diff --git a/vllm/entrypoints/openai/rpc/server.py b/vllm/entrypoints/openai/rpc/server.py index af406d87..471d6263 100644 --- a/vllm/entrypoints/openai/rpc/server.py +++ b/vllm/entrypoints/openai/rpc/server.py @@ -38,6 +38,8 @@ class AsyncEngineRPCServer: self.socket.close() self.context.destroy() self.engine.shutdown_background_loop() + # Clear the engine reference so that it can be GC'ed. + self.engine = None async def get_model_config(self, identity): """Send the ModelConfig"""