Skip to content

fix: prevent orphan VIT model processes on crash or interrupt #1256

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
sufubao wants to merge 2 commits into
base: main
Choose a base branch
from
+27 −5
Open

fix: prevent orphan VIT model processes on crash or interrupt #1256

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8d29866
fix: prevent orphan VIT model processes on crash or interrupt
sufubao Apr 2, 2026
539e733
fix again
sufubao Apr 3, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 12 additions & 2 deletions lightllm/server/visualserver/manager.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,7 @@
import setproctitle
import threading
import collections
import multiprocessing
from typing import List
from lightllm.server.core.objs.io_objs.group_req import GroupReqIndexes
from lightllm.server.core.objs import ShmReqManager, StartArgs
Expand Down Expand Up @@ -62,12 +63,14 @@ def __init__(
async def wait_to_model_ready(self):

self.model_rpcs: List[List[VisualModelRpcClient]] = [[] for _ in range(self.vit_dp)]
self.model_procs: List[multiprocessing.Process] = []
self.vit_attn_backend = init_vit_att_backend(index=0)
for dp_rank_id in range(self.vit_dp):
for tp_rank_id in range(self.vit_tp):

rpc_model = await start_model_process()
rpc_model, proc = await start_model_process()
self.model_rpcs[dp_rank_id].append(rpc_model)
self.model_procs.append(proc)

init_model_ret = []
for dp_rank_id in range(self.vit_dp): # async init model process
Expand Down Expand Up @@ -187,7 +190,14 @@ async def loop_for_netio_req(self):
logger.exception(str(e))

def clean_up(self):
return
for proc in getattr(self, "model_procs", []):
try:
if proc.is_alive():
logger.info(f"Killing VIT model process {proc.pid}")
proc.kill()
proc.join(timeout=5)
except (ProcessLookupError, OSError):
pass
Comment on lines +193 to +200
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist bot Apr 3, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The current cleanup logic kills and joins processes sequentially. If multiple processes are unresponsive, the cumulative timeout (5 seconds per process) could significantly delay the shutdown of the visual server. It is more efficient to signal all processes to terminate first, and then join them.

Suggested change
for proc in getattr(self, "model_procs", []):
try:
if proc.is_alive():
logger.info(f"Killing VIT model process {proc.pid}")
proc.kill()
proc.join(timeout=5)
except (ProcessLookupError, OSError):
pass
procs = getattr(self, "model_procs", [])
for proc in procs:
try:
if proc.is_alive():
logger.info(f"Killing VIT model process {proc.pid}")
proc.kill()
except (ProcessLookupError, OSError):
pass
for proc in procs:
try:
proc.join(timeout=5)
except (ProcessLookupError, OSError):
pass



def start_visual_process(args, pipe_writer):
Expand Down
4 changes: 3 additions & 1 deletion lightllm/server/visualserver/model_infer/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,7 @@
from rpyc.utils.classic import obtain
from rpyc.utils.server import ThreadedServer
from lightllm.utils.graceful_utils import graceful_registry
from lightllm.utils.process_check import start_parent_check_thread
from lightllm.utils.envs_utils import get_env_start_args
from .model_rpc_client import VisualModelRpcClient
from .model_rpc import VisualModelRpcServer
Expand All @@ -18,6 +19,7 @@
def _init_env(socket_path: str, success_event):
# 注册graceful 退出的处理
graceful_registry(inspect.currentframe().f_code.co_name)
start_parent_check_thread()

import lightllm.utils.rpyc_fix_utils as _

Expand Down Expand Up @@ -52,7 +54,7 @@ async def start_model_process():
# 服务端需要调用客户端传入的event所以，客户端需要一个后台线程进行相关的处理。
conn._bg_thread = rpyc.BgServingThread(conn, sleep_interval=0.001)

return VisualModelRpcClient(conn)
return VisualModelRpcClient(conn), proc


def _generate_unix_socket_path() -> str:
Expand Down
14 changes: 12 additions & 2 deletions lightllm/server/visualserver/visual_only_manager.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,6 +12,7 @@
import os
import signal
import time
import multiprocessing
from lightllm.utils.net_utils import get_hostname_ip
from .objs import VIT_Obj
from typing import List
Expand Down Expand Up @@ -94,12 +95,14 @@ async def register_to_config_server_loop(self, args: StartArgs):
async def wait_to_model_ready(self):

self.model_rpcs: List[List[VisualModelRpcClient]] = [[] for _ in range(self.vit_dp)]
self.model_procs: List[multiprocessing.Process] = []
self.vit_attn_backend = init_vit_att_backend(index=0)
for dp_rank_id in range(self.vit_dp):
for tp_rank_id in range(self.vit_tp):

rpc_model = await start_model_process()
rpc_model, proc = await start_model_process()
self.model_rpcs[dp_rank_id].append(rpc_model)
self.model_procs.append(proc)

init_model_ret = []
for dp_rank_id in range(self.vit_dp): # async init model process
Expand Down Expand Up @@ -130,7 +133,14 @@ async def infer_images(self, dp_index: int, images, events):
await VisualManager.infer_images(self, dp_index=dp_index, images=images, events=events)

def clean_up(self):
return
for proc in getattr(self, "model_procs", []):
try:
if proc.is_alive():
logger.info(f"Killing VIT model process {proc.pid}")
proc.kill()
proc.join(timeout=5)
except (ProcessLookupError, OSError):
pass
Comment on lines +136 to +143
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist bot Apr 3, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

Similar to the VisualManager, the cleanup logic here is sequential. Signaling all processes to terminate before joining them would improve shutdown efficiency, especially when dealing with multiple model processes.

        procs = getattr(self, "model_procs", [])
        for proc in procs:
            try:
                if proc.is_alive():
                    logger.info(f"Killing VIT model process {proc.pid}")
                    proc.kill()
            except (ProcessLookupError, OSError):
                pass
        for proc in procs:
            try:
                proc.join(timeout=5)
            except (ProcessLookupError, OSError):
                pass


def exposed_remote_infer_images(self, images: List[ImageItem], ref_event: threading.Event):
try:
Expand Down
Loading