CA: use ForceSchedulePod() instead of SchedulePod() in filterOutExpen…

…dable

This is done to preserve the previous behavior - the predicates weren't
checked here before the refactor.

We need to figure out how to integrate this with DRA - left a TODO.

cluster-autoscaler/core/podlistprocessor/filter_out_expendable.go

@@ -56,7 +56,8 @@ func (p *filterOutExpendable) Process(context *context.AutoscalingContext, pods
 // CA logic from before migration to scheduler framework. So let's keep it for now
 func (p *filterOutExpendable) addPreemptingPodsToSnapshot(pods []*apiv1.Pod, ctx *context.AutoscalingContext) error {
 	for _, p := range pods {
-		if err := ctx.ClusterSnapshot.SchedulePod(p, p.Status.NominatedNodeName); err != nil {
+		// TODO(DRA): Figure out if/how to use the predicate-checking SchedulePod() here instead - otherwise this doesn't work with DRA pods.
+		if err := ctx.ClusterSnapshot.ForceSchedulePod(p, p.Status.NominatedNodeName); err != nil {
 			klog.Errorf("Failed to update snapshot with pod %s/%s waiting for preemption: %v", p.Namespace, p.Name, err)
 			return caerrors.ToAutoscalerError(caerrors.InternalError, err)
 		}