[Core] Priority-based scheduling in async engine (#8850)

vllm/engine/async_llm_engine.py

@@ -420,6 +420,7 @@ class _AsyncLLMEngine(LLMEngine):
         lora_request: Optional[LoRARequest] = None,
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
+        priority: int = 0,
     ) -> None:
         ...
 
@@ -433,6 +434,7 @@ class _AsyncLLMEngine(LLMEngine):
         lora_request: Optional[LoRARequest] = None,
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
+        priority: int = 0,
     ) -> None:
         ...
 
@@ -449,6 +451,7 @@ class _AsyncLLMEngine(LLMEngine):
             lora_request: Optional[LoRARequest] = None,
             trace_headers: Optional[Mapping[str, str]] = None,
             prompt_adapter_request: Optional[PromptAdapterRequest] = None,
+            priority: int = 0,
             *,
             inputs: Optional[PromptType] = None,  # DEPRECATED
     ) -> None:
@@ -460,6 +463,9 @@ class _AsyncLLMEngine(LLMEngine):
         if lora_request is not None and not self.lora_config:
             raise ValueError(f"Got lora_request {lora_request} but LoRA is "
                              "not enabled!")
+        if priority != 0 and not self.scheduler_config.policy == "priority":
+            raise ValueError(f"Got priority {priority} but "
+                             "Priority scheduling is not enabled.")
         if arrival_time is None:
             arrival_time = time.time()
 
@@ -479,6 +485,7 @@ class _AsyncLLMEngine(LLMEngine):
             lora_request=lora_request,
             prompt_adapter_request=prompt_adapter_request,
             trace_headers=trace_headers,
+            priority=priority,
         )
 
     async def check_health_async(self) -> None:
@@ -829,6 +836,7 @@ class AsyncLLMEngine:
         lora_request: Optional[LoRARequest] = None,
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
+        priority: int = 0,
     ) -> Coroutine[None, None, AsyncGenerator[Union[
             RequestOutput, EmbeddingRequestOutput], None]]:
         ...
@@ -843,6 +851,7 @@ class AsyncLLMEngine:
         lora_request: Optional[LoRARequest] = None,
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
+        priority: int = 0,
     ) -> Coroutine[None, None, AsyncGenerator[Union[
             RequestOutput, EmbeddingRequestOutput], None]]:
         ...
@@ -860,6 +869,7 @@ class AsyncLLMEngine:
         lora_request: Optional[LoRARequest] = None,
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
+        priority: int = 0,
         *,
         inputs: Optional[PromptType] = None,  # DEPRECATED
     ) -> AsyncGenerator[Union[RequestOutput, EmbeddingRequestOutput], None]:
@@ -877,6 +887,11 @@ class AsyncLLMEngine:
                     "error that caused the background loop to stop "
                     "(AsyncEngineDeadError).")
 
+        if (priority != 0
+                and not self.engine.scheduler_config.policy == "priority"):
+            raise ValueError(f"Got priority {priority} but "
+                             "Priority scheduling is not enabled.")
+
         stream = self._request_tracker.add_request(
             request_id,
             verbose=self.log_requests,
@@ -885,7 +900,9 @@ class AsyncLLMEngine:
             arrival_time=arrival_time or time.time(),
             lora_request=lora_request,
             trace_headers=trace_headers,
-            prompt_adapter_request=prompt_adapter_request)
+            prompt_adapter_request=prompt_adapter_request,
+            priority=priority,
+        )
 
         return stream.generator()
 
@@ -896,7 +913,8 @@ class AsyncLLMEngine:
         request_id: str,
         lora_request: Optional[LoRARequest] = None,
         trace_headers: Optional[Mapping[str, str]] = None,
-        prompt_adapter_request: Optional[PromptAdapterRequest] = None
+        prompt_adapter_request: Optional[PromptAdapterRequest] = None,
+        priority: int = 0,
     ) -> AsyncGenerator[RequestOutput, None]:
         """Generate outputs for a request.
 
@@ -913,6 +931,8 @@ class AsyncLLMEngine:
             trace_headers: OpenTelemetry trace headers.
             prompt_adapter_request: Prompt Adapter request to use
                                             for generation, if any.
+            priority: The priority of the request.
+                Only applicable with priority scheduling.
 
         Yields:
             The output `RequestOutput` objects from the LLMEngine
@@ -968,6 +988,7 @@ class AsyncLLMEngine:
                 lora_request=lora_request,
                 trace_headers=trace_headers,
                 prompt_adapter_request=prompt_adapter_request,
+                priority=priority,
         ):
             yield LLMEngine.validate_output(output, RequestOutput)
 

vllm/engine/llm_engine.py

@@ -796,7 +796,7 @@ class LLMEngine:
             raise ValueError(f"Got lora_request {lora_request} but LoRA is "
                              "not enabled!")
 
-        if priority > 0 and not self.scheduler_config.policy == "priority":
+        if priority != 0 and not self.scheduler_config.policy == "priority":
             raise ValueError(f"Got priority {priority} but "
                              "Priority scheduling is not enabled.")