Merge pull request #805 from javipolo/not-blacklist-amdgpu #187
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: Training Bootc image builds | |
| on: | |
| push: | |
| branches: [ main ] | |
| paths: | |
| - 'training/**' | |
| - '.github/workflows/training_bootc.yaml' | |
| workflow_dispatch: | |
| concurrency: | |
| group: ${{ github.workflow }} | |
| cancel-in-progress: false | |
| env: | |
| REGISTRY: quay.io | |
| REGISTRY_ORG: ai-lab | |
| REGION: us-east-1 | |
| jobs: | |
| start-runner: | |
| name: Start self-hosted EC2 runner | |
| if: "!contains(github.event.pull_request.labels.*.name, 'hold-tests')" | |
| runs-on: ubuntu-latest | |
| outputs: | |
| label: ${{ steps.start-ec2-runner.outputs.label }} | |
| ec2-instance-id: ${{ steps.start-ec2-runner.outputs.ec2-instance-id }} | |
| steps: | |
| - name: Configure AWS credentials | |
| uses: aws-actions/configure-aws-credentials@v1 | |
| with: | |
| aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }} | |
| aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }} | |
| aws-region: ${{ env.REGION }} | |
| - name: Start EC2 runner | |
| id: start-ec2-runner | |
| uses: machulav/ec2-github-runner@v2 | |
| with: | |
| mode: start | |
| github-token: ${{ secrets.GH_PERSONAL_ACCESS_TOKEN }} | |
| ec2-image-id: ami-0154957ba4ce98784 | |
| ec2-instance-type: m7i.12xlarge | |
| subnet-id: subnet-0b1e1d94240813658 | |
| security-group-id: sg-055105753f5e8bd83 | |
| nvidia-bootc-builder-image: | |
| if: "!contains(github.event.pull_request.labels.*.name, 'hold-tests')" | |
| strategy: | |
| matrix: | |
| include: | |
| - image_name: nvidia-builder | |
| context: training/nvidia-bootc | |
| arch: amd64 | |
| runs-on: ${{ needs.start-runner.outputs.label }} | |
| needs: start-runner | |
| permissions: | |
| contents: read | |
| packages: write | |
| steps: | |
| - uses: actions/[email protected] | |
| - name: mkdir root/.docker directory | |
| run: | | |
| mkdir -p ~/.docker | |
| - name: Login to Container Registry | |
| run: podman login -u ${{ secrets.REGISTRY_USER }} -p ${{ secrets.REGISTRY_PASSWORD }} ${{ env.REGISTRY }} | |
| - name: generate a ssh key - USER SHOULD INJECT THEIR OWN AND REBUILD IF THEY USE THIS IMAGE | |
| run: | | |
| ssh-keygen -t rsa -b 4096 -f ~/.ssh/id_rsa -N "" | |
| - name: Build Image | |
| id: build_image | |
| run: make driver-toolkit ARCH=${{ matrix.arch }} | |
| working-directory: ${{ matrix.context }} | |
| - name: tag image as nvidia-builder | |
| run: podman tag ${{ env.REGISTRY }}/${{ env.REGISTRY_ORG }}/driver-toolkit:latest ${{ env.REGISTRY }}/${{ env.REGISTRY_ORG }}/${{ matrix.image_name}}:latest | |
| - name: Push image | |
| if: github.event_name == 'push' && github.ref == 'refs/heads/main' | |
| uses: redhat-actions/[email protected] | |
| with: | |
| registry: ${{ env.REGISTRY }}/${{ env.REGISTRY_ORG }} | |
| image: driver-toolkit | |
| tags: latest | |
| - name: push the nvidia-builder image | |
| if: github.event_name == 'push' && github.ref == 'refs/heads/main' | |
| uses: redhat-actions/[email protected] | |
| with: | |
| image: ${{ matrix.image_name}} | |
| tags: latest | |
| registry: ${{ env.REGISTRY }}/${{ env.REGISTRY_ORG }} | |
| - name: Publish Job Results to Slack | |
| id: slack | |
| if: always() | |
| uses: slackapi/[email protected] | |
| with: | |
| payload: | | |
| { | |
| "text": "${{ github.workflow }} workflow status: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}" | |
| } | |
| env: | |
| SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }} | |
| nvidia-bootc-image: | |
| strategy: | |
| matrix: | |
| include: | |
| - image_name: nvidia-bootc | |
| driver_version: "550.54.15" | |
| context: training/nvidia-bootc | |
| arch: amd64 | |
| runs-on: ${{ needs.start-runner.outputs.label }} | |
| if: "!contains(github.event.pull_request.labels.*.name, 'hold-tests')" | |
| needs: [ nvidia-bootc-builder-image, start-runner ] | |
| steps: | |
| - uses: actions/[email protected] | |
| - name: mkdir root/.docker directory | |
