xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-14 05:35:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] Nonzero exit code if MQLLMEngine startup fails (#8572)

Browse Source
This commit is contained in:
Nick Hill 2024-09-18 21:17:55 +01:00 committed by GitHub
parent db9120cded
commit d9cd78eb71
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 5 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
vllm/entrypoints/openai/api_server.py
View File

@ -11,7 +11,7 @@ from argparse import Namespace
from contextlib import asynccontextmanager from contextlib import asynccontextmanager
from functools import partial from functools import partial
from http import HTTPStatus from http import HTTPStatus
from typing import AsyncIterator, Optional, Set from typing import AsyncIterator, Set
import uvloop import uvloop
from fastapi import APIRouter, FastAPI, Request from fastapi import APIRouter, FastAPI, Request
@ -95,7 +95,7 @@ async def lifespan(app: FastAPI):
@asynccontextmanager @asynccontextmanager
async def build_async_engine_client( async def build_async_engine_client(
args: Namespace) -> AsyncIterator[Optional[EngineClient]]: args: Namespace) -> AsyncIterator[EngineClient]:
# Context manager to handle engine_client lifecycle # Context manager to handle engine_client lifecycle
# Ensures everything is shutdown and cleaned up on error/exit # Ensures everything is shutdown and cleaned up on error/exit
@ -110,7 +110,7 @@ async def build_async_engine_client(
async def build_async_engine_client_from_engine_args( async def build_async_engine_client_from_engine_args(
engine_args: AsyncEngineArgs, engine_args: AsyncEngineArgs,
disable_frontend_multiprocessing: bool = False, disable_frontend_multiprocessing: bool = False,
) -> AsyncIterator[Optional[EngineClient]]: ) -> AsyncIterator[EngineClient]:
""" """
Create EngineClient, either: Create EngineClient, either:
- in-process using the AsyncLLMEngine Directly - in-process using the AsyncLLMEngine Directly
@ -188,10 +188,8 @@ async def build_async_engine_client_from_engine_args(
break break
except TimeoutError: except TimeoutError:
if not engine_process.is_alive(): if not engine_process.is_alive():
logger.error("Engine process died before responding " raise RuntimeError(
"to readiness probe") "Engine process failed to start") from None
yield None
return
yield mp_engine_client # type: ignore[misc] yield mp_engine_client # type: ignore[misc]
finally: finally:
@ -532,10 +530,6 @@ async def run_server(args, **uvicorn_kwargs) -> None:
signal.signal(signal.SIGTERM, signal_handler) signal.signal(signal.SIGTERM, signal_handler)
async with build_async_engine_client(args) as engine_client: async with build_async_engine_client(args) as engine_client:
# If None, creation of the client failed and we exit.
if engine_client is None:
return
app = build_app(args) app = build_app(args)
model_config = await engine_client.get_model_config() model_config = await engine_client.get_model_config()