github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2026-08-08 23:37:11 +00:00
Code Issues Projects Releases Wiki Activity

Merge pull request #135007 from ania-borowiec/nnn_code_update

Browse Source 
KEP-5278 Bring back clearing NominatedNodeName after scheduling or binding failure
This commit is contained in:
Kubernetes Prow Robot
2025-11-04 07:14:24 -08:00
committed by GitHub
parent ac46677d2a c98804b77f
commit f09cd625bc
3 changed files with 31 additions and 85 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
pkg/scheduler/schedule_one.go
View File

@@ -136,15 +136,7 @@ func (sched *Scheduler) ScheduleOne(ctx context.Context) {
}()
}
// newFailureNominatingInfo returns the appropriate NominatingInfo for scheduling failures.
// When NominatedNodeNameForExpectation feature is enabled, it returns nil (no clearing).
// Otherwise, it returns NominatingInfo to clear the pod's nominated node.
func (sched *Scheduler) newFailureNominatingInfo() *fwk.NominatingInfo {
if sched.nominatedNodeNameForExpectationEnabled {
return nil
}
return &fwk.NominatingInfo{NominatingMode: fwk.ModeOverride, NominatedNodeName: ""}
}
var clearNominatedNode = &fwk.NominatingInfo{NominatingMode: fwk.ModeOverride, NominatedNodeName: ""}
// schedulingCycle tries to schedule a single Pod.
func (sched *Scheduler) schedulingCycle(
@@ -164,13 +156,13 @@ func (sched *Scheduler) schedulingCycle(
}()
if err == ErrNoNodesAvailable {
status := fwk.NewStatus(fwk.UnschedulableAndUnresolvable).WithError(err)
return ScheduleResult{nominatingInfo: sched.newFailureNominatingInfo()}, podInfo, status
return ScheduleResult{nominatingInfo: clearNominatedNode}, podInfo, status
}
fitError, ok := err.(*framework.FitError)
if !ok {
logger.Error(err, "Error selecting node for pod", "pod", klog.KObj(pod))
return ScheduleResult{nominatingInfo: sched.newFailureNominatingInfo()}, podInfo, fwk.AsStatus(err)
return ScheduleResult{nominatingInfo: clearNominatedNode}, podInfo, fwk.AsStatus(err)
}
// SchedulePod() may have failed because the pod would not fit on any host, so we try to
@@ -180,7 +172,7 @@ func (sched *Scheduler) schedulingCycle(
if !schedFramework.HasPostFilterPlugins() {
logger.V(3).Info("No PostFilter plugins are registered, so no preemption will be performed")
return ScheduleResult{}, podInfo, fwk.NewStatus(fwk.Unschedulable).WithError(err)
return ScheduleResult{nominatingInfo: clearNominatedNode}, podInfo, fwk.NewStatus(fwk.Unschedulable).WithError(err)
}
// Run PostFilter plugins to attempt to make the pod schedulable in a future scheduling cycle.
@@ -213,7 +205,7 @@ func (sched *Scheduler) schedulingCycle(
// This relies on the fact that Error will check if the pod has been bound
// to a node and if so will not add it back to the unscheduled pods queue
// (otherwise this would cause an infinite loop).
return ScheduleResult{nominatingInfo: sched.newFailureNominatingInfo()}, assumedPodInfo, fwk.AsStatus(err)
return ScheduleResult{nominatingInfo: clearNominatedNode}, assumedPodInfo, fwk.AsStatus(err)
}
// Run the Reserve method of reserve plugins.
@@ -234,9 +226,9 @@ func (sched *Scheduler) schedulingCycle(
}
fitErr.Diagnosis.NodeToStatus.Set(scheduleResult.SuggestedHost, sts)
fitErr.Diagnosis.AddPluginStatus(sts)
return ScheduleResult{nominatingInfo: sched.newFailureNominatingInfo()}, assumedPodInfo, fwk.NewStatus(sts.Code()).WithError(fitErr)
return ScheduleResult{nominatingInfo: clearNominatedNode}, assumedPodInfo, fwk.NewStatus(sts.Code()).WithError(fitErr)
}
return ScheduleResult{nominatingInfo: sched.newFailureNominatingInfo()}, assumedPodInfo, sts
return ScheduleResult{nominatingInfo: clearNominatedNode}, assumedPodInfo, sts
}
// Run "permit" plugins.
@@ -258,10 +250,10 @@ func (sched *Scheduler) schedulingCycle(
}
fitErr.Diagnosis.NodeToStatus.Set(scheduleResult.SuggestedHost, runPermitStatus)
fitErr.Diagnosis.AddPluginStatus(runPermitStatus)
return ScheduleResult{nominatingInfo: sched.newFailureNominatingInfo()}, assumedPodInfo, fwk.NewStatus(runPermitStatus.Code()).WithError(fitErr)
return ScheduleResult{nominatingInfo: clearNominatedNode}, assumedPodInfo, fwk.NewStatus(runPermitStatus.Code()).WithError(fitErr)
}
return ScheduleResult{nominatingInfo: sched.newFailureNominatingInfo()}, assumedPodInfo, runPermitStatus
return ScheduleResult{nominatingInfo: clearNominatedNode}, assumedPodInfo, runPermitStatus
}
// At the end of a successful scheduling cycle, pop and move up Pods if needed.
@@ -393,7 +385,7 @@ func (sched *Scheduler) handleBindingCycleError(
}
}
sched.FailureHandler(ctx, fwk, podInfo, status, sched.newFailureNominatingInfo(), start)
sched.FailureHandler(ctx, fwk, podInfo, status, clearNominatedNode, start)
}
func (sched *Scheduler) frameworkForPod(pod *v1.Pod) (framework.Framework, error) {

