[RL] Add --nccl-port to prevent port conflict

[zhuzilin]

Motivation

In RL training, we need to start multiple sglang server on one machine, and by default, sglang will allocate an random port to init nccl, which may result in address conflict. Therefore we propose to allocate all ports in advance by adding --nccl-port arguments.

Thank you for your time on reviewing this PR :)

Modifications

Checklist

• Format your code according to the Code Formatting with Pre-Commit.
• Add unit tests as outlined in the Running Unit Tests.
• Update documentation / docstrings / example tutorials as needed, according to Writing Documentation.
• Provide throughput / latency benchmark results and accuracy evaluation results as needed, according to Benchmark and Profiling and Accuracy Results.
• For reviewers: If you haven't made any contributions to this PR and are only assisting with merging the main branch, please remove yourself as a co-author when merging the PR.
• Please feel free to join our Slack channel at https://slack.sglang.ai to discuss your PR.

[gemini-code-assist]

Summary of Changes

Hello @zhuzilin, 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 addresses a common challenge in distributed RL training environments where running multiple sglang servers on a single machine can lead to port conflicts. It introduces new configuration options that allow users to pre-allocate and specify exact ports for NCCL communication and other critical internal services, thereby preventing conflicts and ensuring smoother operation in multi-server setups, especially for data parallel attention.

Highlights

• New CLI Arguments for Port Control: Introduced --nccl-port and --other-ports command-line arguments, allowing users to explicitly specify ports for NCCL communication and other internal services (tokenizer, detokenizer, RPC, scheduler input).
• ServerArgs Configuration Update: Extended the ServerArgs class to include nccl_port (for NCCL communication) and other_ports (a list for various internal IPC ports) as optional integer fields, enabling programmatic configuration of these ports.
• Dynamic Port Allocation Logic Enhancement: Updated the PortArgs.init_new method to prioritize user-defined nccl_port and other_ports for various internal communication channels. If nccl_port is not provided, it falls back to a more robust dynamic allocation with an is_port_available check loop. When other_ports are provided, specific indices are used for tokenizer, detokenizer, RPC, and scheduler input ports.

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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

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 issue 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 is currently in preview and 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 to provide feedback.

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 new command-line arguments --nccl-port and --other-ports to allow users to pre-allocate ports for the sglang server, preventing address conflicts. The changes include modifications to the ServerArgs dataclass, the argument parsing logic, and the port initialization logic. I've provided feedback to improve error handling and robustness.