal scheduling queue", "pod", klog.KObj(pInfo.Pod), "event", event.Label(), "queue", queue, "hint", schedulingHint)
-		if queue == activeQ {
+		if queue == activeQ || (p.isPopFromBackoffQEnabled && queue == backoffQ) {
 			activated = true
 		}
 	}
@@ -1222,11 +1243,13 @@ func (p *PriorityQueue) PendingPods() ([]*v1.Pod, string) {
 	defer p.lock.RUnlock()
 	result := p.activeQ.list()
 	activeQLen := len(result)
-	result = append(result, p.backoffQ.list()...)
+	backoffQPods := p.backoffQ.list()
+	backoffQLen := len(backoffQPods)
+	result = append(result, backoffQPods...)
 	for _, pInfo := range p.unschedulablePods.podInfoMap {
 		result = append(result, pInfo.Pod)
 	}
-	return result, fmt.Sprintf(pendingPodsSummary, activeQLen, p.backoffQ.len(), len(p.unschedulablePods.podInfoMap))
+	return result, fmt.Sprintf(pendingPodsSummary, activeQLen, backoffQLen, len(p.unschedulablePods.podInfoMap))
 }
 
 // Note: this function assumes the caller locks both p.lock.RLock and p.activeQ.getLock().RLock.

pkg/scheduler/backend/queue/scheduling_queue_test.go

@@ -1010,25 +1010,88 @@ func TestPriorityQueue_AddUnschedulableIfNotPresent_Backoff(t *testing.T) {
 	}
 }
 
-func TestPriorityQueue_Pop(t *testing.T) {
-	objs := []runtime.Object{medPriorityPodInfo.Pod}
-	logger, ctx := ktesting.NewTestContext(t)
-	ctx, cancel := context.WithCancel(ctx)
-	defer cancel()
-	q := NewTestQueueWithObjects(ctx, newDefaultQueueSort(), objs)
-	wg := sync.WaitGroup{}
-	wg.Add(1)
+// tryPop tries to pop one pod from the queue and returns it.
+// It waits 5 seconds before timing out, assuming the queue is then empty.
+func tryPop(t *testing.T, logger klog.Logger, q *PriorityQueue) *framework.QueuedPodInfo {
+	t.Helper()
+
+	var gotPod *framework.QueuedPodInfo
+	popped := make(chan struct{}, 1)
 	go func() {
-		defer wg.Done()
-		if p, err := q.Pop(logger); err != nil || p.Pod != medPriorityPodInfo.Pod {
-			t.Errorf("Expected: %v after Pop, but got: %v", medPriorityPodInfo.Pod.Name, p.Pod.Name)
+		pod, err := q.Pop(logger)
+		if err != nil {
+			t.Errorf("Failed to pop pod from scheduling queue: %s", err)
 		}
-		if len(q.nominator.nominatedPods["node1"]) != 1 {
-			t.Errorf("Expected medPriorityPodInfo to be present in nominatedPods: %v", q.nominator.nominatedPods["node1"])
+		if pod != nil {
+			gotPod = pod
 		}
+		popped <- struct{}{}
 	}()
-	q.Add(logger, medPriorityPodInfo.Pod)
-	wg.Wait()
+
+	timer := time.NewTimer(5 * time.Second)
+	select {
+	case <-timer.C:
+		q.Close()
+	case <-popped:
+		timer.Stop()
+	}
+	return gotPod
+}
+
+func TestPriorityQueue_Pop(t *testing.T) {
+	highPriorityPodInfo2 := mustNewPodInfo(
+		st.MakePod().Name("hpp2").Namespace("ns1").UID("hpp2ns1").Priority(highPriority).Obj(),
+	)
+	objs := []runtime.Object{medPriorityPodInfo.Pod, highPriorityPodInfo.Pod, highPriorityPodInfo2.Pod, unschedulablePodInfo.Pod}
+	tests := []struct {
+		name                   string
+		popFromBackoffQEnabled bool
+		wantPods               []string
+	}{
+		{
+			name:                   "Pop pods from both activeQ and backoffQ when PopFromBackoffQ is enabled",
+			popFromBackoffQEnabled: true,
+			wantPods:               []string{medPriorityPodInfo.Pod.Name, highPriorityPodInfo.Pod.Name},
+		},
+		{
+			name:                   "Pop pod only from activeQ when PopFromBackoffQ is disabled",
+			popFromBackoffQEnabled: false,
+			wantPods:               []string{medPriorityPodInfo.Pod.Name},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			featuregatetesting.SetFeatureGateDuringTest(t, utilfeature.DefaultFeatureGate, features.SchedulerPopFromBackoffQ, tt.popFromBackoffQEnabled)
+			logger, ctx := ktesting.NewTestContext(t)
+			ctx, cancel := context.WithCancel(ctx)
+			defer cancel()
+			q := NewTestQueueWithObjects(ctx, newDefaultQueueSort(), objs)
+
+			// Add medium priority pod to the activeQ
+			q.Add(logger, medPriorityPodInfo.Pod)
+			// Add high priority pod to the backoffQ
+			backoffPodInfo := q.newQueuedPodInfo(highPriorityPodInfo.Pod, "plugin")
+			q.backoffQ.add(logger, backoffPodInfo, framework.EventUnscheduledPodAdd.Label())
+			// Add high priority pod to the errorBackoffQ
+			errorBackoffPodInfo := q.newQueuedPodInfo(highPriorityPodInfo2.Pod)
+			q.backoffQ.add(logger, errorBackoffPodInfo, framework.EventUnscheduledPodAdd.Label())
+			// Add pod to the unschedulablePods
+			unschedulablePodInfo := q.newQueuedPodInfo(unschedulablePodInfo.Pod, "plugin")
+			q.unschedulablePods.addOrUpdate(unschedulablePodInfo, framework.EventUnscheduledPodAdd.Label())
+
+			var gotPods []string
+			for i := 0; i < len(tt.wantPods)+1; i++ {
+				gotPod := tryPop(t, logger, q)
+				if gotPod == nil {
+					break
+				}
+				gotPods = append(gotPods, gotPod.Pod.Name)
+			}
+			if diff := cmp.Diff(tt.wantPods, gotPods); diff != "" {
+				t.Errorf("Unexpected popped pods (-want, +got): %s", diff)
+			}
+		})
+	}
 }
 
