Skip to content

fix(scheduler): handle tombstone objects to avoid missing deletes on relist/reconnect #8

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dkeven merged 1 commit into from
Jan 5, 2026
Merged

fix(scheduler): handle tombstone objects to avoid missing deletes on relist/reconnect #8

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 18 additions & 8 deletions pkg/scheduler/scheduler.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -110,20 +110,30 @@ func (s *Scheduler) onUpdatePod(_, newObj any) {
}

func (s *Scheduler) onDelPod(obj any) {
pod, ok := obj.(*corev1.Pod)
if !ok {
klog.Errorf("unknown add object type")
return
}
_, ok = pod.Annotations[util.AssignedNodeAnnotations]
if !ok {
var pod *corev1.Pod
switch t := obj.(type) {
case *corev1.Pod:
pod = t
case cache.DeletedFinalStateUnknown:
// Handle tombstone objects to avoid missing deletes on relist/reconnect
if p, ok := t.Obj.(*corev1.Pod); ok {
pod = p
} else {
klog.Errorf("tombstone contained object that is not a Pod: %#v", t.Obj)
return
}
default:
klog.Errorf("unknown delete object type: %#v", obj)
return
}
s.delPod(pod)

// release node lock if this pod owned one on a best-effort basis.
// this is safe because ReleaseNodeLock checks the lock owner and no-ops if different.
nodeName := pod.Annotations[util.AssignedNodeAnnotations]
nodeName := ""
if pod.Annotations != nil {
nodeName = pod.Annotations[util.AssignedNodeAnnotations]
}
if nodeName == "" {
return
}
Expand Down