[Misc] Add argument to disable FastAPI docs (#8554)

vllm/entrypoints/openai/api_server.py

@@ -417,6 +417,12 @@ if envs.VLLM_ALLOW_RUNTIME_LORA_UPDATING:
 
 
 def build_app(args: Namespace) -> FastAPI:
+    if args.disable_fastapi_docs:
+        app = FastAPI(openapi_url=None,
+                      docs_url=None,
+                      redoc_url=None,
+                      lifespan=lifespan)
+    else:
         app = FastAPI(lifespan=lifespan)
     app.include_router(router)
     app.root_path = args.root_path

vllm/entrypoints/openai/cli_args.py

@@ -190,6 +190,13 @@ def make_arg_parser(parser: FlexibleArgumentParser) -> FlexibleArgumentParser:
                         'ID numbers being printed in log.'
                         '\n\nDefault: Unlimited')
 
+    parser.add_argument(
+        "--disable-fastapi-docs",
+        action='store_true',
+        default=False,
+        help="Disable FastAPI's OpenAPI schema, Swagger UI, and ReDoc endpoint"
+    )
+
     return parser