[CI] slash handler: lookup runs_on from runner_configs.yml (#25394)

This commit is contained in:
Liangsheng Yin
2026-05-15 13:59:00 -07:00
committed by GitHub
parent 54221dd998
commit f9caf43095
33 changed files with 261 additions and 309 deletions
+1 -1
View File
@@ -63,7 +63,7 @@ jobs:
const missingCIText = ':x: **Missing `run-ci` label** — add it to run CI tests.';
const peBlockedByCIText = ':x: **Blocked** — `run-ci` is required first.';
const notExtraEnabledText = ':warning: **Not enabled** — add `run-ci-extra` label to opt in.';
const stalePushText = ':warning: **Not run on latest push** — push again or use `/rerun-failed-ci` to dispatch.';
const stalePushText = ':warning: **Not run on latest push** — push again to dispatch.';
const ptText = !hasCI
? missingCIText
: (isReal(ptRun) ? `[Run #${ptRun.id}](${ptRun.html_url})` : '_Not run yet_');
+110 -62
View File
@@ -4,46 +4,43 @@ run-name: ${{ inputs.pr_head_sha && format('[rerun-test] {0} {1}', inputs.test_c
on:
workflow_dispatch:
inputs:
mode:
description: "Dispatch mode: cuda | multimodal_gen | cpu"
required: true
type: choice
options:
- cuda
- multimodal_gen
- cpu
test_command:
description: "Test command(s) to run, one per line (e.g. 'registered/core/test_srt_endpoint.py TestSRTEndpoint.test_simple_decode')"
required: true
type: string
runner_label:
description: "Runner label"
required: true
type: choice
options:
- 1-gpu-h100
- 1-gpu-5090
- 2-gpu-h100
- 4-gpu-h100
- 4-gpu-a10
- 4-gpu-b200
- 8-gpu-h200
- 8-gpu-h200-deepep
- 8-gpu-h20
- 8-gpu-b200
- ubuntu-latest
runs_on:
description: "GHA runner label (cuda/multimodal_gen only; ignored for cpu)"
required: false
type: string
default: ""
install_script:
description: "Install script path (cuda only). E.g. scripts/ci/cuda/ci_install_dependency.sh"
required: false
type: string
default: ""
install_timeout:
description: "Install-step timeout minutes (cuda only)"
required: false
type: string
default: "20"
rdma_devices:
description: "SGLANG_CI_RDMA_ALL_DEVICES csv (cuda only; empty = unset)"
required: false
type: string
default: ""
pr_head_sha:
description: "PR head SHA to checkout (for /rerun-test on fork PRs)"
required: false
type: string
default: ""
use_deepep:
description: "Use ci_install_deepep.sh instead of ci_install_dependency.sh"
required: false
type: string
default: "false"
is_cpu:
description: "Run as CPU-only test (uses ubuntu-latest with uv pip install)"
required: false
type: string
default: "false"
install_diffusion:
description: "Install diffusion dependencies (for multimodal gen tests)"
required: false
type: string
default: "false"
reply_comment_id:
description: "Reply comment ID to write back result to"
required: false
@@ -69,15 +66,15 @@ permissions:
jobs:
rerun-test-cuda:
if: inputs.is_cpu != 'true'
runs-on: ${{ inputs.runner_label }}
if: inputs.mode == 'cuda'
runs-on: ${{ inputs.runs_on }}
timeout-minutes: 120
permissions:
contents: read
issues: write
env:
RUNNER_LABELS: ${{ inputs.runner_label }}
SGLANG_CI_RDMA_ALL_DEVICES: ${{ inputs.runner_label == '8-gpu-h20' && 'mlx5_1,mlx5_2,mlx5_3,mlx5_4' || '' }}
RUNNER_LABELS: ${{ inputs.runs_on }}
SGLANG_CI_RDMA_ALL_DEVICES: ${{ inputs.rdma_devices }}
steps:
- name: Checkout code
uses: actions/checkout@v4
@@ -90,7 +87,7 @@ jobs:
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
if [[ "${{ inputs.runner_label }}" == "1-gpu-5090" ]]; then
if [[ "${{ inputs.runs_on }}" == "1-gpu-5090" ]]; then
source /etc/profile.d/sglang-ci.sh
fi
python3 scripts/ci/utils/update_rerun_test_status.py \
@@ -102,26 +99,19 @@ jobs:
