Add label to build #258
This file contains hidden or 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: PR - vLLM | |
| on: | |
| pull_request: | |
| branches: [main] | |
| types: [opened, reopened, synchronize] | |
| paths: | |
| - "**vllm**" | |
| - "!docs/**" | |
| permissions: | |
| contents: read | |
| pull-requests: read | |
| env: | |
| # CI environment configuration | |
| FORCE_COLOR: "1" | |
| # Config file paths | |
| EC2_CONFIG: ".github/config/vllm-ec2.yml" | |
| SAGEMAKER_CONFIG: ".github/config/vllm-sagemaker.yml" | |
| RAYSERVE_CONFIG: ".github/config/vllm-rayserve.yml" | |
| jobs: | |
| gatekeeper: | |
| runs-on: ubuntu-latest | |
| concurrency: | |
| group: ${{ github.workflow }}-gate-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout base branch (safe) | |
| uses: actions/checkout@v5 | |
| with: | |
| ref: ${{ github.event.pull_request.base.sha }} | |
| fetch-depth: 1 | |
| - name: Run permission gate (from base) | |
| uses: ./.github/actions/pr-permission-gate | |
| load-config: | |
| needs: [gatekeeper] | |
| if: success() | |
| runs-on: ubuntu-latest | |
| outputs: | |
| ec2-config: ${{ steps.load.outputs.ec2-config }} | |
| sagemaker-config: ${{ steps.load.outputs.sagemaker-config }} | |
| rayserve-config: ${{ steps.load.outputs.rayserve-config }} | |
| steps: | |
| - name: Checkout code | |
| uses: actions/checkout@v5 | |
| - name: Load configurations | |
| id: load | |
| uses: ./.github/actions/load-config | |
| with: | |
| config-files: ${{ env.EC2_CONFIG }},${{ env.SAGEMAKER_CONFIG }},${{ env.RAYSERVE_CONFIG }} | |
| check-changes: | |
| needs: [load-config] | |
| if: success() | |
| runs-on: ubuntu-latest | |
| concurrency: | |
| group: ${{ github.workflow }}-check-changes-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| outputs: | |
| build-change: ${{ steps.changes.outputs.build-change }} | |
| test-change: ${{ steps.changes.outputs.test-change }} | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Setup python | |
| uses: actions/setup-python@v6 | |
| with: | |
| python-version: "3.12" | |
| - name: Run pre-commit | |
| uses: pre-commit/action@v3.0.1 | |
| with: | |
| extra_args: --all-files | |
| - name: Detect file changes | |
| id: changes | |
| uses: dorny/paths-filter@v3 | |
| with: | |
| filters: | | |
| build-change: | |
| - "docker/vllm/**" | |
| - "scripts/vllm/**" | |
| - "scripts/common/**" | |
| - "scripts/telemetry/**" | |
| - ".github/workflows/pr-vllm*" | |
| test-change: | |
| - "test/vllm/**" | |
| # ============================================== | |
| # =============== vLLM EC2 jobs ================ | |
| # ============================================== | |
| build-vllm-ec2-image: | |
| needs: [check-changes, load-config] | |
| if: needs.check-changes.outputs.build-change == 'true' | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-build-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-build-vllm-ec2-image-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| outputs: | |
| ci-image: ${{ steps.build.outputs.image-uri }} | |
| steps: | |
| - uses: actions/checkout@v5 | |
| - name: Parse EC2 config | |
| id: config | |
| run: | | |
| echo '${{ needs.load-config.outputs.ec2-config }}' > config.json | |
| echo "framework=$(jq -r '.common.framework' config.json)" >> $GITHUB_OUTPUT | |
| echo "framework-version=$(jq -r '.common.framework_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "container-type=$(jq -r '.common.job_type' config.json)" >> $GITHUB_OUTPUT | |
| echo "python-version=$(jq -r '.common.python_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "cuda-version=$(jq -r '.common.cuda_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "os-version=$(jq -r '.common.os_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "device-type=$(jq -r '.common.device_type // "gpu"' config.json)" >> $GITHUB_OUTPUT | |
| echo "arch-type=$(jq -r '.common.arch_type // "x86"' config.json)" >> $GITHUB_OUTPUT | |
| echo "contributor=$(jq -r '.common.contributor // "None"' config.json)" >> $GITHUB_OUTPUT | |
| echo "customer-type=$(jq -r '.common.customer_type // ""' config.json)" >> $GITHUB_OUTPUT | |
| - name: Build image | |
| id: build | |
| uses: ./.github/actions/build-image | |
| with: | |
| framework: ${{ steps.config.outputs.framework }} | |
| target: vllm-ec2 | |
| base-image: vllm/vllm-openai:v${{ steps.config.outputs.framework-version }} | |
| framework-version: ${{ steps.config.outputs.framework-version }} | |
| container-type: ${{ steps.config.outputs.container-type }} | |
