Add bisect ci claude code skill#19649
Conversation
Adds a reusable skill that systematically investigates failing CI tests through temporal bisection, runner/hardware analysis, and optional remote reproduction on GPU servers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Document the pr-test.yml scheduled runs on main as the primary data source for regression bisection, with dashboard link and --repo flags. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
github-actions
bot
added
the
documentation
Summary of ChangesHello, 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 introduces a new CI skill aimed at streamlining the debugging process for persistent CI failures. It provides a structured approach to identify whether a regression is due to code changes, hardware specifics, or environmental factors, and offers tools for remote reproduction to facilitate quicker resolution of issues. Highlights
🧠 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
Activity
Using Gemini Code AssistThe 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
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 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
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces a new skill documentation file for bisecting CI regressions. The documentation is well-structured and provides a detailed workflow. My review focuses on improving the correctness and clarity of the shell commands within the guide. I've suggested using grep -E for better portability and have proposed changes to the installation instructions to resolve ambiguity between different update methods (git vs. tarball).
| gh run view {RUN_ID} --repo sgl-project/sglang --json jobs --jq '.jobs[] | select(.conclusion == "failure") | {name, conclusion, databaseId}' | ||
|
|
||
| # Get the failure details | ||
| gh run view {RUN_ID} --repo sgl-project/sglang --job {JOB_ID} --log 2>&1 | grep -B 5 -A 30 "AssertionError\|FAIL\|Error\|{TEST_NAME}" |
There was a problem hiding this comment.
The grep command here uses \| for alternation, which is part of basic regular expressions. For better portability and consistency with other grep commands in this document that use extended regular expressions, it's recommended to use the -E flag and | for alternation.
| gh run view {RUN_ID} --repo sgl-project/sglang --job {JOB_ID} --log 2>&1 | grep -B 5 -A 30 "AssertionError\|FAIL\|Error\|{TEST_NAME}" | |
| gh run view {RUN_ID} --repo sgl-project/sglang --job {JOB_ID} --log 2>&1 | grep -E -B 5 -A 30 "AssertionError|FAIL|Error|{TEST_NAME}" |
| gh run view {RUN_ID} --repo sgl-project/sglang --job {JOB_ID} --log 2>&1 | grep -E "Runner name|Machine name" | head -5 | ||
|
|
||
| # Get GPU/driver info | ||
| gh run view {RUN_ID} --repo sgl-project/sglang --job {JOB_ID} --log 2>&1 | grep -i "NVIDIA-SMI\|Driver Version\|CUDA Version" | head -5 |
There was a problem hiding this comment.
The grep command here uses \| for alternation without the -E flag. This might not be portable across all systems. Using grep -i -E with | is more standard and consistent with other commands in this file.
| gh run view {RUN_ID} --repo sgl-project/sglang --job {JOB_ID} --log 2>&1 | grep -i "NVIDIA-SMI\|Driver Version\|CUDA Version" | head -5 | |
| gh run view {RUN_ID} --repo sgl-project/sglang --job {JOB_ID} --log 2>&1 | grep -i -E "NVIDIA-SMI|Driver Version|CUDA Version" | head -5 |
| # Or download tarball if git auth fails | ||
| ssh {SSH_TARGET} "docker exec {CONTAINER} bash -c 'curl -L https://github.com/sgl-project/sglang/archive/refs/heads/main.tar.gz -o /tmp/sglang-main.tar.gz && cd /tmp && tar xzf sglang-main.tar.gz'" | ||
| # Reinstall | ||
| ssh {SSH_TARGET} "docker exec {CONTAINER} bash -c 'cd /path/to/sglang && pip install -e \"python[all]\"'" |
There was a problem hiding this comment.
The instructions for reinstalling the code are ambiguous. The Reinstall step on line 142 uses /path/to/sglang, which is correct for the git fetch method. However, if the tarball method on line 140 is used, the code is extracted to /tmp/sglang-main, making the reinstall command on line 142 incorrect for that case.
To avoid this ambiguity, I suggest making the tarball installation self-contained and clarifying that the separate reinstall step is for the git method. This makes the instructions clearer and less error-prone.
| # Or download tarball if git auth fails | |
| ssh {SSH_TARGET} "docker exec {CONTAINER} bash -c 'curl -L https://github.com/sgl-project/sglang/archive/refs/heads/main.tar.gz -o /tmp/sglang-main.tar.gz && cd /tmp && tar xzf sglang-main.tar.gz'" | |
| # Reinstall | |
| ssh {SSH_TARGET} "docker exec {CONTAINER} bash -c 'cd /path/to/sglang && pip install -e \"python[all]\"'" | |
| # Or download and install from tarball if git auth fails | |
| ssh {SSH_TARGET} "docker exec {CONTAINER} bash -c 'cd /tmp && curl -L https://github.com/sgl-project/sglang/archive/refs/heads/main.tar.gz | tar xz && cd sglang-main && pip install -e \"python[all]\"'" | |
| # Reinstall (after git fetch) | |
| ssh {SSH_TARGET} "docker exec {CONTAINER} bash -c 'cd /path/to/sglang && pip install -e \"python[all]\"'" |
- Use `grep -E` with `|` instead of `\|` for alternation (3 places) - Clarify tarball vs git fetch reinstall instructions to avoid ambiguity Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
1 participant
Motivation
Modifications
Accuracy Tests
Benchmarking and Profiling
Checklist
Review Process
/tag-run-ci-label,/rerun-failed-ci,/tag-and-rerun-ci