Build and push docker nightly #2259
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: Build and push docker nightly | |
| on: | |
| workflow_dispatch: | |
| inputs: | |
| mode: | |
| description: 'release/nightly/temp, default is nightly' | |
| required: true | |
| default: 'nightly' | |
| arch: | |
| description: 'which images to build [ cpu, cpu-full, pytorch-inf2, pytorch-gpu, tensorrt-llm, lmi, aarch64]' | |
| type: string | |
| required: false | |
| default: '["cpu", "cpu-full", "pytorch-inf2", "pytorch-gpu", "tensorrt-llm", "lmi", "aarch64"]' | |
| workflow_call: | |
| inputs: | |
| mode: | |
| description: 'release/nightly/temp, default is nightly' | |
| type: string | |
| required: true | |
| default: 'nightly' | |
| arch: | |
| description: 'which images to build [ cpu, cpu-full, pytorch-inf2, pytorch-gpu, tensorrt-llm, lmi, aarch64]' | |
| type: string | |
| required: false | |
| default: '["cpu", "cpu-full", "pytorch-inf2", "pytorch-gpu", "tensorrt-llm", "lmi", "aarch64"]' | |
| outputs: | |
| djl_version: | |
| description: "djl version" | |
| value: ${{ jobs.nightly-build.outputs.djl_version }} | |
| permissions: | |
| id-token: write | |
| contents: read | |
| env: | |
| AWS_ECR_REPO: "185921645874.dkr.ecr.us-east-1.amazonaws.com/djl-ci-temp" | |
| DOCKER_HUB_REPO: "deepjavalibrary/djl-serving" | |
| jobs: | |
| create-runners: | |
| runs-on: [ self-hosted, scheduler ] | |
| strategy: | |
| matrix: | |
| arch: ${{ startsWith(inputs.arch, '[') && fromJson(inputs.arch) || fromJson(format('[{0}]', inputs.arch)) }} | |
| steps: | |
| - name: Create new instance | |
| id: create_cpu | |
| run: | | |
| cd /home/ubuntu/djl_benchmark_script/scripts | |
| token=$( curl -X POST -H "Authorization: token ${{ secrets.ACTION_RUNNER_PERSONAL_TOKEN }}" \ | |
| https://api.github.com/repos/deepjavalibrary/djl-serving/actions/runners/registration-token \ | |
| --fail \ | |
| | jq '.token' | tr -d '"' ) | |
| instance_type=action_cpu | |
| if [ "${{matrix.arch}}" == "aarch64" ]; then | |
| instance_type=action_graviton | |
| fi | |
| ./start_instance.sh $instance_type $token djl-serving | |
| instance_id=`grep "^instance_id=" $GITHUB_OUTPUT | cut -d'=' -f2` | |
| echo "instance_id_${{matrix.arch}}=$instance_id" >>"$GITHUB_OUTPUT" | |
| outputs: | |
| instance_id_cpu: ${{ steps.create_cpu.outputs.instance_id_cpu }} | |
| instance_id_cpu-full: ${{ steps.create_cpu.outputs.instance_id_cpu-full }} | |
| instance_id_pytorch-inf2: ${{ steps.create_cpu.outputs.instance_id_pytorch-inf2 }} | |
| instance_id_pytorch-gpu: ${{ steps.create_cpu.outputs.instance_id_pytorch-gpu }} | |
| instance_id_tensorrt-llm: ${{ steps.create_cpu.outputs.instance_id_tensorrt-llm }} | |
| instance_id_lmi: ${{ steps.create_cpu.outputs.instance_id_lmi }} | |
| instance_id_aarch64: ${{ steps.create_cpu.outputs.instance_id_aarch64 }} | |
| nightly-build: | |
| needs: create-runners | |
| timeout-minutes: 120 | |
| strategy: | |
| fail-fast: false | |
| matrix: | |
| arch: ${{ startsWith(inputs.arch, '[') && fromJson(inputs.arch) || fromJson(format('[{0}]', inputs.arch)) }} | |
| runs-on: | |
| - self-hosted | |
| - ${{ matrix.arch != 'aarch64' && 'cpu' || 'aarch64' }} | |
| - RUN_ID-${{ github.run_id }} | |
| - RUN_NUMBER-${{ github.run_number }} | |
| - SHA-${{ github.sha }} | |
| outputs: | |
| djl_version: ${{ steps.get-versions.outputs.DJL_VERSION }} | |
| steps: | |
| - name: Clean disk space | |
| run: | | |
| sudo rm -rf \ | |
| /usr/share/dotnet /usr/local/lib/android /opt/ghc \ | |
| /usr/local/share/powershell /usr/share/swift /usr/local/.ghcup \ | |
| $AGENT_TOOLSDIRECTORY | |
| - uses: actions/checkout@v4 | |
| - name: Set up Python3 | |
