Benchmark Nightly #261
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: Benchmark Nightly | |
| on: | |
| workflow_dispatch: | |
| schedule: | |
| - cron: '0 1 * * *' | |
| permissions: | |
| id-token: write | |
| contents: read | |
| jobs: | |
| create-runners: | |
| runs-on: [ self-hosted, scheduler ] | |
| steps: | |
| - name: Create new G6 instance | |
| id: create_gpu | |
| run: | | |
| cd /home/ubuntu/djl_benchmark_script/scripts | |
| token=$( curl -X POST -H "Authorization: token ${{ secrets.ACTION_RUNNER_PERSONAL_TOKEN }}" \ | |
| https://api.github.com/repos/deepjavalibrary/djl-serving/actions/runners/registration-token \ | |
| --fail \ | |
| | jq '.token' | tr -d '"' ) | |
| ./start_instance.sh action_g6 $token djl-serving | |
| outputs: | |
| gpu_instance_id: ${{ steps.create_gpu.outputs.action_g6_instance_id }} | |
| g5-2xl: | |
| uses: ./.github/workflows/instant_benchmark.yml | |
| secrets: inherit | |
| with: | |
| running_template: ./benchmark/nightly/g5-2xl.txt | |
| instance: g5.2xlarge | |
| record: cloudwatch | |
| g5-12xl: | |
| uses: ./.github/workflows/instant_benchmark.yml | |
| secrets: inherit | |
| with: | |
| running_template: ./benchmark/nightly/g5-12xl.txt | |
| instance: g5.12xlarge | |
| record: cloudwatch | |
| g5-48xl: | |
| uses: ./.github/workflows/instant_benchmark.yml | |
| secrets: inherit | |
| with: | |
| running_template: ./benchmark/nightly/g5-48xl.txt | |
| instance: g5.48xlarge | |
| record: cloudwatch | |
| handler-performance-test: | |
| runs-on: [ self-hosted, g6 ] | |
| timeout-minutes: 60 | |
| needs: create-runners | |
| strategy: | |
| fail-fast: false | |
| matrix: | |
| test: | |
| - test: TestGPUHandlerPerformance | |
| instance: g6 | |
| steps: | |
| - uses: actions/checkout@v4 | |
| - name: Clean env | |
| run: | | |
| yes | docker system prune -a --volumes | |
| sudo rm -rf /home/ubuntu/actions-runner/_work/_tool/Java_Corretto_jdk/ | |
| echo "wait dpkg lock..." | |
| while sudo fuser /var/{lib/{dpkg,apt/lists},cache/apt/archives}/lock >/dev/null 2>&1; do sleep 5; done | |
| - name: Set up Python3 | |
| uses: actions/setup-python@v5 | |
| with: | |
| python-version: '3.10.x' | |
| - name: Install pip dependencies | |
| run: pip3 install pytest requests "numpy<2" pillow huggingface_hub | |
| - name: Install torch | |
| # Use torch to get cuda capability of current device to selectively run tests | |
| # Torch version doesn't really matter that much | |
| run: | | |
| pip3 install torch==2.3.0 | |
| - name: Install awscurl | |
| working-directory: tests/integration | |
| run: | | |
| wget https://publish.djl.ai/awscurl/awscurl | |
| chmod +x awscurl | |
| mkdir outputs | |
| - name: Test | |
| working-directory: tests/integration | |
| env: | |
| TEST_DJL_VERSION: nightly | |
| run: | | |
| python -m pytest -k ${{ matrix.test.test }} tests.py | |
| - name: Cleanup | |
| working-directory: tests/integration | |
| run: | | |
| rm -rf outputs | |
| rm awscurl | |
| - name: On Failure | |
| if: ${{ failure() }} | |
| working-directory: tests/integration | |
| run: | | |
| for file in outputs/*; do if [ -f "$file" ]; then echo "Contents of $file:"; cat "$file"; echo; fi; done | |
| sudo rm -rf outputs && sudo rm -rf models | |
| rm awscurl | |
| ./remove_container.sh | |
| - name: Upload test logs | |
| if: ${{ always() }} | |
| uses: actions/upload-artifact@v4 | |
| with: | |
| name: test-${{ matrix.test.test }}-logs | |
| path: tests/integration/all_logs/ | |
| stop-g6-runners: | |
| if: always() | |
| runs-on: [ self-hosted, scheduler ] | |
| needs: [ create-runners, handler-performance-test ] | |
| steps: | |
| - name: Stop g6 instances | |
| run: | | |
| cd /home/ubuntu/djl_benchmark_script/scripts | |
| instance_id=${{ needs.create-runners.outputs.gpu_instance_id }} | |
| ./stop_instance.sh $instance_id |