diff --git a/python/ray/tests/test_autoscaler_fake_scaledown.py b/python/ray/tests/test_autoscaler_fake_scaledown.py
index 8b8e71c10..98a094b1e 100644
--- a/python/ray/tests/test_autoscaler_fake_scaledown.py
+++ b/python/ray/tests/test_autoscaler_fake_scaledown.py
@@ -28,7 +28,7 @@ def test_scaledown_shared_objects(shutdown_only):
     )
 
     try:
-        cluster.start(_system_config={"scheduler_report_pinned_bytes_only": True})
+        cluster.start()
         ray.init("auto")
 
         # Triggers the addition of a GPU node.
diff --git a/src/ray/common/ray_config_def.h b/src/ray/common/ray_config_def.h
index e4e22f0ea..91a3bfe14 100644
--- a/src/ray/common/ray_config_def.h
+++ b/src/ray/common/ray_config_def.h
@@ -140,7 +140,7 @@ RAY_CONFIG(float, scheduler_spread_threshold, 0.5);
 /// Whether to only report the usage of pinned copies of objects in the
 /// object_store_memory resource. This means nodes holding secondary copies only
 /// will become eligible for removal in the autoscaler.
-RAY_CONFIG(bool, scheduler_report_pinned_bytes_only, false)
+RAY_CONFIG(bool, scheduler_report_pinned_bytes_only, true)
 
 // The max allowed size in bytes of a return object from direct actor calls.
 // Objects larger than this size will be spilled/promoted to plasma.