Skip to content

[Data] Don't downscale actors if the operator hasn't received any inputs #59883

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
bveeramani merged 5 commits into from
Jan 6, 2026
Merged

[Data] Don't downscale actors if the operator hasn't received any inputs #59883

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
9545f20
Initial commit
bveeramani Jan 6, 2026
9732fbd
Merge branch 'master' into no-downscale-if-no-inputs
bveeramani Jan 6, 2026
aa6c735
Address review comments
bveeramani Jan 6, 2026
195f73f
Apply suggestions from code review
owenowenisme Jan 6, 2026
245e5d0
Merge branch 'no-downscale-if-no-inputs' of https://github.com/ray-pr…
bveeramani Jan 6, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions python/ray/data/_internal/actor_autoscaler/default_actor_autoscaler.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,6 +122,11 @@ def _derive_target_scaling_config(
reason="pool exceeding max size",
)

# To prevent unexpected downscaling from the initial size, short-circuit if
# the operator hasn't received any inputs.
if op.metrics.num_inputs_received == 0:
return ActorPoolScalingRequest.no_op(reason="no inputs received")

# Determine whether to scale up based on the actor pool utilization.
util = self._compute_utilization(actor_pool)

Expand Down
9 changes: 8 additions & 1 deletion python/ray/data/tests/test_autoscaler.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,7 +67,7 @@ def test_actor_pool_scaling():
_inputs_complete=False,
input_dependencies=[MagicMock()],
internal_input_queue_num_blocks=MagicMock(return_value=1),
metrics=MagicMock(average_num_inputs_per_task=1),
metrics=MagicMock(average_num_inputs_per_task=1, num_inputs_received=1),
)
op_state = OpState(
op, inqueues=[MagicMock(__len__=MagicMock(return_value=10), num_blocks=10)]
Expand Down Expand Up @@ -217,6 +217,13 @@ def assert_autoscaling_action(
expected_reason="exceeded resource limits",
)

# Should no-op because the op has not received any inputs.
with patch(op.metrics, "num_inputs_received", 0, is_method=False):
assert_autoscaling_action(
delta=0,
expected_reason="no inputs received",
)


@pytest.fixture
def autoscaler_max_upscaling_delta_setup():
Expand Down