Merge pull request kubernetes#124624 from atiratree/fix-flake

[test] statefulset controller: fix requests tracker concurrency

Author: k8s-ci-robot

pkg/controller/statefulset/stateful_set_control_test.go

@@ -20,6 +20,7 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"math"
 	"math/rand"
 	"reflect"
 	"runtime"
@@ -2423,51 +2424,53 @@ type requestTracker struct {
 	err      error
 	after    int
 
-	parallelLock sync.Mutex
-	parallel     int
-	maxParallel  int
-
-	delay time.Duration
+	// this block should be updated consistently
+	parallelLock                sync.Mutex
+	shouldTrackParallelRequests bool
+	parallelRequests            int
+	maxParallelRequests         int
+	parallelRequestDelay        time.Duration
 }
 
-func (rt *requestTracker) errorReady() bool {
-	rt.Lock()
-	defer rt.Unlock()
-	return rt.err != nil && rt.requests >= rt.after
-}
-
-func (rt *requestTracker) inc() {
-	rt.parallelLock.Lock()
-	rt.parallel++
-	if rt.maxParallel < rt.parallel {
-		rt.maxParallel = rt.parallel
+func (rt *requestTracker) trackParallelRequests() {
+	if !rt.shouldTrackParallelRequests {
+		// do not track parallel requests unless specifically enabled
+		return
 	}
-	rt.parallelLock.Unlock()
-
-	rt.Lock()
-	defer rt.Unlock()
-	rt.requests++
-	if rt.delay != 0 {
-		time.Sleep(rt.delay)
+	if rt.parallelLock.TryLock() {
+		// lock acquired: we are the only or the first concurrent request
+		// initialize the next set of parallel requests
+		rt.parallelRequests = 1
+	} else {
+		// lock is held by other requests
+		// now wait for the lock to increase the parallelRequests
+		rt.parallelLock.Lock()
+		rt.parallelRequests++
+	}
+	defer rt.parallelLock.Unlock()
+	// update the local maximum of parallel collisions
+	if rt.maxParallelRequests < rt.parallelRequests {
+		rt.maxParallelRequests = rt.parallelRequests
+	}
+	// increase the chance of collisions
+	if rt.parallelRequestDelay > 0 {
+		time.Sleep(rt.parallelRequestDelay)
 	}
 }
 
-func (rt *requestTracker) reset() {
-	rt.parallelLock.Lock()
-	rt.parallel = 0
-	rt.parallelLock.Unlock()
-
-	rt.Lock()
-	defer rt.Unlock()
-	rt.err = nil
-	rt.after = 0
-	rt.delay = 0
-}
-
-func (rt *requestTracker) getErr() error {
+func (rt *requestTracker) incWithOptionalError() error {
 	rt.Lock()
 	defer rt.Unlock()
-	return rt.err
+	rt.requests++
+	if rt.err != nil && rt.requests >= rt.after {
+		// reset and pass the error
+		defer func() {
+			rt.err = nil
+			rt.after = 0
+		}()
+		return rt.err
+	}
+	return nil
 }
 
 func newRequestTracker(requests int, err error, after int) requestTracker {
@@ -2512,10 +2515,9 @@ func newFakeObjectManager(informerFactory informers.SharedInformerFactory) *fake
 }
 
 func (om *fakeObjectManager) CreatePod(ctx context.Context, pod *v1.Pod) error {
-	defer om.createPodTracker.inc()
-	if om.createPodTracker.errorReady() {
-		defer om.createPodTracker.reset()
-		return om.createPodTracker.getErr()
+	defer om.createPodTracker.trackParallelRequests()
+	if err := om.createPodTracker.incWithOptionalError(); err != nil {
+		return err
 	}
 	pod.SetUID(types.UID(pod.Name + "-uid"))
 	return om.podsIndexer.Update(pod)
@@ -2526,19 +2528,17 @@ func (om *fakeObjectManager) GetPod(namespace, podName string) (*v1.Pod, error)
 }
 
 func (om *fakeObjectManager) UpdatePod(pod *v1.Pod) error {
-	defer om.updatePodTracker.inc()
-	if om.updatePodTracker.errorReady() {
-		defer om.updatePodTracker.reset()
-		return om.updatePodTracker.getErr()
+	defer om.updatePodTracker.trackParallelRequests()
+	if err := om.updatePodTracker.incWithOptionalError(); err != nil {
+		return err
 	}
 	return om.podsIndexer.Update(pod)
 }
 
 func (om *fakeObjectManager) DeletePod(pod *v1.Pod) error {
-	defer om.deletePodTracker.inc()
-	if om.deletePodTracker.errorReady() {
-		defer om.deletePodTracker.reset()
-		return om.deletePodTracker.getErr()
+	defer om.deletePodTracker.trackParallelRequests()
+	if err := om.deletePodTracker.incWithOptionalError(); err != nil {
+		return err
 	}
 	if key, err := controller.KeyFunc(pod); err != nil {
 		return err
@@ -2733,10 +2733,9 @@ func newFakeStatefulSetStatusUpdater(setInformer appsinformers.StatefulSetInform
 }
 
