Fix dual-stack socket handling: IPV6_V6ONLY, IPv4-first, is_port_available all-family check

[hnyls2002]

Changes

• Set IPV6_V6ONLY=1 on IPv6 sockets in try_bind_socket — isolate IPv4/IPv6 to eliminate platform-dependent dual-stack behavior
• Fix is_port_available to check all address families with SO_REUSEADDR — previously only checked first-success family, missing conflicts on the other family under IPV6_V6ONLY
• Sort _get_addrinfos_for_bind IPv4-first for consistent default across platforms
• Refactor get_open_port to use is_port_available (all-family) for SGLANG_PORT path; use default reuse_addr=True for ephemeral path to match server bind behavior

Nits

• Rewrite _get_addrinfos_for_bind dedup from set-comprehension side-effect hack to explicit loop
• Move OverflowError catch from is_port_available down to try_bind_socket
• Add tests for reserve_port, release_port, normalize_base_url, port-occupied detection

Test plan

• Unit tests pass on both macOS and Linux
• /rerun-stage stage-a-cpu-only

[gemini-code-assist]

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request significantly enhances the network handling capabilities, particularly for IPv6, by refining socket binding logic and ensuring more reliable port allocation. The changes aim to improve the stability and predictability of network operations within the system, complemented by new tests to validate the behavior of key utility functions.

Highlights

• Improved IPv6 Handling: Configured IPv6 sockets to explicitly use IPV6_V6ONLY to prevent dual-stack behavior, ensuring more predictable IPv6 network interactions.
• Enhanced Port Binding Logic: Refactored the address information deduplication in _get_addrinfos_for_bind for clarity and robustness, and expanded exception handling in try_bind_socket to include OverflowError.
• Reliable Open Port Retrieval: Modified get_open_port to explicitly disable SO_REUSEADDR when searching for an available port, ensuring that the returned port is truly unique and not in use by a lingering socket.
• New Utility Function Tests: Added comprehensive unit tests for the reserve_port and normalize_base_url utility functions, covering various scenarios including IPv4, IPv6, hostnames, and error conditions.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Changelog

• python/sglang/srt/utils/common.py
  • Refactored address information deduplication in _get_addrinfos_for_bind for improved clarity and robustness.
  • Added IPV6_V6ONLY socket option for IPv6 sockets to prevent dual-stack behavior.
  • Expanded exception handling in try_bind_socket to catch OverflowError during socket binding.
  • Adjusted exception handling in is_port_available to specifically catch OSError.
  • Ensured get_open_port binds to a unique port by setting reuse_addr=False in try_bind_socket.
• test/registered/utils/test_socket_utils.py
  • Imported normalize_base_url, release_port, and reserve_port for testing.
  • Added new test cases for TestReservePort covering functionality, custom ranges, and error handling.
  • Included new test cases for TestNormalizeBaseUrl to validate IPv4, IPv6, hostname, and deprecated scheme handling.

Activity

• No human activity has been recorded on this pull request yet.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request introduces several improvements to IPv6 network handling and related socket utilities. The changes include refactoring address info deduplication for better readability, correctly setting the IPV6_V6ONLY socket option for IPv6 sockets, refining exception handling during port binding, and making get_open_port more reliable by disabling address reuse. Additionally, new tests are added for port reservation and URL normalization, which enhance test coverage. The changes are logical and well-implemented. I've found a minor issue in one of the new tests regarding an incorrect mock patch target and have provided a suggestion to fix it.

[hnyls2002]

/tag-and-rerun-ci

[hnyls2002]

/rerun-ut test_socket_utils.py

[github-actions]

❌ No register_cuda_ci() found in test/registered/utils/test_socket_utils.py.

This file may not be a registered CUDA CI test.

[hnyls2002]

/rerun-ut test/registered/disaggregation/test_disaggregation_basic.py

[github-actions]

❌ Ambiguous filename test_disaggregation_basic.py — matched 2 files:

• test/registered/amd/disaggregation/test_disaggregation_basic.py
• test/registered/disaggregation/test_disaggregation_basic.py

Please provide the full path, e.g. /rerun-ut test/registered/amd/disaggregation/test_disaggregation_basic.py

[github-actions]

✅ Triggered /rerun-ut on 2-gpu-runner runner:

cd test/ && python3 registered/disaggregation/test_disaggregation_basic.py

[github-actions]

🔗 View workflow run

[hnyls2002]

/rerun-ut test_disaggregation_dp_attention.py

[github-actions]

✅ Triggered /rerun-ut on 8-gpu-h20 runner:

cd test/ && python3 registered/distributed/test_disaggregation_dp_attention.py

[github-actions]

🔗 View workflow run