 func TestPriorityQueue_Update(t *testing.T) {
@@ -1951,7 +2014,7 @@ func TestPriorityQueue_MoveAllToActiveOrBackoffQueue(t *testing.T) {
 	// pop out the pods in the backoffQ.
 	// This doesn't make them in-flight pods.
 	c.Step(q.backoffQ.podMaxBackoffDuration())
-	q.backoffQ.popEachBackoffCompleted(logger, nil)
+	_ = q.backoffQ.popAllBackoffCompleted(logger)
 	expectInFlightPods(t, q, medPriorityPodInfo.Pod.UID)
 
 	q.Add(logger, unschedulablePodInfo.Pod)
@@ -2074,7 +2137,7 @@ func TestPriorityQueue_MoveAllToActiveOrBackoffQueueWithOutQueueingHint(t *testi
 	// pop out the pods in the backoffQ.
 	// This doesn't make them in-flight pods.
 	c.Step(q.backoffQ.podMaxBackoffDuration())
-	q.backoffQ.popEachBackoffCompleted(logger, nil)
+	_ = q.backoffQ.popAllBackoffCompleted(logger)
 
 	unschedulableQueuedPodInfo := attemptQueuedPodInfo(q.newQueuedPodInfo(unschedulablePodInfo.Pod, "fooPlugin"))
 	highPriorityQueuedPodInfo := attemptQueuedPodInfo(q.newQueuedPodInfo(highPriorityPodInfo.Pod, "fooPlugin"))
@@ -3883,9 +3946,10 @@ func TestMoveAllToActiveOrBackoffQueue_PreEnqueueChecks(t *testing.T) {
 			q.MoveAllToActiveOrBackoffQueue(logger, tt.event, nil, nil, tt.preEnqueueCheck)
 			got := sets.New[string]()
 			c.Step(2 * q.backoffQ.podMaxBackoffDuration())
-			q.backoffQ.popEachBackoffCompleted(logger, func(pInfo *framework.QueuedPodInfo) {
+			gotPodInfos := q.backoffQ.popAllBackoffCompleted(logger)
+			for _, pInfo := range gotPodInfos {
 				got.Insert(pInfo.Pod.Name)
-			})
+			}
 			if diff := cmp.Diff(tt.want, got); diff != "" {
 				t.Errorf("Unexpected diff (-want, +got):\n%s", diff)
 			}

