Update flags, Improve tests readability & use Bypass instead of ignor…

…e in naming

cluster-autoscaler/config/autoscaling_options.go

@@ -277,11 +277,6 @@ type AutoscalingOptions struct {
 	// dynamicNodeDeleteDelayAfterTaintEnabled is used to enable/disable dynamic adjustment of NodeDeleteDelayAfterTaint
 	// based on the latency between the CA and the api-server
 	DynamicNodeDeleteDelayAfterTaintEnabled bool
-	//IgnoreSchedulerProcessing is used to signal whether CA will/won't wait
-	//for scheduler to mark pods as unschedulable and will process both marked & non-marked pods
-	//it will also signal whether we enable/disable waiting for pod time buffers before triggering a scale-up.
-	IgnoreSchedulerProcessing bool
-	//IgnoredSchedulers are used to specify which schedulers to ignore their processing
-	//if IgnoreSchedulerProcessing is set to true
-	IgnoredSchedulers map[string]bool
+	// BypassedSchedulers are used to specify which schedulers to bypass their processing
+	BypassedSchedulers map[string]bool
 }

cluster-autoscaler/core/podlistprocessor/filter_out_expendable.go

@@ -36,12 +36,14 @@ func NewFilterOutExpendablePodListProcessor() *filterOutExpendable {
 func (p *filterOutExpendable) Process(context *context.AutoscalingContext, pods []*apiv1.Pod) ([]*apiv1.Pod, error) {
 	nodes, err := context.AllNodeLister().List()
 	if err != nil {
+		klog.Warningf("Failed to list all nodes while filtering expendable: %v", err)
 		return nil, err
 	}
 	expendablePodsPriorityCutoff := context.AutoscalingOptions.ExpendablePodsPriorityCutoff
 
 	unschedulablePods, waitingForLowerPriorityPreemption := core_utils.FilterOutExpendableAndSplit(pods, nodes, expendablePodsPriorityCutoff)
 	if err = p.addPreemptingPodsToSnapshot(waitingForLowerPriorityPreemption, context); err != nil {
+		klog.Warningf("Failed to add preempting pods to snapshot: %v", err)
 		return nil, err
 	}
 

cluster-autoscaler/core/static_autoscaler.go

@@ -310,8 +310,9 @@ func (a *StaticAutoscaler) RunOnce(currentTime time.Time) caerrors.AutoscalerErr
 	}
 	originalScheduledPods, unschedulablePods := kube_util.ScheduledPods(pods), kube_util.UnschedulablePods(pods)
 	schedulerUnprocessed := make([]*apiv1.Pod, 0, 0)
-	if a.IgnoreSchedulerProcessing {
-		schedulerUnprocessed = kube_util.SchedulerUnprocessedPods(pods, a.IgnoredSchedulers)
+	isSchedulerProcessingIgnored := len(a.BypassedSchedulers) > 0
+	if isSchedulerProcessingIgnored {
+		schedulerUnprocessed = kube_util.SchedulerUnprocessedPods(pods, a.BypassedSchedulers)
 	}
 
 	// Update cluster resource usage metrics
@@ -456,7 +457,7 @@ func (a *StaticAutoscaler) RunOnce(currentTime time.Time) caerrors.AutoscalerErr
 
 	// SchedulerUnprocessed might be zero here if it was disabled
 	metrics.UpdateUnschedulablePodsCount(len(unschedulablePods), len(schedulerUnprocessed))
-	if a.IgnoreSchedulerProcessing {
+	if isSchedulerProcessingIgnored {
 		// Treat unknown pods as unschedulable, pod list processor will remove schedulable pods
 		unschedulablePods = append(unschedulablePods, schedulerUnprocessed...)
 	}
@@ -543,7 +544,7 @@ func (a *StaticAutoscaler) RunOnce(currentTime time.Time) caerrors.AutoscalerErr
 	} else if a.MaxNodesTotal > 0 && len(readyNodes) >= a.MaxNodesTotal {
 		scaleUpStatus.Result = status.ScaleUpNoOptionsAvailable
 		klog.V(1).Info("Max total nodes in cluster reached")
-	} else if !a.IgnoreSchedulerProcessing && allPodsAreNew(unschedulablePodsToHelp, currentTime) {
+	} else if !isSchedulerProcessingIgnored && allPodsAreNew(unschedulablePodsToHelp, currentTime) {
 		// The assumption here is that these pods have been created very recently and probably there
 		// is more pods to come. In theory we could check the newest pod time but then if pod were created
 		// slowly but at the pace of 1 every 2 seconds then no scale up would be triggered for long time.