| aws-account-id: ${{ vars.CI_AWS_ACCOUNT_ID }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| tag-pr: ${{ steps.config.outputs.framework }}-${{ steps.config.outputs.framework-version }}-gpu-${{ steps.config.outputs.python-version }}-${{ steps.config.outputs.cuda-version }}-${{ steps.config.outputs.os-version }}-ec2-pr-${{ github.event.pull_request.number }} | |
| dockerfile-path: docker/${{ steps.config.outputs.framework }}/Dockerfile | |
| arch-type: ${{ steps.config.outputs.arch-type }} | |
| device-type: ${{ steps.config.outputs.device-type }} | |
| cuda-version: ${{ steps.config.outputs.cuda-version }} | |
| python-version: ${{ steps.config.outputs.python-version }} | |
| os-version: ${{ steps.config.outputs.os-version }} | |
| contributor: ${{ steps.config.outputs.contributor }} | |
| customer-type: ${{ steps.config.outputs.customer-type }} | |
| set-ec2-test-environment: | |
| needs: [check-changes, build-vllm-ec2-image, load-config] | |
| if: | | |
| always() && !failure() && !cancelled() && | |
| (needs.check-changes.outputs.build-change == 'true' || needs.check-changes.outputs.test-change == 'true') | |
| runs-on: ubuntu-latest | |
| concurrency: | |
| group: ${{ github.workflow }}-set-ec2-test-environment-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| outputs: | |
| aws-account-id: ${{ steps.set-env.outputs.AWS_ACCOUNT_ID }} | |
| image-uri: ${{ steps.set-env.outputs.IMAGE_URI }} | |
| framework-version: ${{ steps.config.outputs.framework-version }} | |
| steps: | |
| - name: Checkout code | |
| uses: actions/checkout@v4 | |
| - name: Parse EC2 config | |
| id: config | |
| run: | | |
| echo '${{ needs.load-config.outputs.ec2-config }}' > config.json | |
| echo "framework-version=$(jq -r '.common.framework_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "prod-image=$(jq -r '.common.prod_image' config.json)" >> $GITHUB_OUTPUT | |
| - name: Set test environment | |
| id: set-env | |
| run: | | |
| if [[ "${{ needs.build-vllm-ec2-image.result }}" == "success" ]]; then | |
| AWS_ACCOUNT_ID=${{ vars.CI_AWS_ACCOUNT_ID }} | |
| IMAGE_URI=${{ needs.build-vllm-ec2-image.outputs.ci-image }} | |
| else | |
| AWS_ACCOUNT_ID=${{ vars.PROD_AWS_ACCOUNT_ID }} | |
| IMAGE_URI=${{ vars.PROD_AWS_ACCOUNT_ID }}.dkr.ecr.${{ vars.AWS_REGION }}.amazonaws.com/${{ steps.config.outputs.prod-image }} | |
| fi | |
| echo "Image URI to test: ${IMAGE_URI}" | |
| echo "AWS_ACCOUNT_ID=${AWS_ACCOUNT_ID}" >> ${GITHUB_OUTPUT} | |
| echo "IMAGE_URI=${IMAGE_URI}" >> ${GITHUB_OUTPUT} | |
| vllm-ec2-regression-test: | |
| needs: [build-vllm-ec2-image, set-ec2-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-ec2-regression-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-ec2-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-ec2-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-ec2-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-ec2-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_15_1_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_regression_test.sh | |
| vllm-ec2-cuda-test: | |
| needs: [build-vllm-ec2-image, set-ec2-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-ec2-cuda-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-ec2-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-ec2-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-ec2-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-ec2-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_15_1_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_cuda_test.sh | |
| vllm-ec2-example-test: | |
| needs: [build-vllm-ec2-image, set-ec2-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-ec2-example-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-ec2-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-ec2-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-ec2-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-ec2-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_15_1_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_ec2_examples_test.sh | |
| # =================================================== | |
| # =============== vLLM RayServe jobs ================ | |
| # =================================================== | |
| build-vllm-rayserve-image: | |
| needs: [check-changes, load-config] | |
| if: needs.check-changes.outputs.build-change == 'true' | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-build-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-build-vllm-rayserve-image-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| outputs: | |
| ci-image: ${{ steps.build.outputs.image-uri }} | |
| steps: | |
| - uses: actions/checkout@v5 | |
| - name: Parse RayServe config | |
| id: config | |
| run: | | |
