Update B300 FP4 SGLang (non-MTP) image to latest nightly#1913
Update B300 FP4 SGLang (non-MTP) image to latest nightly#1913hshrivastava-droid wants to merge 2 commits into
Conversation
Bumps dsv4-fp4-b300-sglang image from lmsysorg/sglang:nightly-dev-cu13-20260529-a8cfae0b to lmsysorg/sglang:nightly-dev-cu13-20260624-b2c8f7a2.
hshrivastava-droid
requested review from
Ankur-singh,
jgangani and
kedarpotdar-nv
as code owners
|
Thanks for the contribution! For vLLM & SGLang, please ensure that your recipes is similar to the official vLLM recipes and/or the SGLang cookbook If it is not, please create a PR first before we can merge your single node PR into the master branch. Let's ensure that the documentation is first class such that the entire ML community can benefit from your hard work! Thank you PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. If re-running failed jobs is attempted, PR authors are responsible for ensuring it passes. See GitHub's docs on re-running failed jobs: https://docs.github.com/en/actions/how-tos/manage-workflow-runs/re-run-workflows-and-jobs#re-running-failed-jobs-in-a-workflow As a rule of thumb, generally, PR authors should request a review & get a PR approval from the respective companies' CODEOWNERS before requesting a review from core maintainers. If additional help is needed, PR authors can reach out to core maintainers over Slack. 感谢你的贡献！对于 vLLM 与 SGLang，请确保你的 recipe 与官方 vLLM recipes 和/或 SGLang cookbook 保持一致 如果不一致，请先创建一个 PR，之后我们才能将你的单节点 PR 合并到 master 分支。让我们确保文档保持一流水准，使整个 ML 社区都能从你的辛勤工作中受益！谢谢 PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动（flake），重新运行失败的任务即可解决。如果选择重新运行失败的任务，PR 作者有责任确保其最终通过。参见 GitHub 关于重新运行失败任务的文档：https://docs.github.com/en/actions/how-tos/manage-workflow-runs/re-run-workflows-and-jobs#re-running-failed-jobs-in-a-workflow 一般而言，PR 作者应先向相应公司的 CODEOWNERS 请求审阅并获得 PR 批准，然后再请求核心维护者审阅。 如需更多帮助，PR 作者可通过 Slack 联系核心维护者。 |
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
| - dsv4-fp4-b300-sglang | ||
| description: | ||
| - "Update B300 FP4 SGLang (non-MTP) image to latest nightly: lmsysorg/sglang:nightly-dev-cu13-20260624-b2c8f7a2 (was nightly-dev-cu13-20260529-a8cfae0b)." | ||
| pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXX |
There was a problem hiding this comment.
🔴 The new perf-changelog entry sets pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXX — a placeholder that was never replaced with the actual PR number. Since this is PR #1913, it should read /pull/1913. As-is, the merged changelog will permanently link to a non-existent PR, breaking traceability from this entry back to the originating PR and breaking any tooling that parses pr-link.
Extended reasoning...
What the bug is
The diff adds a new entry to perf-changelog.yaml for the dsv4-fp4-b300-sglang image bump, but its pr-link field still contains the literal placeholder XXX:
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXXThis was clearly intended to be replaced with the real PR number before the PR was opened/merged.
Why this matters
Every other entry in perf-changelog.yaml uses a real PR number (e.g. #1891, #1894, #1792, #1676), and the convention is clearly to point pr-link at the PR that introduced the change. The placeholder breaks that contract in two concrete ways:
- The rendered URL
https://github.com/SemiAnalysisAI/InferenceX/pull/XXXis a permanent dead link — GitHub will 404 on it. - Any tooling that parses
pr-link(changelog generators, sweep dashboards, traceability scripts) will either crash on the non-numeric segment or silently fail to look up the originating PR.
Because the changelog is append-only and the placeholder will be carried in git history, fixing it post-merge requires a follow-up commit; catching it pre-merge is essentially free.
Step-by-step proof
- Open
perf-changelog.yamland look at line 4161 in the diff — the new block ends withpr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXX. - The PR metadata at the top of this review shows this is PR Update B300 FP4 SGLang (non-MTP) image to latest nightly #1913 (
<pr number="1913">). - Scroll up in
perf-changelog.yaml: the immediately preceding entry useshttps://github.com/SemiAnalysisAI/InferenceX/pull/1891, matching its own PR number. Other recent entries follow the same pattern (1894, 1792, 1676). - Therefore the new entry should use
/pull/1913, not/pull/XXX.
Fix
Replace XXX with 1913:
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1913
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=28078823761 |
2 similar comments
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=28078823761 |
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=28078823761 |
1 participant
Summary
dsv4-fp4-b300-sglangimage:lmsysorg/sglang:nightly-dev-cu13-20260529-a8cfae0b→lmsysorg/sglang:nightly-dev-cu13-20260624-b2c8f7a2(latestnightly-dev-cu13-*on Docker Hub).perf-changelog.yamlentry to trigger a sweep.