 func (ssu *fakeStatefulSetStatusUpdater) UpdateStatefulSetStatus(ctx context.Context, set *apps.StatefulSet, status *apps.StatefulSetStatus) error {
-	defer ssu.updateStatusTracker.inc()
-	if ssu.updateStatusTracker.errorReady() {
-		defer ssu.updateStatusTracker.reset()
-		return ssu.updateStatusTracker.err
+	defer ssu.updateStatusTracker.trackParallelRequests()
+	if err := ssu.updateStatusTracker.incWithOptionalError(); err != nil {
+		return err
 	}
 	set.Status = *status
 	ssu.setsIndexer.Update(set)
@@ -2942,50 +2941,61 @@ func fakeResourceVersion(object interface{}) {
 		obj.SetResourceVersion(strconv.FormatInt(intValue+1, 10))
 	}
 }
-
 func TestParallelScale(t *testing.T) {
 	for _, tc := range []struct {
-		desc            string
-		replicas        int32
-		desiredReplicas int32
+		desc                        string
+		replicas                    int32
+		desiredReplicas             int32
+		expectedMinParallelRequests int
 	}{
 		{
-			desc:            "scale up from 3 to 30",
-			replicas:        3,
-			desiredReplicas: 30,
+			desc:                        "scale up from 3 to 30",
+			replicas:                    3,
+			desiredReplicas:             30,
+			expectedMinParallelRequests: 2,
 		},
 		{
-			desc:            "scale down from 10 to 1",
-			replicas:        10,
-			desiredReplicas: 1,
+			desc:                        "scale down from 10 to 1",
+			replicas:                    10,
+			desiredReplicas:             1,
+			expectedMinParallelRequests: 2,
 		},
 
 		{
-			desc:            "scale down to 0",
-			replicas:        501,
-			desiredReplicas: 0,
+			desc:                        "scale down to 0",
+			replicas:                    501,
+			desiredReplicas:             0,
+			expectedMinParallelRequests: 10,
 		},
 		{
-			desc:            "scale up from 0",
-			replicas:        0,
-			desiredReplicas: 1000,
+			desc:                        "scale up from 0",
+			replicas:                    0,
+			desiredReplicas:             1000,
+			expectedMinParallelRequests: 20,
 		},
 	} {
 		t.Run(tc.desc, func(t *testing.T) {
 			set := burst(newStatefulSet(0))
 			set.Spec.VolumeClaimTemplates[0].ObjectMeta.Labels = map[string]string{"test": "test"}
-			parallelScale(t, set, tc.replicas, tc.desiredReplicas, assertBurstInvariants)
+			parallelScale(t, set, tc.replicas, tc.desiredReplicas, tc.expectedMinParallelRequests, assertBurstInvariants)
 		})
 	}
 
 }
 
-func parallelScale(t *testing.T, set *apps.StatefulSet, replicas, desiredReplicas int32, invariants invariantFunc) {
+func parallelScale(t *testing.T, set *apps.StatefulSet, replicas, desiredReplicas int32, expectedMinParallelRequests int, invariants invariantFunc) {
 	var err error
 	diff := desiredReplicas - replicas
+
+	// maxParallelRequests: MaxBatchSize of the controller is 500, We divide the diff by 4 to allow maximum of the half of the last batch.
+	if maxParallelRequests := min(500, math.Abs(float64(diff))/4); expectedMinParallelRequests < 2 || float64(expectedMinParallelRequests) > maxParallelRequests {
+		t.Fatalf("expectedMinParallelRequests should be between 2 and %v. Batch size of the controller is expontially increasing until 500. "+
+			"Got expectedMinParallelRequests %v, ", maxParallelRequests, expectedMinParallelRequests)
+	}
 	client := fake.NewSimpleClientset(set)
 	om, _, ssc := setupController(client)
-	om.createPodTracker.delay = time.Millisecond
+	om.createPodTracker.shouldTrackParallelRequests = true
+	om.createPodTracker.parallelRequestDelay = time.Millisecond
 
 	*set.Spec.Replicas = replicas
 	if err := parallelScaleUpStatefulSetControl(set, ssc, om, invariants); err != nil {
@@ -3017,8 +3027,8 @@ func parallelScale(t *testing.T, set *apps.StatefulSet, replicas, desiredReplica
 		t.Errorf("Failed to scale statefulset to %v replicas, got %v replicas", desiredReplicas, set.Status.Replicas)
 	}
 
-	if (diff < -1 || diff > 1) && om.createPodTracker.maxParallel <= 1 {
-		t.Errorf("want max parallel requests > 1, got %v", om.createPodTracker.maxParallel)
+	if om.createPodTracker.maxParallelRequests < expectedMinParallelRequests {
+		t.Errorf("want max parallelRequests requests >= %v, got %v", expectedMinParallelRequests, om.createPodTracker.maxParallelRequests)
 	}
 }