| echo '${{ needs.load-config.outputs.rayserve-config }}' > config.json | |
| echo "framework=$(jq -r '.common.framework' config.json)" >> $GITHUB_OUTPUT | |
| echo "framework-version=$(jq -r '.common.framework_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "container-type=$(jq -r '.common.job_type' config.json)" >> $GITHUB_OUTPUT | |
| echo "python-version=$(jq -r '.common.python_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "cuda-version=$(jq -r '.common.cuda_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "os-version=$(jq -r '.common.os_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "device-type=$(jq -r '.common.device_type // "gpu"' config.json)" >> $GITHUB_OUTPUT | |
| echo "arch-type=$(jq -r '.common.arch_type // "x86"' config.json)" >> $GITHUB_OUTPUT | |
| echo "contributor=$(jq -r '.common.contributor // "None"' config.json)" >> $GITHUB_OUTPUT | |
| echo "customer-type=$(jq -r '.common.customer_type // ""' config.json)" >> $GITHUB_OUTPUT | |
| - name: Build image | |
| id: build | |
| uses: ./.github/actions/build-image | |
| with: | |
| framework: ${{ steps.config.outputs.framework }} | |
| target: vllm-rayserve-ec2 | |
| base-image: vllm/vllm-openai:v${{ steps.config.outputs.framework-version }} | |
| framework-version: ${{ steps.config.outputs.framework-version }} | |
| container-type: ${{ steps.config.outputs.container-type }} | |
| aws-account-id: ${{ vars.CI_AWS_ACCOUNT_ID }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| tag-pr: ${{ steps.config.outputs.framework }}-${{ steps.config.outputs.framework-version }}-gpu-${{ steps.config.outputs.python-version }}-${{ steps.config.outputs.cuda-version }}-${{ steps.config.outputs.os-version }}-rayserve-ec2-pr-${{ github.event.pull_request.number }} | |
| dockerfile-path: docker/${{ steps.config.outputs.framework }}/Dockerfile | |
| arch-type: ${{ steps.config.outputs.arch-type }} | |
| device-type: ${{ steps.config.outputs.device-type }} | |
| cuda-version: ${{ steps.config.outputs.cuda-version }} | |
| python-version: ${{ steps.config.outputs.python-version }} | |
| os-version: ${{ steps.config.outputs.os-version }} | |
| contributor: ${{ steps.config.outputs.contributor }} | |
| customer-type: ${{ steps.config.outputs.customer-type }} | |
| set-rayserve-test-environment: | |
| needs: [check-changes, build-vllm-rayserve-image, load-config] | |
| if: | | |
| always() && !failure() && !cancelled() && | |
| (needs.check-changes.outputs.build-change == 'true' || needs.check-changes.outputs.test-change == 'true') | |
| runs-on: ubuntu-latest | |
| concurrency: | |
| group: ${{ github.workflow }}-set-rayserve-test-environment-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| outputs: | |
| aws-account-id: ${{ steps.set-env.outputs.AWS_ACCOUNT_ID }} | |
| image-uri: ${{ steps.set-env.outputs.IMAGE_URI }} | |
| framework-version: ${{ steps.config.outputs.framework-version }} | |
| steps: | |
| - name: Checkout code | |
| uses: actions/checkout@v4 | |
| - name: Parse RayServe config | |
| id: config | |
| run: | | |
| echo '${{ needs.load-config.outputs.rayserve-config }}' > config.json | |
| echo "framework-version=$(jq -r '.common.framework_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "prod-image=$(jq -r '.common.prod_image' config.json)" >> $GITHUB_OUTPUT | |
| - name: Set test environment | |
| id: set-env | |
| run: | | |
| if [[ "${{ needs.build-vllm-rayserve-image.result }}" == "success" ]]; then | |
| AWS_ACCOUNT_ID=${{ vars.CI_AWS_ACCOUNT_ID }} | |
| IMAGE_URI=${{ needs.build-vllm-rayserve-image.outputs.ci-image }} | |
| else | |
| AWS_ACCOUNT_ID=${{ vars.PROD_AWS_ACCOUNT_ID }} | |
| IMAGE_URI=${{ vars.PROD_AWS_ACCOUNT_ID }}.dkr.ecr.${{ vars.AWS_REGION }}.amazonaws.com/${{ steps.config.outputs.prod-image }} | |
| fi | |
| echo "Image URI to test: ${IMAGE_URI}" | |
| echo "AWS_ACCOUNT_ID=${AWS_ACCOUNT_ID}" >> ${GITHUB_OUTPUT} | |
| echo "IMAGE_URI=${IMAGE_URI}" >> ${GITHUB_OUTPUT} | |
| vllm-rayserve-regression-test: | |
| needs: [build-vllm-rayserve-image, set-rayserve-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-rayserve-regression-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-rayserve-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-rayserve-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-rayserve-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-rayserve-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_10_2_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_regression_test.sh | |
| vllm-rayserve-cuda-test: | |
