Skip to content

Allow for pre-bound socket to be passed in server #5301

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

Merged
tdene merged 2 commits into from
Jun 12, 2026
Merged

Allow for pre-bound socket to be passed in server #5301

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d451807
Allow for pre-bound socket to be passed in server
tdene Jun 11, 2026
9c69e51
Address reviewer comments
tdene Jun 11, 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
36 changes: 24 additions & 12 deletions ...n/core/inference/text_generation_server/dynamic_text_gen_server/text_generation_server.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
# Copyright (c) 2025, NVIDIA CORPORATION. All rights reserved.
# Copyright (c) 2026, NVIDIA CORPORATION. All rights reserved.

import asyncio
import logging
Expand Down Expand Up @@ -150,6 +150,7 @@ def start_text_gen_server(
verbose: bool = False,
num_replicas: int = 4,
hostname: Optional[str] = None,
sock: Optional[socket.socket] = None,
):
"""Start the text generation server."""
global _SERVER_PROCESSES
Expand All @@ -159,18 +160,29 @@ def start_text_gen_server(
logger.warning("Text gen server processes are already running.")
return

_SHARED_SOCKET = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
_SHARED_SOCKET.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)

if hasattr(socket, 'SO_REUSEPORT'):
try:
_SHARED_SOCKET.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEPORT, 1)
except OSError:
pass
# The caller may pass in a socket it has already bound ahead of time.
if sock is not None:
bound_port = sock.getsockname()[1]
if bound_port == 0:
raise ValueError(
"socket must be bound to a real port before being passed to start_text_gen_server"
)
_SHARED_SOCKET = sock
server_port = bound_port
_SHARED_SOCKET.setblocking(False)
else:
_SHARED_SOCKET = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
_SHARED_SOCKET.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)

if hasattr(socket, 'SO_REUSEPORT'):
try:
_SHARED_SOCKET.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEPORT, 1)
except OSError:
pass
Comment on lines +177 to +181

@santhnm2 santhnm2 Jun 11, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Can you add a small comment explaining what this is trying to do?

@tdene tdene Jun 11, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

Done!


bind_address = hostname if hostname is not None else "0.0.0.0"
_SHARED_SOCKET.bind((bind_address, server_port))
_SHARED_SOCKET.setblocking(False)
bind_address = hostname if hostname is not None else "0.0.0.0"
_SHARED_SOCKET.bind((bind_address, server_port))
_SHARED_SOCKET.setblocking(False)

_SHARED_SOCKET.set_inheritable(True)
fd = _SHARED_SOCKET.fileno()
Expand Down
Loading