Enhance prefill PP node robustness

[qhsc]

Motivation

When deploy PP2 + PD for online serving, we found that the assertion assert poll == KVPoll.WaitingForInput or poll == KVPoll.Failed of prefill node would fail in some situation. The root cause of this problem is PP1 and PP0 have different bootstrapped requests due to unstable network latency. We add a small change to enable the higher PP ranks to be bootstrapped earlier to reduce the influence of network latency and make system more reliable.

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.
• Work with maintainers to merge your PR. See the PR Merge Process

[gemini-code-assist]

Summary of Changes

Hello @qhsc, 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 enhances the robustness of the prefill node in a disaggregated serving setup (PP + PD). It specifically targets an assertion failure that could occur during the bootstrapping process. The core modification involves reordering how pipeline parallelism ranks are initialized, ensuring that higher ranks are processed earlier to establish more stable connections and prevent runtime issues.

Highlights

• Improved Bootstrapping Order: The order of bootstrapping for pipeline parallelism (PP) ranks has been reversed, now processing higher ranks first.
• Enhanced Robustness: This change aims to make the bootstrapping of PP PD requests more robust, addressing an assertion failure observed during online serving.

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 enhances the robustness of the prefill pipeline parallelism node by changing the bootstrapping order of pipeline ranks. The change iterates over pipeline ranks in descending order, which, as the comment suggests, helps prevent an assertion failure. The logic seems sound.

I have one minor suggestion to improve the implementation slightly.

Also, there are a few typos in the pull request title and description:

• Title: "Enanche" -> "Enhance"
• Description: "onling" -> "online", "foud" -> "found"

[ShangmingCai]

LGTM. A very clean and smart fix.

[ShangmingCai]

/tag-and-rerun-ci