Skip to content

Latest commit

 

History

History
312 lines (262 loc) · 13.1 KB

bootstrap.md

File metadata and controls

312 lines (262 loc) · 13.1 KB

Set Up OpenVINO DL Workbench

  • If you are a developer and want to configure your system for DL Workbench development, refer to Configure for Development section.

Configure for Development

  1. Install prerequisites:

    • git
    • python3.6
  2. Clone the DL Workbench repository:

    git clone https://github.com/openvinotoolkit/workbench.git
  3. Run the script to bootstrap os from the root folder of the DL Workbench repository:

    bash automation/bootstrap/bootstrap_os.sh
  4. (Optional) Set up the workspace in PyCharm

    1. Open the workbench project in PyCharm
    2. Open the automation/bootstrap/bootstrap_pycharm.py
    3. Right click inside the script and choose Run 'bootstrap_pycharm.py'
    4. Restart PyCharm

Steps if you want to set up the environment manually

  1. Install prerequisites:

    • git
    • python3.6
  2. Download the OpenVINO package and install it (follow the instruction) The full link to the current package you can find in openvino_version.yml in the private CI repository.

  3. Open the terminal and clone the DL Workbench repository:

    git clone https://github.com/openvinotoolkit/workbench.git
    cd workbench
    export OPENVINO_WORKBENCH_ROOT=$(pwd)
  4. Initialize and clone submodules of the repository:

    git submodule update --init --recursive
  5. Install the virtualenv python package:

    python3 -m pip install virtualenv
  6. Create a python virtual environment:

    python3 -m virtualenv ${OPENVINO_WORKBENCH_ROOT}/.venv
  7. Initialize the OpenVINO environment:

    source ~/intel/openvino_2022.4.653/setupvars.sh
  8. Install OpenVINO console tools (Accuracy Checker, POT, Benchmark App, Model Optimizer) from wheels:

    source ${OPENVINO_WORKBENCH_ROOT}/.venv/bin/activate
    python -m pip install ${INTEL_OPENVINO_DIR}/tools/wheels/openvino-2022.1.0-5583-cp36-cp36m-manylinux2014_x86_64.whl
    python -m pip install ${INTEL_OPENVINO_DIR}/tools/wheels/openvino_dev-2022.1.0-5583-py3-none-any.whl
    deactivate

    Make sure to substitute cp36-cp36m in openvino runtime wheel with your Python version.

  9. Install requirements of DL Workbench:

    source ${OPENVINO_WORKBENCH_ROOT}/.venv/bin/activate
    python -m pip install -r ${OPENVINO_WORKBENCH_ROOT}/requirements/requirements.txt
    python -m pip install -r ${OPENVINO_WORKBENCH_ROOT}/requirements/requirements_dev.txt
    python -m pip install -r ${OPENVINO_WORKBENCH_ROOT}/requirements/requirements_jupyter.txt
    python -m pip install -r ${OPENVINO_WORKBENCH_ROOT}/client/automation/requirements_dev.txt
    python -m pip install -r ${OPENVINO_WORKBENCH_ROOT}/model_analyzer/requirements.txt
    deactivate
  10. Create a environment for models conversion:

    python3 -m virtualenv ${OPENVINO_WORKBENCH_ROOT}/.unified_venv
    source ${OPENVINO_WORKBENCH_ROOT}/.unified_venv/bin/activate
    python3 -m pip install -r ${INTEL_OPENVINO_DIR}/tools/model_optimizer/requirements_tf2.txt
    python3 -m pip install -r ${INTEL_OPENVINO_DIR}/extras/open_model_zoo/tools/model_tools/requirements.in
    python3 -m pip install -r ${INTEL_OPENVINO_DIR}/extras/open_model_zoo/tools/model_tools/requirements-pytorch.in
    python3 -m pip install -r ${INTEL_OPENVINO_DIR}/extras/open_model_zoo/tools/model_tools/requirements-caffe2.in
    deactivate
  11. Install Node.js and install packages for the client:

    wget -qO- https://raw.githubusercontent.com/nvm-sh/nvm/v0.38.0/install.sh | bash
    export NVM_DIR="$([ -z "${XDG_CONFIG_HOME-}" ] && printf %s "${HOME}/.nvm" || printf %s "${XDG_CONFIG_HOME}/nvm")"
    [ -s "$NVM_DIR/nvm.sh" ] && \. "$NVM_DIR/nvm.sh" # This loads nvm
    
    pushd ${OPENVINO_WORKBENCH_ROOT}/client
        nvm use --delete-prefix v14 --silent
        npm ci
        npm run init-netron
    popd
  12. Install Git hooks:

    source ${OPENVINO_WORKBENCH_ROOT}/.venv/bin/activate
    pre-commit install
    pre-commit install --hook-type pre-push
    deactivate
  13. Install Docker

    1. Update the apt package index and install packages to allow apt to use a repository over HTTPS
    sudo -E apt-get update
    
    sudo -E apt-get install \
        apt-transport-https \
        ca-certificates \
        curl \
        gnupg \
        lsb-release
    1. Add Docker’s official GPG key
    curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg
    1. Use the following command to set up the stable repository
    echo \
    "deb [arch=amd64 signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/ubuntu \
    $(lsb_release -cs) stable" | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
    1. Install Docker engine
    sudo -E apt-get update
    
    sudo -E apt-get install docker-ce docker-ce-cli containerd.io
    1. Verify that Docker Engine is installed correctly by running the hello-world image
    sudo docker run hello-world
    1. To manage Docker as a non-root user create the docker group
    sudo groupadd docker
    1. Add your user to the docker group
    sudo usermod -aG docker $USER
    1. Log out and log back in so that your group membership is re-evaluated
  14. Install Docker Compose

