From c49c43d51cdcd1c61182c2182bc87458d6264f19 Mon Sep 17 00:00:00 2001
From: chenjian <1435317881@qq.com>
Date: Mon, 1 Sep 2025 14:06:09 +0800
Subject: [PATCH] [Bug fix] Fix perf in mixed deployment with yiyan adpater
 (#3703)

Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com>
---
 fastdeploy/output/token_processor.py    | 3 +++
 fastdeploy/scheduler/local_scheduler.py | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/fastdeploy/output/token_processor.py b/fastdeploy/output/token_processor.py
index dcf889d29..c5511104f 100644
--- a/fastdeploy/output/token_processor.py
+++ b/fastdeploy/output/token_processor.py
@@ -360,6 +360,7 @@ class TokenProcessor:
                 metrics = RequestMetrics(
                     arrival_time=task.arrival_time,
                     inference_start_time=task.inference_start_time,
+                    model_execute_time=time.time() - task.inference_start_time,
                     first_token_time=time.time() - task.inference_start_time,
                     time_in_queue=task.schedule_start_time - task.preprocess_end_time,
                     preprocess_cost_time=task.preprocess_end_time - task.preprocess_start_time,
@@ -503,6 +504,7 @@ class TokenProcessor:
                 metrics = RequestMetrics(
                     arrival_time=task.arrival_time,
                     inference_start_time=task.inference_start_time,
+                    model_execute_time=time.time() - task.inference_start_time,
                     first_token_time=time.time() - task.inference_start_time,
                     time_in_queue=task.schedule_start_time - task.preprocess_end_time,
                     preprocess_cost_time=task.preprocess_end_time - task.preprocess_start_time,
@@ -514,6 +516,7 @@ class TokenProcessor:
             else:
                 metrics = RequestMetrics(
                     arrival_time=time.time(),
+                    model_execute_time=time.time() - task.inference_start_time,
                     request_start_time=task.arrival_time,
                 )
             self.number_of_output_tokens += len(token_ids)
diff --git a/fastdeploy/scheduler/local_scheduler.py b/fastdeploy/scheduler/local_scheduler.py
index 20e53317b..159dd447d 100644
--- a/fastdeploy/scheduler/local_scheduler.py
+++ b/fastdeploy/scheduler/local_scheduler.py
@@ -209,7 +209,7 @@ class LocalScheduler:
         return (token_num + block_size - 1) // block_size
 
     def get_unhandled_request_num(self):
-        return len(self.requests)
+        return len(self.ids) - self.ids_read_cursor
 
     def get_requests(
         self,