[Feat] Add received_time in serving_base (#13432)

Signed-off-by: zhanghaotong <zhanghaotong.zht@antgroup.com>
This commit is contained in:
zhanghaotong
2025-12-09 21:00:24 +08:00
committed by GitHub
parent cef5ba65b1
commit fe7f91ef82
3 changed files with 16 additions and 1 deletions

View File

@@ -204,6 +204,12 @@ class GenerateReqInput(BaseReq):
# Validation step duration
validation_time: Optional[float] = None
# For metrics
received_time: Optional[float] = None
# Perf_counter equivalents for accurate time calculations
received_time_perf: Optional[float] = None
# For data parallel rank routing
data_parallel_rank: Optional[int] = None

View File

@@ -431,7 +431,7 @@ class TokenizerManager(TokenizerCommunicatorMixin):
obj: Union[GenerateReqInput, EmbeddingReqInput],
request: Optional[fastapi.Request] = None,
):
created_time = time.time()
created_time = obj.received_time if obj.received_time else time.time()
self.auto_create_handle_loop()
obj.normalize_batch_and_arguments()