Skip to content

fix(pd): disable overlap for spec+grammar in disagg decode loop #28039

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
hnyls2002 merged 2 commits into from
Jun 12, 2026
Merged

fix(pd): disable overlap for spec+grammar in disagg decode loop #28039

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2d7e01b
fix(pd): disable overlap for spec+grammar in disagg decode loop
Jun 12, 2026
189cc54
add json grammar kit to disagg spec test
hnyls2002 Jun 12, 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
13 changes: 11 additions & 2 deletions python/sglang/srt/disaggregation/decode.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1750,6 +1750,10 @@ def event_loop_overlap_disagg_decode(self: Scheduler):
self.result_queue = deque()
self.last_batch: Optional[ScheduleBatch] = None

def pop_and_process():
tmp_batch, tmp_result = self.result_queue.popleft()
self.process_batch_result(tmp_batch, tmp_result)

while True:
# Receive requests
recv_reqs = self.request_receiver.recv_requests()
Expand All @@ -1765,6 +1769,11 @@ def event_loop_overlap_disagg_decode(self: Scheduler):
# Get the next batch to run
batch = self.get_next_disagg_decode_batch_to_run()
self.cur_batch = batch
# overlap + spec + grammar is unsupported (would desync DP ranks).
disable_overlap_for_batch = self.is_disable_overlap_for_batch(batch)

if disable_overlap_for_batch and self.last_batch:
pop_and_process()

# Launch the current batch
if batch:
Expand All @@ -1775,8 +1784,8 @@ def event_loop_overlap_disagg_decode(self: Scheduler):

# Process the last batch
if self.last_batch:
tmp_batch, tmp_result = self.result_queue.popleft()
self.process_batch_result(tmp_batch, tmp_result)
if not disable_overlap_for_batch:
pop_and_process()
elif batch is None:
self.on_idle()

Expand Down
5 changes: 3 additions & 2 deletions test/registered/disaggregation/test_disaggregation_basic.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@
from transformers import AutoTokenizer

from sglang.test.ci.ci_register import register_cuda_ci
from sglang.test.kits.json_constrained_kit import JSONConstrainedMixin
from sglang.test.kits.pause_generation_kit import PauseResumeInPlaceMixin
from sglang.test.run_eval import run_eval
from sglang.test.server_fixtures.disaggregation_fixture import (
Expand All @@ -21,7 +22,7 @@
DEFAULT_TARGET_MODEL_EAGLE3,
)

register_cuda_ci(est_time=509, stage="base-b", runner_config="2-gpu-large")
register_cuda_ci(est_time=560, stage="base-b", runner_config="2-gpu-large")


class TestDisaggregationAccuracy(PauseResumeInPlaceMixin, PDDisaggregationServerBase):
Expand Down Expand Up @@ -215,7 +216,7 @@ def test_gsm8k(self):
raise e from health_check_error


class TestDisaggregationMooncakeSpec(PDDisaggregationServerBase):
class TestDisaggregationMooncakeSpec(JSONConstrainedMixin, PDDisaggregationServerBase):
@classmethod
def setUpClass(cls):
super().setUpClass()
Expand Down
Loading