[BugFix] prevent requests from entering running state without a slot (#7141)

* [fix] prevent requests from entering running state without a slot

* [fix] count abort set

* [fix] count preempted task in waiting list

fastdeploy/engine/common_engine.py

@@ -1152,8 +1152,7 @@ class EngineService:
                     time.sleep(0.005)
 
             except RuntimeError as e:
-                if "cannot schedule new futures after shutdown" in str(e):
-                    break
+                raise e
             except Exception as e:
                 err_msg = "Error happened while insert task to engine: {}, {}.".format(e, str(traceback.format_exc()))
                 self.llm_logger.error(err_msg)

fastdeploy/engine/sched/resource_manager_v1.py

@@ -941,7 +941,13 @@ class ResourceManagerV1(ResourceManager):
             if not preempted_reqs:
                 skip_requests: list[Request] = []
                 while self.waiting and token_budget > 0:
-                    if len(self.running) == self.max_num_seqs:
+                    if (
+                        len(self.running)
+                        + len(self.to_be_rescheduled_request_id_set)
+                        + len(self.to_be_aborted_req_id_set)
+                        + sum([req.status == RequestStatus.PREEMPTED for req in self.waiting])
+                        >= self.max_num_seqs
+                    ):
                         break
 
                     request = self.waiting[0]