github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-09-26 12:46:06 +00:00
Code Issues Projects Releases Wiki Activity

forget the pod when the reserve plugins fail

Browse Source 
and move the metrics function calls before all other functions
This commit is contained in:
Yecheng Fu
2020-08-09 20:28:23 +08:00
parent 9b78bd5979
commit 1176ef9c7d
4 changed files with 157 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
pkg/scheduler/scheduler.go
View File

@@ -422,9 +422,6 @@ func (sched *Scheduler) finishBinding(prof *profile.Profile, assumed *v1.Pod, ta
}
if err != nil {
klog.V(1).Infof("Failed to bind pod: %v/%v", assumed.Namespace, assumed.Name)
if err := sched.SchedulerCache.ForgetPod(assumed); err != nil {
klog.Errorf("scheduler cache ForgetPod failed: %v", err)
}
return
}
@@ -504,22 +501,25 @@ func (sched *Scheduler) scheduleOne(ctx context.Context) {
// assume modifies `assumedPod` by setting NodeName=scheduleResult.SuggestedHost
err = sched.assume(assumedPod, scheduleResult.SuggestedHost)
if err != nil {
metrics.PodScheduleError(prof.Name, metrics.SinceInSeconds(start))
// This is most probably result of a BUG in retrying logic.
// We report an error here so that pod scheduling can be retried.
// 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).
sched.recordSchedulingFailure(prof, assumedPodInfo, err, SchedulerError, "")
metrics.PodScheduleError(prof.Name, metrics.SinceInSeconds(start))
return
}
// Run the Reserve method of reserve plugins.
if sts := prof.RunReservePluginsReserve(schedulingCycleCtx, state, assumedPod, scheduleResult.SuggestedHost); !sts.IsSuccess() {
sched.recordSchedulingFailure(prof, assumedPodInfo, sts.AsError(), SchedulerError, "")
metrics.PodScheduleError(prof.Name, metrics.SinceInSeconds(start))
// trigger un-reserve to clean up state associated with the reserved Pod
prof.RunReservePluginsUnreserve(schedulingCycleCtx, state, assumedPod, scheduleResult.SuggestedHost)
if forgetErr := sched.Cache().ForgetPod(assumedPod); forgetErr != nil {
klog.Errorf("scheduler cache ForgetPod failed: %v", forgetErr)
}
sched.recordSchedulingFailure(prof, assumedPodInfo, sts.AsError(), SchedulerError, "")
return
}
@@ -587,6 +587,9 @@ func (sched *Scheduler) scheduleOne(ctx context.Context) {
metrics.PodScheduleError(prof.Name, metrics.SinceInSeconds(start))
// trigger un-reserve plugins to clean up state associated with the reserved Pod
prof.RunReservePluginsUnreserve(bindingCycleCtx, state, assumedPod, scheduleResult.SuggestedHost)
if err := sched.SchedulerCache.ForgetPod(assumedPod); err != nil {
klog.Errorf("scheduler cache ForgetPod failed: %v", err)
}
sched.recordSchedulingFailure(prof, assumedPodInfo, fmt.Errorf("Binding rejected: %v", err), SchedulerError, "")
} else {
// Calculating nodeResourceString can be heavy. Avoid it if klog verbosity is below 2.