Files
xc-llm-ascend/.github/workflows/schedule_update_estimated_time.yaml
Li Wang d431d7d526 [CI] Enable auto upgrade e2e estimated time for auto-partition suites (#6840)
### What this PR does / why we need it?
This patch add a schedule triggered workflow for auto upgrade e2e
estimated-time for batter load balance
1. The workflow will run the full e2e test to get the duration of each
test.
2. The script `update_estimated_time.py` will upgrade the
[config.json](https://github.com/vllm-project/vllm-ascend/blob/main/.github/workflows/scripts/config.yaml)
according to the latest time
3. The workflow will submit a pull request that includes changes to
`config.json` automatically
<img width="2484" height="764" alt="image"
src="https://github.com/user-attachments/assets/02f3459c-bb3b-4f8e-9966-8bb2e5c1bbea"
/>


### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

- vLLM version: v0.15.0
- vLLM main:
83b47f67b1
- 
### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

- vLLM version: v0.15.0
- vLLM main:
83b47f67b1

---------

Signed-off-by: wangli <wangli858794774@gmail.com>
2026-03-04 10:38:34 +08:00

112 lines
3.8 KiB
YAML

name: Update estimated test times
on:
schedule:
- cron: '0 2 * * 1' # Every Monday at 02:00 UTC
workflow_dispatch:
pull_request:
branches:
- 'main'
paths:
- '.github/workflows/schedule_update_estimated_time.yaml'
permissions:
contents: write
pull-requests: write
concurrency:
group: update-estimated-times-${{ github.ref }}
cancel-in-progress: true
jobs:
e2e-test:
name: e2e-test
strategy:
matrix:
vllm_version: [15d76f74e2fdb12a95ea00f0ca283acf6219a2b7]
type: [full, light]
uses: ./.github/workflows/_e2e_test.yaml
with:
vllm: ${{ matrix.vllm_version }}
image: swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:main
contains_310: false
type: ${{ matrix.type }}
continue_on_error: true # Continue even if some tests fail, we want to collect as much timing data as possible
update-estimated-times:
name: Update estimated_time in config.yaml
needs: [e2e-test]
runs-on: ubuntu-latest
steps:
- name: Checkout repo
uses: actions/checkout@v4
- name: Download all timing artifacts
uses: actions/download-artifact@v4
with:
pattern: timing-data-*
path: timing-artifacts/
merge-multiple: false
- name: Set up Python
uses: actions/setup-python@v5
with:
python-version: '3.11'
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install pyyaml
- name: Update config.yaml from timing data
run: |
python3 .github/workflows/scripts/update_estimated_time.py \
--timing-dir timing-artifacts/ \
--config .github/workflows/scripts/config.yaml
- name: Check for changes
id: check_changes
run: |
if git diff --quiet .github/workflows/scripts/config.yaml; then
echo "changed=false" >> "$GITHUB_OUTPUT"
echo "No changes to config.yaml."
else
echo "changed=true" >> "$GITHUB_OUTPUT"
echo "config.yaml has been updated:"
git diff .github/workflows/scripts/config.yaml
fi
- name: Create pull request
if: steps.check_changes.outputs.changed == 'true' && github.event_name != 'pull_request'
env:
GH_TOKEN: ${{ github.token }}
run: |
BRANCH="auto/update-estimated-times-${{ github.run_id }}"
git config user.name "github-actions[bot]"
git config user.email "github-actions[bot]@users.noreply.github.com"
git checkout -b "$BRANCH"
git add .github/workflows/scripts/config.yaml
git commit -m "[CI] Auto-update estimated test times in config.yaml
Computed from timing-data artifacts of workflow run ${{ github.run_id }}.
Buffer ratio: 1.1x median, rounded to the nearest 10 s."
git push origin "$BRANCH"
gh pr create \
--repo "${{ github.repository }}" \
--base main \
--head "$BRANCH" \
--title "chore: Auto-update estimated test times in config.yaml" \
--body "## Summary
This PR was auto-generated by the **Update estimated test times** workflow.
It updates the \`estimated_time\` values in \`.github/workflows/scripts/config.yaml\`
based on actual elapsed times collected from workflow run \`${{ github.run_id }}\`.
### Methodology
- Timing data is uploaded as \`timing-data-*\` artifacts by each e2e test job.
- For each test file, the **median** of all collected elapsed times is taken.
- A **10 % safety buffer** is applied and the result is rounded to the nearest 10 s.
Please review the diff and merge if the new values look reasonable."