Fix scaleTarget name reference when resolving pods (#1483)

* fix object name reference when resolving pods

Signed-off-by: Zbynek Roubalik <zroubali@redhat.com>

* update changelog

Signed-off-by: Zbynek Roubalik <zroubali@redhat.com>

CHANGELOG.md

@@ -32,7 +32,7 @@
 - Optimize Kafka scaler's `getLagForPartition` function. ([#1464](https://github.com/kedacore/keda/pull/1464))
 - Reduce unnecessary /scale requests from ScaledObject controller ([#1453](https://github.com/kedacore/keda/pull/1453))
 - Add support for the WATCH_NAMESPACE environment variable to the operator ([#1474](https://github.com/kedacore/keda/pull/1474))
-- Automatically determine the RabbitMQ protocol when possible, and support setting the protocl via TriggerAuthentication ([#1459](https://github.com/kedacore/keda/pulls/1459))
+- Automatically determine the RabbitMQ protocol when possible, and support setting the protocl via TriggerAuthentication ([#1459](https://github.com/kedacore/keda/pulls/1459),[#1483](https://github.com/kedacore/keda/pull/1483))
 - Improve performance when fetching pod information ([#1457](https://github.com/kedacore/keda/pull/1457))
 
 ### Breaking Changes

pkg/scaling/scale_handler.go

@@ -377,7 +377,7 @@ func (h *scaleHandler) getPods(scalableObject interface{}) (*corev1.PodTemplateS
 		// Try to get a real object instance for better cache usage, but fall back to an Unstructured if needed.
 		podTemplateSpec := corev1.PodTemplateSpec{}
 		gvk := obj.Status.ScaleTargetGVKR.GroupVersionKind()
-		objKey := client.ObjectKey{Namespace: obj.Namespace, Name: obj.Status.ScaleTargetGVKR.Resource}
+		objKey := client.ObjectKey{Namespace: obj.Namespace, Name: obj.Spec.ScaleTargetRef.Name}
 		switch {
 		// For core types, use a typed client so we get an informer-cache-backed Get to reduce API load.
 		case gvk.Group == "apps" && gvk.Kind == "Deployment":