    1. Run this command to download the current stable release of Docker Compose
    sudo -E curl -L "https://github.com/docker/compose/releases/download/1.29.2/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose
    1. Apply executable permissions to the binary
    sudo chmod +x /usr/local/bin/docker-compose
    1. Install command completion for the bash and zsh shell
    sudo -E curl \
    -L https://raw.githubusercontent.com/docker/compose/1.29.2/contrib/completion/bash/docker-compose \
    -o /etc/bash_completion.d/docker-compose
    1. Reload your terminal. You can close and then open a new terminal, or reload your setting with source ~/.bashrc command in your current terminal.
  15. Install PyCharm

  16. Setup PyCharm environment

    1. Open 'Edit Configurations' window and add new configuration(press plus button)

    2. Select Docker->Docker-compose from the list

    3. Setup settings as on the imageDocker configuration

      Property Value
      Compose files ./docker/docker-compose.local.yml;
      Environment variables LOCAL_ADDR=172.17.0.1
      Services nginx, postgres, rabbitmq,
    4. Add another configuration, select 'npm' from the list

    5. Setup settings as on the imagenpm configuration

      Property Value
      package.json OPENVINO_WORKBENCH_ROOT/client/package.json
      Command start
      Arguments --scripts-prepend-node-path
    6. Create Python configuration for Gunicorn

    7. Setup settings as on the imageGunicorn configuration

      Property Value
      Script path OPENVINO_WORKBENCH_ROOT/.venv/bin/gunicorn
      Parameters --worker-class eventlet -w 1 -b 0.0.0.0:5676 workbench:APP --log-level DEBUG --no-sendfile --timeout 500
      Working configuration OPENVINO_WORKBENCH_ROOT/.venv/bin
    8. Add these user environment variables for the configuration:

      Property Value
      PYTHONUNBUFFERED 1
      PYTHONPATH
      LD_LIBRARY_PATH
      INTEL_OPENVINO_DIR
      VENV_TF2_PYTHON OPENVINO_WORKBENCH_ROOT/.unified_venv
      SERVER_MODE development
      PUBLIC_PORT 4200
    9. In terminal activate OpenVINO environment

      source ~/intel/openvino_2022/setupvars.sh
    10. Copy output of commands as values to the same named environment variables in PyCharm

      echo $PYTHONPATH
      echo $LD_LIBRARY_PATH
      echo $INTEL_OPENVINO_DIR
    11. Copy this configuration(this one will be for Celery)

    12. Setup settings as on the imageCelery configuration

      Property Value
      Script path OPENVINO_WORKBENCH_ROOT/.venv/bin/celery
      Parameters -A wb.main.tasks.task worker --loglevel=DEBUG
      Working configuration OPENVINO_WORKBENCH_ROOT/
    13. Set to user environment variable VENV_TF2_PYTHON this value OPENVINO_WORKBENCH_ROOT/.unified_venv/bin/python

    14. Copy this configuration(this one will be for database upgrading)

    15. Setup settings as on the imageDB upgrade configuration

      Property Value
      Script path OPENVINO_WORKBENCH_ROOT/.venv/bin/flask
      Parameters db upgrade
      Working configuration OPENVINO_WORKBENCH_ROOT/
    16. Set to user environment variable FLASK_APP this value OPENVINO_WORKBENCH_ROOT/migrations/migration:APP

    17. Remove PUBLIC_PORT user environment variable

  17. Run all of these configurations

  18. Check http://127.0.0.1:4200 in your browser, you should see DL Workbench app

How to download bundles for setuping remote targets and creating deployment package

  1. Copy the value of the field openvino_package from the openvino_version.yml from the private CI repository file
  2. Run the following command to download bundles from root of the repository:
python ./wb/main/utils/bundle_creator/bundle_downloader.py \
    --link ${PACKAGE_LINK}/deployment_archives \
    -os ubuntu18 ubuntu20 \
    --output-path ./bundles \
    --targets cpu gpu vpu hddl opencv python3.6 python3.7 python3.8 

Setup VM for working with remote targets

  1. Create new virtual machine with Ubuntu 18.04 OS
  2. Configure SSH server in guest Ubuntu OS
    1. Install openssh-server:
      sudo apt update 
      sudo apt install openssh-server
    2. Check SSH service is running:
      sudo systemctl status ssh 
    3. Update firewall rules:
      sudo ufw allow ssh
    4. Configure network in virtual machine as bridged network. Your VM's IP address should be in the same network as your host machine (e.g 192.168.*.*), not in the virtual subnet.
    5. Test SSH connection:
      ssh -p $user $user@$machine_ip_address 
      Example:
  3. Generate and setup SSH keys:
    1. Generate SSH public and private key pair:
      ssh-keygen -t rsa -f ~/destination_file_path
    2. Copy public key to virtual machine:
      ssh-copy-id -i ~/destination_file_path -p $user $user@$machine_ip_address
    3. Test SSH connection with private key:
      ssh -i ~/destination_file_path -p $user $user@$machine_ip_address
  4. Make sure that your VM satisfy requirement:
    • Installed python (v3) and pip (as a python module:python3 -m pip --version)
    • Configured Internet connection (ping -c 1 google.com)