| needs: [build-vllm-rayserve-image, set-rayserve-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-rayserve-cuda-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-rayserve-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-rayserve-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-rayserve-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-rayserve-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_10_2_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_cuda_test.sh | |
| vllm-rayserve-example-test: | |
| needs: [build-vllm-rayserve-image, set-rayserve-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-rayserve-example-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-rayserve-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-rayserve-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-rayserve-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-rayserve-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_10_2_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_rayserve_examples_test.sh | |
| # ==================================================== | |
| # =============== vLLM SageMaker jobs ================ | |
| # ==================================================== | |
| build-vllm-sagemaker-image: | |
| needs: [check-changes, load-config] | |
| if: needs.check-changes.outputs.build-change == 'true' | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-build-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-build-vllm-sagemaker-image-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| outputs: | |
| ci-image: ${{ steps.build.outputs.image-uri }} | |
| steps: | |
| - uses: actions/checkout@v5 | |
| - name: Parse SageMaker config | |
| id: config | |
| run: | | |
| echo '${{ needs.load-config.outputs.sagemaker-config }}' > config.json | |
| echo "framework=$(jq -r '.common.framework' config.json)" >> $GITHUB_OUTPUT | |
| echo "framework-version=$(jq -r '.common.framework_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "container-type=$(jq -r '.common.job_type' config.json)" >> $GITHUB_OUTPUT | |
| echo "python-version=$(jq -r '.common.python_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "cuda-version=$(jq -r '.common.cuda_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "os-version=$(jq -r '.common.os_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "device-type=$(jq -r '.common.device_type // "gpu"' config.json)" >> $GITHUB_OUTPUT | |
| echo "arch-type=$(jq -r '.common.arch_type // "x86"' config.json)" >> $GITHUB_OUTPUT | |
| echo "contributor=$(jq -r '.common.contributor // "None"' config.json)" >> $GITHUB_OUTPUT | |
| echo "customer-type=$(jq -r '.common.customer_type // ""' config.json)" >> $GITHUB_OUTPUT | |
| - name: Build image | |
| id: build | |
| uses: ./.github/actions/build-image | |
| with: | |
| framework: ${{ steps.config.outputs.framework }} | |
| target: vllm-sagemaker | |
| base-image: vllm/vllm-openai:v${{ steps.config.outputs.framework-version }} | |
| framework-version: ${{ steps.config.outputs.framework-version }} | |
| container-type: ${{ steps.config.outputs.container-type }} | |
| aws-account-id: ${{ vars.CI_AWS_ACCOUNT_ID }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| tag-pr: ${{ steps.config.outputs.framework }}-${{ steps.config.outputs.framework-version }}-gpu-${{ steps.config.outputs.python-version }}-${{ steps.config.outputs.cuda-version }}-${{ steps.config.outputs.os-version }}-sagemaker-pr-${{ github.event.pull_request.number }} | |
| dockerfile-path: docker/${{ steps.config.outputs.framework }}/Dockerfile | |
| arch-type: ${{ steps.config.outputs.arch-type }} | |
| device-type: ${{ steps.config.outputs.device-type }} | |
| cuda-version: ${{ steps.config.outputs.cuda-version }} | |
| python-version: ${{ steps.config.outputs.python-version }} | |
| os-version: ${{ steps.config.outputs.os-version }} | |
| contributor: ${{ steps.config.outputs.contributor }} | |
| customer-type: ${{ steps.config.outputs.customer-type }} | |
| set-sagemaker-test-environment: | |
| needs: [check-changes, build-vllm-sagemaker-image, load-config] | |
| if: | | |
| always() && !failure() && !cancelled() && | |
| (needs.check-changes.outputs.build-change == 'true' || needs.check-changes.outputs.test-change == 'true') | |
| runs-on: ubuntu-latest | |
| concurrency: | |
| group: ${{ github.workflow }}-set-sagemaker-test-environment-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| outputs: | |
| aws-account-id: ${{ steps.set-env.outputs.AWS_ACCOUNT_ID }} | |
| image-uri: ${{ steps.set-env.outputs.IMAGE_URI }} | |
| framework-version: ${{ steps.config.outputs.framework-version }} | |
| steps: | |
| - name: Checkout code | |
| uses: actions/checkout@v4 | |
