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

[XPU][CI] Enable nightly-xpu-8-gpu suite: declare + wire runner job (#37230)

Browse Source 
Signed-off-by: P V R K Jyothendra Varma <polisetty.v.r.k.jyothendra.varma@intel.com>
Co-authored-by: P V R K Jyothendra Varma <polisetty.v.r.k.jyothendra.varma@intel.com>
This commit is contained in:
ashwini rathi
2026-09-02 10:14:40 +08:00
committed by GitHub
co-authored by P V R K Jyothendra Varma
parent b83bf7a65f
commit e874ae64cd
2 changed files with 76 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/nightly-test-intel.yml
+75
View File
@@ -297,12 +297,87 @@ jobs:
docker rmi -f "${CI_SGLANG_XPU_IMAGE}" || true
fi
nightly-xpu-8-gpu:
if: github.repository == 'sgl-project/sglang'
runs-on: intel-bmg-nightly
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: ${{ inputs.ref || github.sha }}
- name: Start CI container (pull intel/sglang-dev:latest)
run: |
export HF_TOKEN="$(cat ~/huggingface_token.txt)"
bash scripts/ci/xpu/xpu_ci_start_container.sh
env:
GITHUB_WORKSPACE: ${{ github.workspace }}
- name: HF login + install run_suite extras
timeout-minutes: 10
run: |
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip install --no-cache-dir tabulate
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip install --no-cache-dir --no-deps xgrammar==0.1.33
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip install --no-cache-dir "lm-eval==0.4.9"
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip install --no-cache-dir pytest
# Diffusion extras
# Pins mirror python/pyproject_xpu.toml [project.optional-dependencies.diffusion].
docker exec ci_sglang_xpu /opt/venv/bin/python3 -m pip install --no-cache-dir "sglang[diffusion]"
# HF /whoami-v2 is aggressively rate-limited and the three matrix
# jobs (1/2/4/8-GPU) all hit it at the same second under a scheduled
# run. Retry with linear backoff + jitter to desynchronize them.
docker exec ci_sglang_xpu /bin/bash -c '
for i in 1 2 3 4 5; do
/opt/venv/bin/hf auth login --token "${HF_TOKEN}" && exit 0
sleep_secs=$((i * 15 + RANDOM % 15))
echo "hf auth login failed (attempt $i), retrying in ${sleep_secs}s..."
sleep "${sleep_secs}"
done
exit 1
'
- name: Nightly Test (8-GPU XPU)
timeout-minutes: 480
run: |
touch github_summary.md
docker exec ci_sglang_xpu bash -c "
source /opt/venv/bin/activate &&
cd /sglang-checkout/test &&
GITHUB_STEP_SUMMARY=/sglang-checkout/github_summary.md \
python3 run_suite.py --hw xpu --suite nightly-xpu-8-gpu --nightly --timeout-per-file 7200 ${{ (github.event_name == 'schedule' || inputs.continue_on_error) && '--continue-on-error' || '' }}
" || TEST_EXIT_CODE=$?
echo "$(<github_summary.md)" >> $GITHUB_STEP_SUMMARY || true
exit ${TEST_EXIT_CODE:-0}
- name: Cleanup container
if: always()
run: |
docker run --rm -v "${{ github.workspace }}:/w" busybox:latest \
chown -R "$(id -u):$(id -g)" /w || true
rm -rf 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
docker rm -f ci_sglang_xpu || true
if [[ -n "${CI_SGLANG_XPU_IMAGE:-}" ]]; then
docker rmi -f "${CI_SGLANG_XPU_IMAGE}" || true
fi
check-all-jobs:
if: always() && (github.repository == 'sgl-project/sglang' || github.event_name == 'workflow_dispatch')
needs:
- nightly-xpu-1-gpu
- nightly-xpu-2-gpu
- nightly-xpu-4-gpu
- nightly-xpu-8-gpu
runs-on: ubuntu-latest
steps:
- name: Check if any job failed
test/run_suite.py
+1
View File
@@ -181,6 +181,7 @@ NIGHTLY_SUITES = {
"nightly-xpu-1-gpu",
"nightly-xpu-2-gpu",
"nightly-xpu-4-gpu",
"nightly-xpu-8-gpu",
],
}