- uses: ./.github/actions/check-maintenance
- name: Install dependencies
timeout-minutes: 20
timeout-minutes: ${{ fromJson(inputs.install_timeout) }}
run: |
if [[ "${{ inputs.runner_label }}" == "1-gpu-5090" ]]; then
if [[ "${{ inputs.runs_on }}" == "1-gpu-5090" ]]; then
source /etc/profile.d/sglang-ci.sh
fi
if [[ "${{ inputs.use_deepep }}" == "true" ]]; then
bash scripts/ci/cuda/ci_install_deepep.sh
elif [[ "${{ inputs.install_diffusion }}" == "true" ]]; then
bash scripts/ci/cuda/ci_install_dependency.sh diffusion
else
bash scripts/ci/cuda/ci_install_dependency.sh
fi
bash ${{ inputs.install_script }}
- name: Run test
timeout-minutes: 60
run: |
if [[ "${{ inputs.runner_label }}" == "1-gpu-5090" ]]; then
if [[ "${{ inputs.runs_on }}" == "1-gpu-5090" ]]; then
source /etc/profile.d/sglang-ci.sh
fi
# Collect non-empty commands into an array for counting.
cmds=()
while IFS= read -r cmd; do
[ -z "$cmd" ] && continue
@@ -134,17 +124,74 @@ jobs:
cmd="${cmds[$idx]}"
echo ""
echo "."
if [[ "${{ inputs.install_diffusion }}" == "true" ]]; then
echo "Begin ($i/$total): python3 -m pytest $cmd -x"
echo "."
file_start=$SECONDS
python3 -m pytest $cmd -x || exit 1
else
echo "Begin ($i/$total): python3 $cmd"
echo "."
file_start=$SECONDS
(cd test/ && python3 $cmd -f) || exit 1
fi
echo "Begin ($i/$total): python3 $cmd"
echo "."
file_start=$SECONDS
(cd test/ && python3 $cmd -f) || exit 1
elapsed=$(( SECONDS - file_start ))
echo "."
echo "End ($i/$total): elapsed=${elapsed}s"
echo "."
echo ""
done
total_elapsed=$(( SECONDS - suite_start ))
echo "All $total test(s) passed in ${total_elapsed}s"
- uses: ./.github/actions/upload-cuda-coredumps
if: failure()
rerun-test-multimodal-gen:
if: inputs.mode == 'multimodal_gen'
runs-on: ${{ inputs.runs_on }}
timeout-minutes: 120
permissions:
contents: read
issues: write
env:
RUNNER_LABELS: ${{ inputs.runs_on }}
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ inputs.pr_head_sha || github.sha }}
- name: Mark runner picked up
if: inputs.reply_comment_id != '' && inputs.reply_marker != ''
continue-on-error: true
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
python3 scripts/ci/utils/update_rerun_test_status.py \
--comment-id "${{ inputs.reply_comment_id }}" \
--marker "${{ inputs.reply_marker }}" \
--status running \
--repo "${{ github.repository }}"
- uses: ./.github/actions/check-maintenance
- name: Install dependencies (diffusion)
timeout-minutes: 20
run: bash scripts/ci/cuda/ci_install_dependency.sh diffusion
- name: Run test
timeout-minutes: 60
run: |
cmds=()
while IFS= read -r cmd; do
[ -z "$cmd" ] && continue
cmds+=("$cmd")
done <<< "${{ inputs.test_command }}"
total=${#cmds[@]}
suite_start=$SECONDS
for idx in "${!cmds[@]}"; do
i=$((idx + 1))
cmd="${cmds[$idx]}"
echo ""
echo "."
echo "Begin ($i/$total): python3 -m pytest $cmd -x"
echo "."
file_start=$SECONDS
python3 -m pytest $cmd -x || exit 1
elapsed=$(( SECONDS - file_start ))
echo "."
echo "End ($i/$total): elapsed=${elapsed}s"
@@ -158,7 +205,7 @@ jobs:
if: failure()
rerun-test-cpu:
if: inputs.is_cpu == 'true'
if: inputs.mode == 'cpu'
runs-on: ubuntu-latest
timeout-minutes: 120
permissions:
@@ -214,7 +261,6 @@ jobs:
timeout-minutes: 60
run: |
cd test/
# Collect non-empty commands into an array for counting.
cmds=()
while IFS= read -r cmd; do
[ -z "$cmd" ] && continue
@@ -241,7 +287,7 @@ jobs:
echo "All $total test(s) passed in ${total_elapsed}s"
write-back-result:
needs: [rerun-test-cuda, rerun-test-cpu]
needs: [rerun-test-cuda, rerun-test-multimodal-gen, rerun-test-cpu]
if: always() && inputs.reply_comment_id != '' && inputs.reply_marker != ''
runs-on: ubuntu-latest
permissions:
@@ -263,7 +309,9 @@ jobs:
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
if [[ "${{ needs.rerun-test-cuda.result }}" == "success" || "${{ needs.rerun-test-cpu.result }}" == "success" ]]; then
if [[ "${{ needs.rerun-test-cuda.result }}" == "success" \
|| "${{ needs.rerun-test-multimodal-gen.result }}" == "success" \
|| "${{ needs.rerun-test-cpu.result }}" == "success" ]]; then
STATUS=success
else
STATUS=failure
+1 -1
View File
@@ -86,7 +86,7 @@ jobs:
- name: Install dependencies
run: |
pip install PyGithub
pip install PyGithub PyYAML
- name: Handle Slash Command
env: