name: Distribute-stable-Test

on:
  workflow_call:
    inputs:
      docker_distribute_image:
        type: string
        required: true
      clone-can-skip:
        type: string
        required: false
        default: "false"

env:
  PR_ID: ${{ github.event.pull_request.number }}
  COMMIT_ID: ${{ github.event.pull_request.head.sha }}
  work_dir: /paddle
  PADDLE_ROOT: /paddle
  TASK: paddle-CI-${{ github.event.pull_request.number }}-distribute-test
  ci_scripts: /paddle/ci
  BRANCH: ${{ github.event.pull_request.base.ref }}
  CI_name: distribute
  no_proxy: bcebos.com,apiin.im.baidu.com,gitee.com,aliyun.com,.baidu.com,.tuna.tsinghua.edu.cn,paddlepaddle.org.cn
  docker_image: ${{ inputs.docker_distribute_image }}

defaults:
  run:
    shell: bash

jobs:
  check-bypass:
    name: Check bypass
    uses: ./.github/workflows/check-bypass.yml
    with:
      workflow-name: distribute-test
    secrets:
      github-token: ${{ secrets.GITHUB_TOKEN }}

  test:
    name: Test
    needs: [check-bypass]
    if: ${{ needs.check-bypass.outputs.can-skip != 'true' && inputs.clone-can-skip != 'true' }}
    runs-on:
      group: Distribute
    timeout-minutes: 120
    steps:
      - name: Check docker image and run container
        env:
          FLAGS_fraction_of_gpu_memory_to_use: 0.15
          CTEST_OUTPUT_ON_FAILURE: 1
          CTEST_PARALLEL_LEVEL: 4
          WITH_GPU: "ON"
          WITH_AVX: "ON"
          WITH_DISTRIBUTE: "ON"
          WITH_TESTING: "ON"
          WITH_COVERAGE: "OFF"
          CMAKE_BUILD_TYPE: Release
          PADDLE_FRACTION_GPU_MEMORY_TO_USE: 0.15
          PRECISION_TEST: "OFF"
          WITH_UNITY_BUILD: "ON"
          AGILE_COMPILE_BRANCH: ${{ github.event.pull_request.base.ref }}
          AGILE_REVISION: ${{ github.event.pull_request.head.sha }}
          WITH_INCREMENTAL_COVERAGE: "OFF"
          WITH_ONNXRUNTIME: "OFF"
          COVERALLS_UPLOAD: "ON"
          PADDLE_VERSION: 0.0.0
          GIT_PR_ID: ${{ github.event.pull_request.number }}
          PY_VERSION: "3.10"
          CUDA_ARCH_NAME: Auto
          WITH_CUDNN_FRONTEND: "ON"
          FLAGS_enable_cudnn_frontend: 1
          CACHE_DIR: /root/.cache/build
          CCACHE_DIR: /root/.ccache/gpubox
        run: |
          export CUDA_SO="$(\ls -d /usr/lib64/libcuda* | xargs -I{} echo '-v {}:{}') $(\ls -d /usr/lib64/libnvidia* | xargs -I{} echo '-v {}:{}')"
          export DEVICES="$(\ls -d /dev/nvidia* | xargs -I{} echo "-v {}:{}") $(\ls /dev/nvidia-caps/* | xargs -I{} echo "-v {}:{}")"
          export SMI="-v /usr/bin/nvidia-smi:/usr/bin/nvidia-smi"
          container_name=${TASK}-test-$(date +%Y%m%d-%H%M%S)
          echo "container_name=${container_name}" >> ${{ github.env }}
          docker run -d -t --name ${container_name} ${CUDA_SO} ${DEVICES} ${SMI} --runtime=nvidia --shm-size=32G \
          -v "/home/data/cfs:/home/data/cfs" \
          -v "/home/data/cfs/.cache/:/root/.cache" \
          -v "/home/data/cfs/.ccache:/root/.ccache" \
          -v "/ssd1/root:/root" \
          -v "/dev/shm:/dev/shm"  \
          -v ${{ github.workspace }}/../../..:${{ github.workspace }}/../../.. \
          -v ${{ github.workspace }}:/paddle \
          -e BRANCH \
          -e PR_ID \
          -e COMMIT_ID \
          -e work_dir \
          -e PADDLE_ROOT \
          -e ci_scripts \
          -e CI_name \
          -e FLAGS_fraction_of_gpu_memory_to_use \
          -e CTEST_OUTPUT_ON_FAILURE \
          -e CTEST_PARALLEL_LEVEL \
          -e WITH_GPU \
          -e WITH_AVX \
          -e WITH_DISTRIBUTE \
          -e WITH_TESTING \
          -e WITH_COVERAGE \
          -e CMAKE_BUILD_TYPE \
          -e PADDLE_FRACTION_GPU_MEMORY_TO_USE \
          -e PRECISION_TEST \
          -e WITH_UNITY_BUILD \
          -e AGILE_COMPILE_BRANCH \
          -e AGILE_REVISION \
          -e WITH_INCREMENTAL_COVERAGE \
          -e WITH_ONNXRUNTIME \
          -e COVERALLS_UPLOAD \
          -e PADDLE_VERSION \
          -e GIT_PR_ID \
          -e PY_VERSION \
          -e CUDA_ARCH_NAME \
          -e WITH_CUDNN_FRONTEND \
          -e FLAGS_enable_cudnn_frontend \
          -e CACHE_DIR \
          -e CCACHE_DIR \
          -e no_proxy \
          -w /paddle --network host ${docker_image}