41
pkg/scheduler/schedule_one_test.go
View File

@@ -916,33 +916,6 @@ func TestSchedulerScheduleOne(t *testing.T) {
mockScheduleResult: emptyScheduleResult,
eventReason: "FailedScheduling",
},
{
name: "pod with existing nominated node name on scheduling error keeps nomination",
sendPod: func() *v1.Pod {
p := podWithID("foo", "")
p.Status.NominatedNodeName = "existing-node"
return p
}(),
injectSchedulingError: schedulingErr,
mockScheduleResult: scheduleResultOk,
expectError: schedulingErr,
expectErrorPod: func() *v1.Pod {
p := podWithID("foo", "")
p.Status.NominatedNodeName = "existing-node"
return p
}(),
expectPodInBackoffQ: func() *v1.Pod {
p := podWithID("foo", "")
p.Status.NominatedNodeName = "existing-node"
return p
}(),
// Depending on the timing, if asyncAPICallsEnabled, the NNN update might not be sent yet while checking the expectNominatedNodeName.
// So, asyncAPICallsEnabled is set to false.
asyncAPICallsEnabled: ptr.To(false),
nominatedNodeNameForExpectationEnabled: ptr.To(true),
expectNominatedNodeName: "existing-node",
eventReason: "FailedScheduling",
},
{
name: "pod with existing nominated node name on scheduling error clears nomination",
sendPod: func() *v1.Pod {
@@ -965,9 +938,9 @@ func TestSchedulerScheduleOne(t *testing.T) {
}(),
// Depending on the timing, if asyncAPICallsEnabled, the NNN update might not be sent yet while checking the expectNominatedNodeName.
// So, asyncAPICallsEnabled is set to false.
asyncAPICallsEnabled: ptr.To(false),
nominatedNodeNameForExpectationEnabled: ptr.To(false),
eventReason: "FailedScheduling",
asyncAPICallsEnabled: ptr.To(false),
expectNominatedNodeName: "",
eventReason: "FailedScheduling",
},
}
@@ -986,7 +959,7 @@ func TestSchedulerScheduleOne(t *testing.T) {
for _, nominatedNodeNameForExpectationEnabled := range nominatedNodeNameForExpectationEnabled {
if (asyncAPICallsEnabled || nominatedNodeNameForExpectationEnabled) && !qHintEnabled {
// If the QHint feature gate is disabled, NominatedNodeNameForExpectation and SchedulerAsyncAPICalls cannot be enabled
// because that means users set the emilation version to 1.33 or later.
// because that means users set the emulation version to 1.33 or later.
continue
}
t.Run(fmt.Sprintf("%s (Queueing hints enabled: %v, Async API calls enabled: %v, NominatedNodeNameForExpectation enabled: %v)", item.name, qHintEnabled, asyncAPICallsEnabled, nominatedNodeNameForExpectationEnabled), func(t *testing.T) {
@@ -1157,11 +1130,7 @@ func TestSchedulerScheduleOne(t *testing.T) {
t.Errorf("Unexpected error. Wanted %v, got %v", item.expectError.Error(), gotError.Error())
}
if item.expectError != nil {
var expectedNominatingInfo *fwk.NominatingInfo
// Check nominatingInfo expectation based on feature gate
if !nominatedNodeNameForExpectationEnabled {
expectedNominatingInfo = &fwk.NominatingInfo{NominatingMode: fwk.ModeOverride, NominatedNodeName: ""}
}
expectedNominatingInfo := &fwk.NominatingInfo{NominatingMode: fwk.ModeOverride, NominatedNodeName: ""}
if diff := cmp.Diff(expectedNominatingInfo, gotNominatingInfo); diff != "" {
t.Errorf("Unexpected nominatingInfo (-want,+got):\n%s", diff)
}