_Auto-Parallel.yml

name: Auto-Parallel

on:
  workflow_call:
    inputs:
      docker_build_image:
        type: string
        required: true
      can-skip:
        type: string
        required: false

env:
  PR_ID: ${{ github.event.pull_request.number }}
  COMMIT_ID: ${{ github.event.pull_request.head.sha }}
  work_dir: /workspace/Paddle
  PADDLE_ROOT: /workspace/Paddle
  TASK: paddle-CI-${{ github.event.pull_request.number }}-Auto-Parallel
  ci_scripts: /workspace/Paddle/ci
  BRANCH: ${{ github.event.pull_request.base.ref }}
  CI_name: auto-parallel
  no_proxy: "localhost,bj.bcebos.com,su.bcebos.com,bcebos.com,apiin.im.baidu.com,gitee.com,aliyun.com,.baidu.com,.tuna.tsinghua.edu.cn"

defaults:
  run:
    shell: bash

jobs:
  check-bypass:
    name: Check bypass
    if: ${{ inputs.can-skip != 'true' }}
    uses: ./.github/workflows/check-bypass.yml
    with:
      workflow-name: "auto-parallel"
    secrets:
      github-token: ${{ secrets.GITHUB_TOKEN }}

  parallel-test:
    name: Parallel test
    needs: check-bypass
    if: ${{ needs.check-bypass.outputs.can-skip != 'true' }}
    runs-on:
      group: Auto-Parallel
    timeout-minutes: 120
    steps:
      - name: Check docker image and run container
        env:
          FLAGS_dynamic_static_unified_comm: "True"
          python_version: "3.10"
          paddle_whl: https://paddle-github-action.bj.bcebos.com/PR/build/${{ github.event.pull_request.number }}/${{ github.event.pull_request.head.sha }}/paddlepaddle_gpu-0.0.0-cp310-cp310-linux_x86_64.whl
        run: |
          container_name=${TASK}-${core_index}-$(date +%Y%m%d-%H%M%S)
          echo "container_name=${container_name}" >> ${{ github.env }}
          docker_image=${{ inputs.docker_build_image }}
          nvidia-docker run -d -t --name ${container_name} --net=host -v /dev/shm:/dev/shm --shm-size=32G \
            -v ${{ github.workspace }}/../../..:${{ github.workspace }}/../../.. \
            -v ${{ github.workspace }}:/workspace \
            -v /home/FleetX_CI:/gpt_data \
            -v /home/Llm_gpt_CI:/llm_gpt_data \
            -v /home/Llama_CI:/llama_data \
            -v /home/.cache/pip:/home/.cache/pip \
            -e BRANCH \
            -e PR_ID \
            -e COMMIT_ID \
            -e work_dir \
            -e PADDLE_ROOT \
            -e ci_scripts \
            -e no_proxy \
            -e CI_name \
            -e paddle_whl \
            -e FLAGS_dynamic_static_unified_comm \
            -e python_version \
            -w /workspace --runtime=nvidia ${docker_image}

      - name: Download Paddle
        env:
          work_dir: ${{ github.workspace }}
        timeout-minutes: 30
        run: |
          docker exec -t ${{ env.container_name }} /bin/bash -c '
          rm -rf * .[^.]*
          echo "Downloading build.tar.gz"
          wget -q --tries=5 --no-proxy https://paddle-github-action.bj.bcebos.com/PR/build/${{ github.event.pull_request.number }}/${{ github.event.pull_request.head.sha }}/build.tar.gz --no-check-certificate
          echo "Extracting build.tar.gz"
          tar --use-compress-program="pzstd" -xpf build.tar.gz
          mv paddle Paddle
          rm -f build.tar.gz
          '

      - name: Download PaddleTest
        env:
          work_dir: ${{ github.workspace }}
        timeout-minutes: 30
        run: |
          docker exec -t ${{ env.container_name }} /bin/bash -c '
          echo "Download and extract PaddleNLP.tar.gz"
          wget -q --tries=5 --no-proxy  https://paddle-qa.bj.bcebos.com/CodeSync/develop/PaddleNLP.tar --no-check-certificate
          tar xf PaddleNLP.tar && rm -rf PaddleNLP.tar
          source ${{ github.workspace }}/../../../proxy
          cd PaddleNLP
          git config --global user.name "PaddleCI"
          git config --global user.email "paddle_ci@example.com"
          git pull
          git checkout 6ac04028757dfbcc089916997493611f62de81b2
          git switch -c 6ac04028757dfbcc089916997493611f62de81b2
          git cherry-pick bc08aeec91d2c992c3d8d39755bea7c6213b0e82
          git cherry-pick 7ab35ce94eca977bcf3b44bfb42deb0e0b5ef158
          git cherry-pick 2ac85997c2dafe3d67a4aac01d553ad76d6024bf
          git submodule update --init --recursive --force
          '

      - name: Test
        env:
          work_dir: ${{ github.workspace }}
        run: |
          docker exec -t ${{ env.container_name }} /bin/bash -c '
          ldconfig
          set -e
          pip config set global.cache-dir "/home/.cache/pip"
          ln -sf $(which python${python_version}) /usr/bin/python
          python -c "import sys; print(sys.version_info[:])"
          source ${{ github.workspace }}/../../../proxy
          python -m pip install bce-python-sdk==0.8.74
          timeout 80m bash Paddle/ci/auto_parallel/ci_auto_parallel.sh ${paddle_whl}
          '

      - 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 ${{ env.home_path }}/bos_retry
          fi
          cd /workspace/case_logs
          for FILE in /workspace/case_logs/*; do
            file=$(basename "$FILE")
            python ${{ env.bos_file }} $file paddle-github-action/PR/Auto-Parallel/${PR_ID}/${COMMIT_ID}/logs
            echo "$file: https://paddle-github-action.bj.bcebos.com/PR/Auto-Parallel/${PR_ID}/${COMMIT_ID}/logs/$file"
          done
          '

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