pkg/scheduler/framework/events.go

@@ -31,6 +31,8 @@ const (
 	ScheduleAttemptFailure = "ScheduleAttemptFailure"
 	// BackoffComplete is the event when a pod finishes backoff.
 	BackoffComplete = "BackoffComplete"
+	// PopFromBackoffQ is the event when a pod is popped from backoffQ when activeQ is empty.
+	PopFromBackoffQ = "PopFromBackoffQ"
 	// ForceActivate is the event when a pod is moved from unschedulablePods/backoffQ
 	// to activeQ. Usually it's triggered by plugin implementations.
 	ForceActivate = "ForceActivate"

test/integration/scheduler/eventhandler/eventhandler_test.go

@@ -212,6 +212,8 @@ func TestUpdateNominatedNodeName(t *testing.T) {
 		for _, qHintEnabled := range []bool{false, true} {
 			t.Run(fmt.Sprintf("%s, with queuehint(%v)", tt.name, qHintEnabled), func(t *testing.T) {
 				featuregatetesting.SetFeatureGateDuringTest(t, utilfeature.DefaultFeatureGate, features.SchedulerQueueingHints, qHintEnabled)
+				// Set the SchedulerPopFromBackoffQ feature to false, because when it's enabled, we can't be sure the pod won't be popped from the backoffQ.
+				featuregatetesting.SetFeatureGateDuringTest(t, utilfeature.DefaultFeatureGate, features.SchedulerPopFromBackoffQ, false)
 
 				testCtx, teardown := schedulerutils.InitTestSchedulerForFrameworkTest(t, testContext, 0, true,
 					scheduler.WithClock(fakeClock),

test/integration/scheduler/plugins/plugins_test.go

@@ -309,14 +309,18 @@ var _ framework.PreFilterPlugin = &PreFilterPlugin{}
 var _ framework.PostFilterPlugin = &PostFilterPlugin{}
 var _ framework.ScorePlugin = &ScorePlugin{}
 var _ framework.FilterPlugin = &FilterPlugin{}
+var _ framework.EnqueueExtensions = &FilterPlugin{}
 var _ framework.ScorePlugin = &ScorePlugin{}
 var _ framework.ScorePlugin = &ScoreWithNormalizePlugin{}
+var _ framework.EnqueueExtensions = &ScorePlugin{}
 var _ framework.ReservePlugin = &ReservePlugin{}
 var _ framework.PreScorePlugin = &PreScorePlugin{}
 var _ framework.PreBindPlugin = &PreBindPlugin{}
+var _ framework.EnqueueExtensions = &PreBindPlugin{}
 var _ framework.BindPlugin = &BindPlugin{}
 var _ framework.PostBindPlugin = &PostBindPlugin{}
 var _ framework.PermitPlugin = &PermitPlugin{}
+var _ framework.EnqueueExtensions = &PermitPlugin{}
 var _ framework.QueueSortPlugin = &QueueSortPlugin{}
 
 func (ep *QueueSortPlugin) Name() string {
@@ -377,6 +381,10 @@ func (sp *ScorePlugin) ScoreExtensions() framework.ScoreExtensions {
 	return nil
 }
 
+func (sp *ScorePlugin) EventsToRegister(_ context.Context) ([]framework.ClusterEventWithHint, error) {
+	return nil, nil
+}
+
 // Name returns name of the score plugin.
 func (sp *ScoreWithNormalizePlugin) Name() string {
 	return scoreWithNormalizePluginName
@@ -427,6 +435,12 @@ func (fp *FilterPlugin) Filter(ctx context.Context, state *framework.CycleState,
 	return nil
 }
 
