Skip to content

[CI] Support branch-based nightly testing with dynamic vLLM version#7627

Open
zhangxinyuehfad wants to merge 1 commit intovllm-project:mainfrom
zhangxinyuehfad:zxy_branch_nightly
Open

[CI] Support branch-based nightly testing with dynamic vLLM version#7627
zhangxinyuehfad wants to merge 1 commit intovllm-project:mainfrom
zhangxinyuehfad:zxy_branch_nightly

Conversation

@zhangxinyuehfad
Copy link
Collaborator

@zhangxinyuehfad zhangxinyuehfad commented Mar 25, 2026

What this PR does / why we need it?

Nightly CI workflows previously hardcoded the vLLM version (v0.18.0) and image tag (nightly-a2/nightly-a3) in multiple places. This PR introduces a unified mechanism to derive the version from the branch name, enabling proper multi-branch nightly testing.
Key changes:

  • _parse_trigger.yaml: Add branch input; auto-derive vllm_version from branch name (releases-v0.13.0v0.13.0); detect PR target branch to select the matching nightly image; getRef() prefers inputs.branch for dispatch triggers
  • _nightly_image_build.yaml: Image tag now includes version (e.g., nightly-v0.18.0-a2)
  • _e2e_nightly_single_node.yaml: Add vllm_ascend_ref input; on dispatch with a ref, checkout and reinstall vllm-ascend from the branch; consolidate PR setup steps (7 → 4 steps); rename is_runshould_run
  • schedule_nightly_test_a2.yaml / schedule_nightly_test_a3.yaml: Pass branch input; derive vllm_version dynamically via parse-trigger; propagate vllm_ascend_ref to single-node tests
  • nightly_image_build.yaml: Add vllm_version input to fix missing required parameter
  • lws*.yaml.jinja2: Image default constructed from vllm_version variable
  • multi_node_test.md: Use |ci_vllm_version| substitution instead of hardcoded version
    这个翻译

Does this PR introduce any user-facing change?

How was this patch tested?

Signed-off-by: hfadzxy <starmoon_zhang@163.com>
@github-actions github-actions bot added documentation Improvements or additions to documentation ci/build module:tests labels Mar 25, 2026
@github-actions
Copy link
Contributor

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

  • A PR should do only one thing, smaller PRs enable faster reviews.
  • Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
  • Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

@gemini-code-assist
Copy link
Contributor

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 refactors the Continuous Integration (CI) workflows to support branch-based nightly testing with dynamic vLLM versions. Previously, CI workflows hardcoded vLLM versions and image tags, limiting flexibility. The changes introduce a unified approach to derive versions from branch names, ensuring that nightly tests and image builds correctly correspond to the specific vLLM version being tested across different branches. This significantly improves the maintainability and accuracy of the nightly CI system.

Highlights

  • Dynamic Version Derivation: Introduced a mechanism in _parse_trigger.yaml to automatically derive the vLLM version from the branch name (e.g., releases-v0.13.0 -> v0.13.0), supporting proper multi-branch nightly testing.
  • Unified Image Tagging: Updated _nightly_image_build.yaml to include the vLLM version in the image tag (e.g., nightly-v0.18.0-a2), ensuring version-specific image builds.
  • Enhanced Single-Node Testing: Modified _e2e_nightly_single_node.yaml to add vllm_ascend_ref input, allowing vllm-ascend to be checked out and reinstalled from a specific branch on dispatch, and consolidated PR setup steps.
  • Scheduled Test Updates: Updated schedule_nightly_test_a2.yaml and schedule_nightly_test_a3.yaml to pass the branch input, dynamically derive vllm_version via parse-trigger, and propagate vllm_ascend_ref to single-node tests.
  • Configuration Template Updates: Adjusted lws*.yaml.jinja2 templates to construct image defaults using the dynamic vllm_version variable, and updated multi_node_test.md to use |ci_vllm_version| substitution.

🧠 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.

Ignored Files
  • Ignored by pattern: .github/workflows/** (6)
    • .github/workflows/_e2e_nightly_single_node.yaml
    • .github/workflows/_nightly_image_build.yaml
    • .github/workflows/_parse_trigger.yaml
    • .github/workflows/nightly_image_build.yaml
    • .github/workflows/schedule_nightly_test_a2.yaml
    • .github/workflows/schedule_nightly_test_a3.yaml
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.

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.

Copy link
Contributor

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request updates VLLM Docker image references in multi-node test documentation and Kubernetes deployment templates. It replaces hardcoded 'nightly-a2' or 'nightly-a3' image tags with dynamic tags that incorporate a 'vllm_version' variable. This change enables more flexible and configurable image versioning for CI/CD pipelines and local testing environments. No user-facing changes are introduced, and the modifications are implicitly tested by existing CI/CD pipelines. There is no feedback to provide as no review comments were submitted.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ci/build documentation Improvements or additions to documentation module:tests

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant