fix(grpc): send last chunk before completion during streaming (#19895)

This commit is contained in:
Chang Su
2026-03-04 13:21:21 -08:00
committed by GitHub
parent 329817e262
commit 0ee9d3c8e9

View File

@@ -227,12 +227,17 @@ class SGLangSchedulerServicer(sglang_scheduler_pb2_grpc.SglangSchedulerServicer)
),
),
)
elif output.get("finished", False):
elif request.stream:
yield self._create_chunk_response(request.request_id, output)
if output.get("finished", False):
yield self._create_completion_response(
request.request_id, output
)
else:
# Non-streaming n=1: single completion response
yield self._create_completion_response(
request.request_id, output
)
else:
yield self._create_chunk_response(request.request_id, output)
except Exception as e:
logger.error(