+func (fp *FilterPlugin) EventsToRegister(_ context.Context) ([]framework.ClusterEventWithHint, error) {
+	return []framework.ClusterEventWithHint{
+		{Event: framework.ClusterEvent{Resource: framework.Pod, ActionType: framework.Delete}},
+	}, nil
+}
+
 // Name returns name of the plugin.
 func (rp *ReservePlugin) Name() string {
 	return rp.name
@@ -491,6 +505,10 @@ func (pp *PreBindPlugin) PreBind(ctx context.Context, state *framework.CycleStat
 	return nil
 }
 
+func (pp *PreBindPlugin) EventsToRegister(_ context.Context) ([]framework.ClusterEventWithHint, error) {
+	return nil, nil
+}
+
 const bindPluginAnnotation = "bindPluginName"
 
 func (bp *BindPlugin) Name() string {
@@ -651,6 +669,10 @@ func (pp *PermitPlugin) rejectAllPods() {
 	pp.fh.IterateOverWaitingPods(func(wp framework.WaitingPod) { wp.Reject(pp.name, "rejectAllPods") })
 }
 
+func (pp *PermitPlugin) EventsToRegister(_ context.Context) ([]framework.ClusterEventWithHint, error) {
+	return nil, nil
+}
+
 // TestPreFilterPlugin tests invocation of prefilter plugins.
 func TestPreFilterPlugin(t *testing.T) {
 	testContext := testutils.InitTestAPIServer(t, "prefilter-plugin", nil)

test/integration/scheduler/queueing/queue_test.go

@@ -581,3 +581,38 @@ func (p *fakePermitPlugin) EventsToRegister(_ context.Context) ([]framework.Clus
 		{Event: framework.ClusterEvent{Resource: framework.Node, ActionType: framework.UpdateNodeLabel}, QueueingHintFn: p.schedulingHint},
 	}, nil
 }
+
+func TestPopFromBackoffQWhenActiveQEmpty(t *testing.T) {
+	// Set initial backoff to 1000s to make sure pod won't go to the activeQ after being requeued.
+	testCtx := testutils.InitTestSchedulerWithNS(t, "pop-from-backoffq", scheduler.WithPodInitialBackoffSeconds(1000), scheduler.WithPodMaxBackoffSeconds(1000))
+	cs, ns, ctx := testCtx.ClientSet, testCtx.NS.Name, testCtx.Ctx
+
+	// Create node, so we can schedule pods.
+	node := st.MakeNode().Name("node").Obj()
+	if _, err := cs.CoreV1().Nodes().Create(ctx, node, metav1.CreateOptions{}); err != nil {
+		t.Fatal("Failed to create node")
+	}
+
+	// Create a pod that will be unschedulable.
+	pod := st.MakePod().Namespace(ns).Name("pod").NodeAffinityIn("foo", []string{"bar"}, st.NodeSelectorTypeMatchExpressions).Container(imageutils.GetPauseImageName()).Obj()
+	if _, err := cs.CoreV1().Pods(ns).Create(ctx, pod, metav1.CreateOptions{}); err != nil {
+		t.Fatalf("Failed to create pod: %v", err)
+	}
+
+	err := wait.PollUntilContextTimeout(ctx, 200*time.Millisecond, wait.ForeverTestTimeout, false, testutils.PodUnschedulable(cs, ns, pod.Name))
+	if err != nil {
+		t.Fatalf("Expected pod to be unschedulable: %v", err)
+	}
+
+	// Create node with label to make the pod schedulable.
+	node2 := st.MakeNode().Name("node-schedulable").Label("foo", "bar").Obj()
+	if _, err := cs.CoreV1().Nodes().Create(ctx, node2, metav1.CreateOptions{}); err != nil {
+		t.Fatal("Failed to create node-schedulable")
+	}
+
+	// Pod should be scheduled, even if it was in the backoffQ, because PopFromBackoffQ feature is enabled.
+	err = wait.PollUntilContextTimeout(ctx, 200*time.Millisecond, wait.ForeverTestTimeout, false, testutils.PodScheduled(cs, ns, pod.Name))
+	if err != nil {
+		t.Fatalf("Expected pod to be scheduled: %v", err)
+	}
+}