github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-07-24 12:15:52 +00:00
Code Issues Projects Releases Wiki Activity

Fix metrics time durations in schedulerCycle and bindingCycle

Browse Source 
Signed-off-by: kerthcet <kerthcet@gmail.com>
This commit is contained in:
kerthcet 2022-10-17 23:24:24 +08:00
parent 64b98495ec
commit 1582c42e2b
1 changed files with 35 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

64
pkg/scheduler/schedule_one.go
View File

@ -176,12 +176,13 @@ func (sched *Scheduler) schedulingCycle(ctx context.Context, state *framework.Cy
// Run the Reserve method of reserve plugins. // Run the Reserve method of reserve plugins.
if sts := fwk.RunReservePluginsReserve(ctx, state, assumedPod, scheduleResult.SuggestedHost); !sts.IsSuccess() { if sts := fwk.RunReservePluginsReserve(ctx, state, assumedPod, scheduleResult.SuggestedHost); !sts.IsSuccess() {
metrics.PodScheduleError(fwk.ProfileName(), metrics.SinceInSeconds(start))
// trigger un-reserve to clean up state associated with the reserved Pod // trigger un-reserve to clean up state associated with the reserved Pod
fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost) fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost)
if forgetErr := sched.Cache.ForgetPod(assumedPod); forgetErr != nil { if forgetErr := sched.Cache.ForgetPod(assumedPod); forgetErr != nil {
klog.ErrorS(forgetErr, "Scheduler cache ForgetPod failed") klog.ErrorS(forgetErr, "Scheduler cache ForgetPod failed")
} }
metrics.PodScheduleError(fwk.ProfileName(), metrics.SinceInSeconds(start))
sched.FailureHandler(ctx, fwk, assumedPodInfo, sts.AsError(), v1.PodReasonSchedulerError, clearNominatedNode) sched.FailureHandler(ctx, fwk, assumedPodInfo, sts.AsError(), v1.PodReasonSchedulerError, clearNominatedNode)
return ScheduleResult{}, nil return ScheduleResult{}, nil
} }
@ -189,6 +190,12 @@ func (sched *Scheduler) schedulingCycle(ctx context.Context, state *framework.Cy
// Run "permit" plugins. // Run "permit" plugins.
runPermitStatus := fwk.RunPermitPlugins(ctx, state, assumedPod, scheduleResult.SuggestedHost) runPermitStatus := fwk.RunPermitPlugins(ctx, state, assumedPod, scheduleResult.SuggestedHost)
if !runPermitStatus.IsWait() && !runPermitStatus.IsSuccess() { if !runPermitStatus.IsWait() && !runPermitStatus.IsSuccess() {
// One of the plugins returned status different from success or wait.
fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost)
if forgetErr := sched.Cache.ForgetPod(assumedPod); forgetErr != nil {
klog.ErrorS(forgetErr, "Scheduler cache ForgetPod failed")
}
var reason string var reason string
if runPermitStatus.IsUnschedulable() { if runPermitStatus.IsUnschedulable() {
metrics.PodUnschedulable(fwk.ProfileName(), metrics.SinceInSeconds(start)) metrics.PodUnschedulable(fwk.ProfileName(), metrics.SinceInSeconds(start))
@ -197,11 +204,7 @@ func (sched *Scheduler) schedulingCycle(ctx context.Context, state *framework.Cy
metrics.PodScheduleError(fwk.ProfileName(), metrics.SinceInSeconds(start)) metrics.PodScheduleError(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonSchedulerError reason = v1.PodReasonSchedulerError
} }
// One of the plugins returned status different from success or wait.
fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost)
if forgetErr := sched.Cache.ForgetPod(assumedPod); forgetErr != nil {
klog.ErrorS(forgetErr, "Scheduler cache ForgetPod failed")
}
sched.FailureHandler(ctx, fwk, assumedPodInfo, runPermitStatus.AsError(), reason, clearNominatedNode) sched.FailureHandler(ctx, fwk, assumedPodInfo, runPermitStatus.AsError(), reason, clearNominatedNode)
return ScheduleResult{}, nil return ScheduleResult{}, nil
} }
@ -222,14 +225,6 @@ func (sched *Scheduler) bindingCycle(ctx context.Context, state *framework.Cycle
waitOnPermitStatus := fwk.WaitOnPermit(ctx, assumedPod) waitOnPermitStatus := fwk.WaitOnPermit(ctx, assumedPod)
if !waitOnPermitStatus.IsSuccess() { if !waitOnPermitStatus.IsSuccess() {
var reason string
if waitOnPermitStatus.IsUnschedulable() {
metrics.PodUnschedulable(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonUnschedulable
} else {
metrics.PodScheduleError(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonSchedulerError
}
// trigger un-reserve plugins to clean up state associated with the reserved Pod // trigger un-reserve plugins to clean up state associated with the reserved Pod
fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost) fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost)