| uses: actions/setup-python@v5 | |
| with: | |
| python-version: '3.10.x' | |
| - name: Install pip dependencies | |
| run: | | |
| python -m pip install --upgrade pip | |
| pip install awscli | |
| - name: Configure AWS Credentials | |
| uses: aws-actions/configure-aws-credentials@v4 | |
| with: | |
| role-to-assume: arn:aws:iam::185921645874:role/github-actions-djl-serving | |
| aws-region: us-east-1 | |
| - name: Set up JDK 17 | |
| uses: actions/setup-java@v4 | |
| with: | |
| distribution: 'corretto' | |
| java-version: 17 | |
| - uses: actions/cache@v4 | |
| with: | |
| path: ~/.gradle/caches | |
| key: ${{ runner.os }}-gradle-${{ hashFiles('**/*.gradle*') }} | |
| - name: Set env vars for container build and tag | |
| id: get-versions | |
| run: | | |
| DJL_VERSION=$(awk -F '=' '/djl / {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml) | |
| SERVING_VERSION=$(awk -F '=' '/serving / {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml) | |
| echo "DJL_VERSION=$DJL_VERSION" >> $GITHUB_ENV | |
| echo "DJL_VERSION=$DJL_VERSION" >> $GITHUB_OUTPUT | |
| echo "SERVING_VERSION=$SERVING_VERSION" >> $GITHUB_ENV | |
| mode=${{ inputs.mode }} | |
| if [[ "$mode" != "release" ]]; then | |
| export NIGHTLY="-nightly" | |
| echo "NIGHTLY=$NIGHTLY" >> $GITHUB_ENV | |
| ./gradlew --refresh-dependencies :serving:dockerDeb -Psnapshot | |
| export BUILD_VERSION_ARG_DJL="${DJL_VERSION}-SNAPSHOT" | |
| export BUILD_VERSION_ARG_SERVING="${SERVING_VERSION}-SNAPSHOT" | |
| else | |
| rm -rf serving/docker/distributions | |
| export BUILD_VERSION_ARG_DJL="${DJL_VERSION}-SNAPSHOT" | |
| export BUILD_VERSION_ARG_SERVING="${SERVING_VERSION}-SNAPSHOT" | |
| fi | |
| echo "BUILD_VERSION_ARG_DJL=$BUILD_VERSION_ARG_DJL" >> $GITHUB_ENV | |
| echo "BUILD_VERSION_ARG_SERVING=$BUILD_VERSION_ARG_SERVING" >> $GITHUB_ENV | |
| - name: Build docker image | |
| run: | | |
| export RELEASE_VERSION="${{ env.SERVING_VERSION }}-" | |
| cd serving/docker | |
| docker compose build --no-cache \ | |
| --build-arg djl_version=${{ env.BUILD_VERSION_ARG_DJL }} \ | |
| --build-arg djl_serving_version=${{ env.BUILD_VERSION_ARG_SERVING }} \ | |
| ${{ matrix.arch }} | |
| - name: Tag and push temp image to ECR repo | |
| working-directory: serving/docker | |
| run: | | |
| ECR_REGION=$(echo "${{ env.AWS_ECR_REPO }}" | awk -F. '{print $4}') | |
| aws ecr get-login-password --region $ECR_REGION | docker login --username AWS --password-stdin ${{env.AWS_ECR_REPO}} | |
| baseDockerTag="${{ env.AWS_ECR_REPO }}:${{ env.SERVING_VERSION}}-${{ matrix.arch }}${{ env.NIGHTLY }}" | |
| tempRunIdTag="${baseDockerTag}-${GITHUB_RUN_ID}" | |
| tempCommitTag="${baseDockerTag}-${GITHUB_SHA}" | |
| docker tag ${{ env.DOCKER_HUB_REPO }}:${{ env.SERVING_VERSION}}-${{ matrix.arch }}${{ env.NIGHTLY }} $baseDockerTag | |
| docker tag ${{ env.DOCKER_HUB_REPO }}:${{ env.SERVING_VERSION}}-${{ matrix.arch }}${{ env.NIGHTLY }} $tempRunIdTag | |
| docker tag ${{ env.DOCKER_HUB_REPO }}:${{ env.SERVING_VERSION}}-${{ matrix.arch }}${{ env.NIGHTLY }} $tempCommitTag | |
| time docker push --all-tags ${{ env.AWS_ECR_REPO }} | |
| stop-runners: | |
| if: always() | |
| runs-on: [ self-hosted, scheduler ] | |
| needs: [nightly-build, create-runners] | |
| env: | |
| runner_output: ${{ toJson(needs.create-runners.outputs) }} | |
| steps: | |
| - name: Stop all instances | |
| continue-on-error: true | |
| run: | | |
| cd /home/ubuntu/djl_benchmark_script/scripts | |
| for key in $(echo $runner_output | jq -r 'keys[]'); do | |
| instance_id=$(echo $runner_output | jq -r ".[\"$key\"]") | |
| echo "Key: $key, instance_id: $instance_id" | |
| ./stop_instance.sh $instance_id | |
| done |