minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[ci] xpu: trigger pr-test-xpu on multimodal_gen changes (#36100)

Browse Source 
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: Ma Mingfei <mingfei.ma@intel.com>
This commit is contained in:
ashwini rathi
2026-08-27 14:50:26 +08:00
committed by GitHub
co-authored by Claude Opus 4.7 Ma Mingfei
parent adcf73d7f7
commit 72bf8c4d53
5 changed files with 363 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/pr-test-xpu.yml
+95 -3
View File
@@ -28,7 +28,9 @@ jobs:
check-changes:
runs-on: ubuntu-latest
outputs:
main_package: ${{ steps.filter.outputs.main_package || steps.run-mode.outputs.run_all_tests }}
changes_exist: ${{ steps.filter.outputs.main_package == 'true' || steps.filter.outputs.multimodal_gen == 'true' || steps.run-mode.outputs.run_all_tests == 'true' }}
main_package: ${{ steps.filter.outputs.main_package == 'true' || steps.run-mode.outputs.run_all_tests == 'true' }}
multimodal_gen: ${{ steps.filter.outputs.multimodal_gen == 'true' || steps.run-mode.outputs.run_all_tests == 'true' }}
steps:
- name: Checkout code
uses: actions/checkout@v4
@@ -60,11 +62,16 @@ jobs:
- "python/sglang/kernels/aot/**/!(*.md|THIRDPARTYNOTICES.txt|LICENSE)"
- ".github/workflows/pr-test-xpu.yml"
- "docker/xpu.Dockerfile"
multimodal_gen:
- "python/sglang/multimodal_gen/**/!(*.md|*.ipynb)"
- "python/pyproject_xpu.toml"
- ".github/workflows/pr-test-xpu.yml"
- "docker/xpu.Dockerfile"
# ==================== PR Gate ==================== #
pr-gate:
needs: check-changes
if: needs.check-changes.outputs.main_package == 'true'
if: needs.check-changes.outputs.changes_exist == 'true'
uses: ./.github/workflows/pr-gate.yml
secrets: inherit
@@ -249,15 +256,96 @@ jobs:
docker rmi -f "${CI_SGLANG_XPU_IMAGE}" || true
fi
# ==================== Multimodal Gen ==================== #
multimodal-gen-test-1-gpu-xpu:
needs: [check-changes, pr-gate]
if: needs.check-changes.outputs.multimodal_gen == 'true'
runs-on: bmg-multigen-models
env:
DOCKERHUB_INTEL_USERNAME: ${{ secrets.DOCKERHUB_INTEL_USERNAME }}
DOCKERHUB_INTEL_TOKEN: ${{ secrets.DOCKERHUB_INTEL_TOKEN }}
steps:
- name: Reset workspace ownership
run: |
docker run --rm -v "${{ github.workspace }}:/w" busybox:latest \
chown -R "$(id -u):$(id -g)" /w || true
- name: Checkout code
uses: actions/checkout@v4
with:
fetch-depth: 0
ref: ${{ inputs.ref || github.ref }}
- name: Start CI container
run: |
export HF_TOKEN="$(cat ~/huggingface_token.txt)"
bash scripts/ci/xpu/xpu_ci_start_container.sh
env:
GITHUB_WORKSPACE: ${{ github.workspace }}
- name: Install Dependency
timeout-minutes: 60
run: |
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip install --upgrade pip
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip install pytest expecttest ray huggingface_hub tabulate "lmcache>=0.3.9"
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip uninstall -y flashinfer-python sgl-kernel sglang
docker exec ci_sglang_xpu cp /sglang-checkout/python/pyproject_xpu.toml /sglang-checkout/python/pyproject.toml
# Fetch tags so setuptools_scm resolves a real version instead of
# falling back to 0.0.0 on a shallow/tag-less checkout.
docker exec -w /sglang-checkout ci_sglang_xpu git fetch origin '+refs/tags/*:refs/tags/*' --force
docker exec -w /sglang-checkout/python ci_sglang_xpu /opt/venv/bin/python3 -m pip install --no-cache-dir . --extra-index-url https://download.pytorch.org/whl/xpu
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip install --no-cache-dir --no-deps xgrammar==0.1.33
docker exec ci_sglang_xpu /bin/bash -c '/opt/venv/bin/hf auth login --token ${HF_TOKEN}'
- name: Run diffusion server tests (1-GPU)
timeout-minutes: 60
run: |
docker exec ci_sglang_xpu bash -c "source /opt/venv/bin/activate && cd /sglang-checkout/python && python3 sglang/multimodal_gen/test/run_suite.py --suite 1-gpu-xpu"
- name: Cleanup container
if: always()
run: |
docker run --rm -v "${{ github.workspace }}:/w" busybox:latest \
chown -R "$(id -u):$(id -g)" /w || true
rm -rf \
python/build \
python/dist \
python/sglang.egg-info \
python/sglang/*.egg-info \
test/result.jsonl \
test/results \
test/.pytest_cache \
.pytest_cache || true
find . -type d -name "__pycache__" -prune -exec rm -rf {} + || true
find . -type f -name "*.pyc" -delete || true
# SIGTERM sglang and drain GPU context before `docker rm -f`;
# SIGKILL leaves the xe/GuC exec queue registered and triggers a
# GT reset (+ devcoredump) on B580.
if docker ps --format '{{.Names}}' | grep -qx ci_sglang_xpu; then
docker exec ci_sglang_xpu bash -c '
pkill -TERM -f "sglang|run_suite|python3.*test_" 2>/dev/null || true
for _ in $(seq 1 30); do
pgrep -f "sglang::|sglang.launch_server" >/dev/null || break
sleep 1
done
pkill -KILL -f "sglang|run_suite" 2>/dev/null || true
' || true
fi
docker rm -f ci_sglang_xpu || true
if [[ -n "${CI_SGLANG_XPU_IMAGE:-}" ]]; then
docker rmi -f "${CI_SGLANG_XPU_IMAGE}" || true
fi
finish:
if: always()
needs: [stage-a-test-1-gpu-xpu, stage-b-test-1-gpu-xpu, pr-gate]
needs: [stage-a-test-1-gpu-xpu, stage-b-test-1-gpu-xpu, multimodal-gen-test-1-gpu-xpu, pr-gate]
runs-on: ubuntu-latest
steps:
- name: Check job status
run: |
stage_a="${{ needs.stage-a-test-1-gpu-xpu.result }}"
stage_b="${{ needs.stage-b-test-1-gpu-xpu.result }}"
multimodal_gen="${{ needs.multimodal-gen-test-1-gpu-xpu.result }}"
if [ "$stage_a" != "success" ] && [ "$stage_a" != "skipped" ]; then
echo "stage-a failed with result: $stage_a"
exit 1
@@ -266,5 +354,9 @@ jobs:
echo "stage-b failed with result: $stage_b"
exit 1
fi
if [ "$multimodal_gen" != "success" ] && [ "$multimodal_gen" != "skipped" ]; then
echo "multimodal-gen failed with result: $multimodal_gen"
exit 1
fi
echo "All jobs completed successfully"
exit 0