[Failover][BugFix] Fix duplicate requests caused by failover

llumnix/backends/vllm/llm_engine.py

@@ -251,9 +251,6 @@ def add_request(self,
                     server_info: ServerInfo,
                     *args,
                     **kwargs) -> None:
-        # When manager is unavailable, api server might dispatch the request that has already been dispatched.
-        if request_id in self.engine.request_server_info:
-            return
         # Store the server information of each request to put the request outputs back to the corresponding api server correctly.
         self.engine.request_server_info[request_id] = server_info
         self.engine.add_request(request_id, *args, **kwargs)

llumnix/entrypoints/vllm/api_server.py

@@ -46,6 +46,7 @@
 log_requests = None
 num_finished_requests = 0
 WAIT_MANAGER_INTERVAL = 5
+manager_available = True
 
 
 async def _background_process_outputs():
@@ -78,10 +79,17 @@ async def manager_generate(prompt, sampling_params, request_id) -> AsyncStream:
     # This request's outputs will be put to the request_output_queue of this api server no matter which instance it's running in.
     server_info = ServerInfo(server_id, request_output_queue)
     # If manager is unavailable, request will be directly added to the llumlet held by api server.
+    global manager_available
     try:
         # await to catch exception
         await engine_manager.generate.remote(request_id, server_info, prompt, sampling_params)
+        manager_available = True
     except ray.exceptions.RayActorError:
+        # Do not re-generate the request to avoid duplicate requests.
+        if manager_available:
+            manager_available = False
+            return results_generator
+
         try:
             if instance_num_requests:
                 instance_id = min(instance_num_requests, key=instance_num_requests.get)

llumnix/llm_engine_manager.py

@@ -122,8 +122,6 @@ async def generate(
         except (ray.exceptions.RayActorError, KeyError):
             logger.info("[generate] instance {} is dead, regenerate request {}".format(instance_id, request_id))
             self.scale_down(instance_id)
-            if self.num_instances != 0:
-                asyncio.create_task(self.generate(request_id, server_info, *args, **kwargs))
 
     async def abort(self, request_id: Union[str, Iterable[str]]) -> None:
         if isinstance(request_id, str):