      - name: Download paddle.tar.gz and merge target branch
        timeout-minutes: 30
        run: |
          docker exec -t ${{ env.container_name }} /bin/bash -c '
          rm -rf * .[^.]*
          echo "Downloading Paddle.tar.gz"
          wget -q --tries=5 --no-proxy https://paddle-github-action.bj.bcebos.com/PR/gpups/${{ env.PR_ID }}/${{ env.COMMIT_ID }}/Paddle.tar.gz --no-check-certificate
          echo "Extracting Paddle.tar.gz"
          tar --use-compress-program="pzstd" -xf Paddle.tar.gz --strip-components=1
          rm Paddle.tar.gz
          git checkout test
          '

      - name: Test
        run: |
          docker exec -t ${{ env.container_name }} /bin/bash -c '
          source ${{ github.workspace }}/../../../proxy
          bash ${ci_scripts}/distribute_test.sh
          '

      - name: Upload and display logs
        if: always()
        env:
          home_path: ${{ github.workspace }}/..
          bos_file: ${{ github.workspace }}/../bos_retry/BosClient.py
        run: |
          docker exec -t ${{ env.container_name }} /bin/bash -c '
          export AK=paddle
          export SK=paddle
          if [ ! -f "${{ env.bos_file }}" ]; then
            wget -q --no-proxy -O ${{ env.home_path }}/bos_retry.tar.gz https://xly-devops.bj.bcebos.com/home/bos_retry.tar.gz --no-check-certificate
            mkdir ${{ env.home_path }}/bos_retry
            tar xf ${{ env.home_path }}/bos_retry.tar.gz -C ${{ env.home_path }}/bos_retry
          fi
          cd /case_logs
          for FILE in /case_logs/*; do
            file=$(basename "$FILE")
            python ${{ env.bos_file }} $file paddle-github-action/PR/Distribute-Stable/${PR_ID}/${COMMIT_ID}/logs
            echo "$file: https://paddle-github-action.bj.bcebos.com/PR/Distribute-Stable/${PR_ID}/${COMMIT_ID}/logs/$file"
          done
          '

      - name: Terminate and delete the container
        if: always()
        run: |
          set +e
          docker exec -t ${{ env.container_name }} /bin/bash -c 'rm -rf * .[^.]*'
          docker rm -f ${{ env.container_name }}