Port metrics from aioprometheus to prometheus_client (#2730)

vllm/entrypoints/openai/api_server.py

@@ -6,8 +6,7 @@ import os
 import importlib
 import inspect
 
-from aioprometheus import MetricsMiddleware
-from aioprometheus.asgi.starlette import metrics
+from prometheus_client import make_asgi_app
 import fastapi
 import uvicorn
 from http import HTTPStatus
@@ -18,7 +17,6 @@ from fastapi.responses import JSONResponse, StreamingResponse, Response
 
 from vllm.engine.arg_utils import AsyncEngineArgs
 from vllm.engine.async_llm_engine import AsyncLLMEngine
-from vllm.engine.metrics import add_global_metrics_labels
 from vllm.entrypoints.openai.protocol import CompletionRequest, ChatCompletionRequest, ErrorResponse
 from vllm.logger import init_logger
 from vllm.entrypoints.openai.serving_chat import OpenAIServingChat
@@ -141,8 +139,9 @@ def parse_args():
     return parser.parse_args()
 
 
-app.add_middleware(MetricsMiddleware)  # Trace HTTP server metrics
-app.add_route("/metrics", metrics)  # Exposes HTTP metrics
+# Add prometheus asgi middleware to route /metrics requests
+metrics_app = make_asgi_app()
+app.mount("/metrics", metrics_app)
 
 
 @app.exception_handler(RequestValidationError)
@@ -242,9 +241,6 @@ if __name__ == "__main__":
     openai_serving_completion = OpenAIServingCompletion(
         engine, served_model, args.lora_modules)
 
-    # Register labels for metrics
-    add_global_metrics_labels(model_name=engine_args.model)
-
     app.root_path = args.root_path
     uvicorn.run(app,
                 host=args.host,