if forgetErr := sched.Cache.ForgetPod(assumedPod); forgetErr != nil { if forgetErr := sched.Cache.ForgetPod(assumedPod); forgetErr != nil {
@ -249,6 +244,15 @@ func (sched *Scheduler) bindingCycle(ctx context.Context, state *framework.Cycle
sched.SchedulingQueue.MoveAllToActiveOrBackoffQueue(internalqueue.AssignedPodDelete, nil) sched.SchedulingQueue.MoveAllToActiveOrBackoffQueue(internalqueue.AssignedPodDelete, nil)
} }
} }
var reason string
if waitOnPermitStatus.IsUnschedulable() {
metrics.PodUnschedulable(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonUnschedulable
} else {
metrics.PodScheduleError(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonSchedulerError
}
sched.FailureHandler(ctx, fwk, assumedPodInfo, waitOnPermitStatus.AsError(), reason, clearNominatedNode) sched.FailureHandler(ctx, fwk, assumedPodInfo, waitOnPermitStatus.AsError(), reason, clearNominatedNode)
return return
} }
@ -256,15 +260,6 @@ func (sched *Scheduler) bindingCycle(ctx context.Context, state *framework.Cycle
// Run "prebind" plugins. // Run "prebind" plugins.
preBindStatus := fwk.RunPreBindPlugins(ctx, state, assumedPod, scheduleResult.SuggestedHost) preBindStatus := fwk.RunPreBindPlugins(ctx, state, assumedPod, scheduleResult.SuggestedHost)
if !preBindStatus.IsSuccess() { if !preBindStatus.IsSuccess() {
var reason string
if preBindStatus.IsUnschedulable() {
metrics.PodUnschedulable(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonUnschedulable
} else {
metrics.PodScheduleError(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonSchedulerError
}
// trigger un-reserve plugins to clean up state associated with the reserved Pod // trigger un-reserve plugins to clean up state associated with the reserved Pod
fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost) fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost)
if forgetErr := sched.Cache.ForgetPod(assumedPod); forgetErr != nil { if forgetErr := sched.Cache.ForgetPod(assumedPod); forgetErr != nil {
@ -281,14 +276,9 @@ func (sched *Scheduler) bindingCycle(ctx context.Context, state *framework.Cycle
sched.SchedulingQueue.MoveAllToActiveOrBackoffQueue(internalqueue.AssignedPodDelete, nil) sched.SchedulingQueue.MoveAllToActiveOrBackoffQueue(internalqueue.AssignedPodDelete, nil)
} }
} }
sched.FailureHandler(ctx, fwk, assumedPodInfo, preBindStatus.AsError(), reason, clearNominatedNode)
return
}
bindStatus := sched.bind(ctx, fwk, assumedPod, scheduleResult.SuggestedHost, state)
if !bindStatus.IsSuccess() {
var reason string var reason string
if bindStatus.IsUnschedulable() { if preBindStatus.IsUnschedulable() {
metrics.PodUnschedulable(fwk.ProfileName(), metrics.SinceInSeconds(start)) metrics.PodUnschedulable(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonUnschedulable reason = v1.PodReasonUnschedulable
} else { } else {
@ -296,6 +286,12 @@ func (sched *Scheduler) bindingCycle(ctx context.Context, state *framework.Cycle
reason = v1.PodReasonSchedulerError reason = v1.PodReasonSchedulerError
} }
sched.FailureHandler(ctx, fwk, assumedPodInfo, preBindStatus.AsError(), reason, clearNominatedNode)
return
}
bindStatus := sched.bind(ctx, fwk, assumedPod, scheduleResult.SuggestedHost, state)
if !bindStatus.IsSuccess() {
// trigger un-reserve plugins to clean up state associated with the reserved Pod // trigger un-reserve plugins to clean up state associated with the reserved Pod
fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost) fwk.RunReservePluginsUnreserve(ctx, state, assumedPod, scheduleResult.SuggestedHost)
if err := sched.Cache.ForgetPod(assumedPod); err != nil { if err := sched.Cache.ForgetPod(assumedPod); err != nil {
@ -312,6 +308,16 @@ func (sched *Scheduler) bindingCycle(ctx context.Context, state *framework.Cycle
sched.SchedulingQueue.MoveAllToActiveOrBackoffQueue(internalqueue.AssignedPodDelete, nil) sched.SchedulingQueue.MoveAllToActiveOrBackoffQueue(internalqueue.AssignedPodDelete, nil)
} }
} }
var reason string
if bindStatus.IsUnschedulable() {
metrics.PodUnschedulable(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonUnschedulable
} else {
metrics.PodScheduleError(fwk.ProfileName(), metrics.SinceInSeconds(start))
reason = v1.PodReasonSchedulerError
}
sched.FailureHandler(ctx, fwk, assumedPodInfo, fmt.Errorf("binding rejected: %w", bindStatus.AsError()), reason, clearNominatedNode) sched.FailureHandler(ctx, fwk, assumedPodInfo, fmt.Errorf("binding rejected: %w", bindStatus.AsError()), reason, clearNominatedNode)
return return
} }