Skip to content
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

[V1][Core] Remove should_shutdown to simplify core process termination #11113

Merged
merged 3 commits into from
Dec 11, 2024
Merged

[V1][Core] Remove should_shutdown to simplify core process termination #11113

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9c713b4
[V1][Core] Remove should_shutdown to simplify core process termination
tlrmchlsmth Dec 11, 2024
1bc2fc3
fixup
tlrmchlsmth Dec 11, 2024
5758a20
imports
tlrmchlsmth Dec 11, 2024
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
13 changes: 2 additions & 11 deletions vllm/v1/engine/core.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,6 @@
import threading
import time
from multiprocessing.process import BaseProcess
from multiprocessing.sharedctypes import Synchronized
from typing import List, Tuple, Type, Union

import zmq
Expand Down Expand Up @@ -133,13 +132,9 @@ def __init__(
input_path: str,
output_path: str,
ready_path: str,
should_shutdown: Synchronized,
):
super().__init__(vllm_config, executor_class, usage_context)

# Signal from main process to shutdown (multiprocessing.Value).
self.should_shutdown = should_shutdown

# Background Threads and Queues for IO. These enable us to
# overlap ZMQ socket IO with GPU since they release the GIL,
# and to overlap some serialization/deserialization with the
Expand Down Expand Up @@ -195,7 +190,6 @@ def make_engine_core_process(
input_path: str,
output_path: str,
ready_path: str,
should_shutdown: Synchronized,
) -> BaseProcess:
# The current process might have CUDA context,
# so we need to spawn a new process.
Expand All @@ -210,7 +204,6 @@ def make_engine_core_process(
"vllm_config": vllm_config,
"executor_class": executor_class,
"usage_context": usage_context,
"should_shutdown": should_shutdown
}
# Run EngineCore busy loop in background process.
proc = context.Process(target=EngineCoreProc.run_engine_core,
Expand Down Expand Up @@ -260,8 +253,8 @@ def signal_handler(signum, frame):
def run_busy_loop(self):
"""Core busy loop of the EngineCore."""

# Loop until we get a shutdown signal.
while not self.should_shutdown:
# Loop until process is sent a SIGINT or SIGTERM
while True:
# 1) Poll the input queue until there is work to do.
if not self.scheduler.has_unfinished_requests():
while True:
Expand All @@ -272,8 +265,6 @@ def run_busy_loop(self):
except queue.Empty:
self._log_stats()
logger.debug("EngineCore busy loop waiting.")
if self.should_shutdown:
return
except BaseException:
raise

Expand Down
6 changes: 0 additions & 6 deletions vllm/v1/engine/core_client.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,4 @@
import atexit
import multiprocessing
from typing import List, Union

import msgspec
Expand Down Expand Up @@ -149,21 +148,16 @@ def __init__(
self.input_socket.bind(input_path)

# Start EngineCore in background process.
self.should_shutdown = multiprocessing.Value('b', False, lock=False)
self.proc = EngineCoreProc.make_engine_core_process(
*args,
input_path=input_path,
output_path=output_path,
ready_path=ready_path,
should_shutdown=self.should_shutdown,
**kwargs,
)
atexit.register(self.shutdown)

def shutdown(self):
# Send shutdown signal to background process.
self.should_shutdown = True

# Shut down the zmq context.
self.ctx.destroy(linger=0)

Expand Down
Loading