| - name: Parse SageMaker config | |
| id: config | |
| run: | | |
| echo '${{ needs.load-config.outputs.sagemaker-config }}' > config.json | |
| echo "framework-version=$(jq -r '.common.framework_version' config.json)" >> $GITHUB_OUTPUT | |
| echo "prod-image=$(jq -r '.common.prod_image' config.json)" >> $GITHUB_OUTPUT | |
| - name: Set test environment | |
| id: set-env | |
| run: | | |
| if [[ "${{ needs.build-vllm-sagemaker-image.result }}" == "success" ]]; then | |
| AWS_ACCOUNT_ID=${{ vars.CI_AWS_ACCOUNT_ID }} | |
| IMAGE_URI=${{ needs.build-vllm-sagemaker-image.outputs.ci-image }} | |
| else | |
| AWS_ACCOUNT_ID=${{ vars.PROD_AWS_ACCOUNT_ID }} | |
| IMAGE_URI=${{ vars.PROD_AWS_ACCOUNT_ID }}.dkr.ecr.${{ vars.AWS_REGION }}.amazonaws.com/${{ steps.config.outputs.prod-image }} | |
| fi | |
| echo "Image URI to test: ${IMAGE_URI}" | |
| echo "AWS_ACCOUNT_ID=${AWS_ACCOUNT_ID}" >> ${GITHUB_OUTPUT} | |
| echo "IMAGE_URI=${IMAGE_URI}" >> ${GITHUB_OUTPUT} | |
| vllm-sagemaker-regression-test: | |
| needs: [build-vllm-sagemaker-image, set-sagemaker-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-sagemaker-regression-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-sagemaker-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-sagemaker-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-sagemaker-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-sagemaker-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_15_1_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_regression_test.sh | |
| vllm-sagemaker-cuda-test: | |
| needs: [build-vllm-sagemaker-image, set-sagemaker-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-sagemaker-cuda-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-sagemaker-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-sagemaker-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-sagemaker-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-sagemaker-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_15_1_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_cuda_test.sh | |
| vllm-sagemaker-example-test: | |
| needs: [build-vllm-sagemaker-image, set-sagemaker-test-environment] | |
| if: success() | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:x86-g6xl-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-sagemaker-example-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: true | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Container pull | |
| uses: ./.github/actions/ecr-authenticate | |
| with: | |
| aws-account-id: ${{ needs.set-sagemaker-test-environment.outputs.aws-account-id }} | |
| aws-region: ${{ vars.AWS_REGION }} | |
| image-uri: ${{ needs.set-sagemaker-test-environment.outputs.image-uri }} | |
| - name: Checkout vLLM tests | |
| uses: actions/checkout@v5 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v${{ needs.set-sagemaker-test-environment.outputs.framework-version }} | |
| path: vllm_source | |
| - name: Start container | |
| run: | | |
| CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \ | |
| -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \ | |
| -v ${HOME}/.cache/vllm:/root/.cache/vllm \ | |
| -v .:/workdir --workdir /workdir \ | |
| -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \ | |
| ${{ needs.set-sagemaker-test-environment.outputs.image-uri }}) | |
| echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV | |
| - name: Setup for vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_0_15_1_test_setup.sh | |
| - name: Run vLLM tests | |
| run: | | |
| docker exec ${CONTAINER_ID} scripts/vllm/vllm_sagemaker_examples_test.sh | |
| vllm-sagemaker-endpoint-test: | |
| needs: [set-sagemaker-test-environment] | |
| if: | | |
| always() && !failure() && !cancelled() && | |
| needs.set-sagemaker-test-environment.result == 'success' | |
| runs-on: | |
| - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }} | |
| fleet:default-runner | |
| buildspec-override:true | |
| concurrency: | |
| group: ${{ github.workflow }}-vllm-sagemaker-endpoint-test-${{ github.event.pull_request.number }} | |
| cancel-in-progress: false | |
| steps: | |
| - name: Checkout DLC source | |
| uses: actions/checkout@v5 | |
| - name: Install test dependencies | |
| run: | | |
| uv venv --python 3.12 | |
| source .venv/bin/activate | |
| uv pip install -r test/requirements.txt | |
| uv pip install -r test/vllm/sagemaker/requirements.txt | |
| - name: Run sagemaker endpoint test | |
| run: | | |
| source .venv/bin/activate | |
| cd test/ | |
| python3 -m pytest -vs -rA --image-uri ${{ needs.set-sagemaker-test-environment.outputs.image-uri }} vllm/sagemaker |