PlacementDecision events are recorded even when there are no changes

[haoqing0110]

Describe the bug
The scheduling controller records Kubernetes Events for PlacementDecision even when there are no actual changes, causing unnecessary event accumulation.

Current Behavior

File: pkg/placement/controllers/scheduling/scheduling_controller.go
Function: createOrUpdatePlacementDecision()
Lines: 711-754

updated, err := placementDecisionPatcher.PatchStatus(ctx, newPlacementDecision, newPlacementDecision.Status, existPlacementDecision.Status)
if updated {
return err // Early return when status is updated
}

// Return value ignored - cannot tell if labels actually changed
_, err = placementDecisionPatcher.PatchLabelAnnotations(ctx, newPlacementDecision, newPlacementDecision.ObjectMeta, existPlacementDecision.ObjectMeta)
if err != nil {
return err
}

// ❌ PROBLEM: Events are ALWAYS recorded here, even when nothing changed
c.eventsRecorder.Eventf(..., "DecisionUpdate", "DecisionUpdated", ...)
c.eventsRecorder.Eventf(..., "ScoreUpdate", "ScoreUpdated", ...)

To Reproduce
Steps to reproduce the behavior:

Expected behavior
A clear and concise description of what you expected to happen.

Environment ie: OCM version, Kubernetes version and provider:

Additional context
Add any other context about the problem here.

[dosubot]

You're right—events like "DecisionUpdate" and "ScoreUpdate" are always recorded in createOrUpdatePlacementDecision, even when nothing actually changes. This happens because the function ignores the boolean return value from PatchLabelAnnotations, so it can't tell if labels or annotations were updated. As a result, events are emitted unconditionally after patching, which leads to unnecessary event accumulation and noise in the event log.

To avoid this, you could check the return value of PatchLabelAnnotations and only record events if either the status or labels/annotations actually changed. This would suppress redundant events when the PlacementDecision is already up to date. The current logic and this improvement opportunity are visible in the code here: scheduling_controller.go#L670-L754. No existing issues or PRs address this yet.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}