| run: | | |
| mkdir -p ~/.docker | |
| - name: generate a ssh key - USER SHOULD INJECT THEIR OWN AND REBUILD IF THEY USE THIS IMAGE and overwrite the existing one | |
| run: | | |
| ssh-keygen -t rsa -b 4096 -f ~/.ssh/id_rsa -N "" <<<y | |
| - name: Login to Container Registry | |
| run: podman login -u ${{ secrets.REGISTRY_USER }} -p ${{ secrets.REGISTRY_PASSWORD }} ${{ env.REGISTRY }} | |
| - name: generate the local OCI assets | |
| run: | | |
| cd training | |
| make -j vllm | |
| make -j deepspeed | |
| make -j instruct-nvidia | |
| - name: Build Image | |
| id: build_image | |
| run: make bootc DRIVER_VERSION=${{ matrix.driver_version }} ARCH=${{ matrix.arch }} SSH_PUBKEY=~/.ssh/id_rsa.pub | |
| working-directory: ${{ matrix.context }} | |
| - name: Push image | |
| if: github.event_name == 'push' && github.ref == 'refs/heads/main' | |
| uses: redhat-actions/[email protected] | |
| with: | |
| image: ${{ matrix.image_name }} | |
| tags: latest | |
| registry: ${{ env.REGISTRY }}/${{ env.REGISTRY_ORG }} | |
| - name: Publish Job Results to Slack | |
| id: slack | |
| if: always() | |
| uses: slackapi/[email protected] | |
| with: | |
| payload: | | |
| { | |
| "text": "${{ github.workflow }} workflow status: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}" | |
| } | |
| env: | |
| SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }} | |
| bootc-images: | |
| strategy: | |
| matrix: | |
| include: | |
| - image_name: intel-bootc | |
| context: training/intel-bootc | |
| arch: amd64 | |
| gpu: intel | |
| - image_name: amd-bootc | |
| context: training/amd-bootc | |
| arch: amd64 | |
| gpu: amd | |
| if: "!contains(github.event.pull_request.labels.*.name, 'hold-tests')" | |
| runs-on: ${{ needs.start-runner.outputs.label }} | |
| needs: [ start-runner, nvidia-bootc-builder-image ] | |
| continue-on-error: true | |
| steps: | |
| - uses: actions/[email protected] | |
| - name: mkdir root/.docker directory | |
| run: | | |
| mkdir -p ~/.docker | |
| - name: Login to Container Registry | |
| run: podman login -u ${{ secrets.REGISTRY_USER }} -p ${{ secrets.REGISTRY_PASSWORD }} ${{ env.REGISTRY }} | |
| - name: generate the local OCI assets | |
| run: | | |
| cd training | |
| make -j vllm | |
| make -j deepspeed | |
| make -j instruct-${{ matrix.gpu}} | |
| - name: Build Image | |
| id: build_image | |
| run: make bootc ARCH=${{ matrix.arch }} INSTRUCTLAB_IMAGE=${{env.REGISTRY}}/${{env.REGISTRY_ORG}}/instruct-${{ matrix.gpu }}:latest | |
| working-directory: ${{ matrix.context }} | |
| - name: Push image | |
| if: github.event_name == 'push' && github.ref == 'refs/heads/main' | |
| uses: redhat-actions/[email protected] | |
| with: | |
| image: ${{ matrix.image_name }} | |
| tags: latest | |
| registry: ${{ env.REGISTRY }}/${{ env.REGISTRY_ORG }} | |
| - name: Publish Job Results to Slack | |
| id: slack | |
| if: always() | |
| uses: slackapi/[email protected] | |
| with: | |
| payload: | | |
| { | |
| "text": "${{ github.workflow }} workflow status: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}" | |
| } | |
| env: | |
| SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }} | |
| stop-runner: | |
| name: Stop self-hosted EC2 runner | |
| needs: | |
| - start-runner | |
| - bootc-images | |
| - nvidia-bootc-image | |
| runs-on: ubuntu-latest | |
| if: ${{ always() }} # required to stop the runner even if the error happened in the previous jobs | |
| steps: | |
| - name: Configure AWS credentials | |
| uses: aws-actions/configure-aws-credentials@v1 | |
| with: | |
| aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }} | |
| aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }} | |
| aws-region: ${{ secrets.AWS_REGION }} | |
| - name: Stop EC2 runner | |
| uses: machulav/ec2-github-runner@v2 | |
| with: | |
| mode: stop | |
| github-token: ${{ secrets.GH_PERSONAL_ACCESS_TOKEN }} | |
| label: ${{ needs.start-runner.outputs.label }} | |
| ec2-instance-id: ${{ needs.start-runner.outputs.ec2-instance-id }} |