如何从 Dockerfile 运行环境初始化 shell 脚本

Question

I am trying to build an API wrapped in a docker image that serves Openvino model. How do I run the "setupvars.sh" from Dockerfile itself so that my application can access it?

I have tried running the script using RUN. For ex: RUN /bin/bash setupvars.sh or RUN ./setupvars.sh . However, none of them work and I get ModelNotFoundError: no module named openvino

RUN $INSTALL_DIR/install_dependencies/install_openvino_dependencies.sh
RUN cd /opt/intel/openvino/deployment_tools/model_optimizer/install_prerequisites && sudo ./install_prerequisites_tf.sh
COPY . /app
WORKDIR /app
RUN apt autoremove -y && \
    rm -rf /openvino /var/lib/apt/lists/*
RUN /bin/bash -c "source $INSTALL_DIR/bin/setupvars.sh"
RUN echo "source $INSTALL_DIR/bin/setupvars.sh" >> /root/.bashrc
CMD ["/bin/bash"]
RUN python3 -m pip install opencv-python
RUN python3 test.py

I want OpenVino accessible to my gunicorn application that will serve the model in a docker image

Answer 1

Next commands works for me.

ARG OPENVINO_DIR=/opt/intel/computer_vision_sdk


# Unzip the OpenVINO installer
RUN cd ${APP_DIR} && tar -xvzf l_openvino_toolkit*

# installing OpenVINO dependencies
RUN cd ${APP_DIR}/l_openvino_toolkit* && \
    ./install_cv_sdk_dependencies.sh

# installing OpenVINO itself
RUN cd ${APP_DIR}/l_openvino_toolkit* && \
    sed -i 's/decline/accept/g' silent.cfg && \
    ./install.sh --silent silent.cfg

# Setup the OpenVINO environment
RUN /bin/bash -c "source ${OPENVINO_DIR}/bin/setupvars.sh"

Answer 2

You need to re-run it every time you start the container, because those variables are only for the session.

Option 1: Run your application something like this:

CMD /bin/bash -c "source /opt/intel/openvino/bin/setupvars.sh && python test.py"

Option 2 (not tested): Add the source command to your .bashrc so it will be run every time on startup

# Assuming running as root
RUN echo "/bin/bash -c 'source /opt/intel/openvino/bin/setupvars.sh'" >> ~root/.bashrc
CMD python test.py

For the rest of the Dockerfile, there is a guide here (also not tested, and it doesn't cover the above): https://docs.openvinotoolkit.org/latest/_docs_install_guides_installing_openvino_docker_linux.html

Answer 3

As mentioned in the two previous answers, the setupvars.sh script sets the environment variables required by OpenVINO.

But rather than running this every time, you can add the variables to your Dockerfile. While writing your Dockerfile run:

CMD /bin/bash -c "source /opt/intel/openvino/bin/setupvars.sh && printenv"

This will give you the values that the environment variables are set to. You might also want to run printenv without setting the OpenVINO variables:

CMD /bin/bash printenv

Comparing the two outputs will let you figure out exactly what the setupvars.sh script is setting.

Once you know the values set by the script, you can set these as part of the Dockerfile using the ENV instruction. I wouldn't copy this because it's likely to be specific to your setup, but in my case, this ended up looking like:

ENV PATH=/opt/intel/openvino/deployment_tools/model_optimizer:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
ENV LD_LIBRARY_PATH=/opt/intel/openvino/opencv/lib:/opt/intel/openvino/deployment_tools/ngraph/lib:/opt/intel/openvino/deployment_tools/inference_engine/external/tbb/lib::/opt/intel/openvino/deployment_tools/inference_engine/external/hddl/lib:/opt/intel/openvino/deployment_tools/inference_engine/external/omp/lib:/opt/intel/openvino/deployment_tools/inference_engine/external/gna/lib:/opt/intel/openvino/deployment_tools/inference_engine/external/mkltiny_lnx/lib:/opt/intel/openvino/deployment_tools/inference_engine/lib/intel64
ENV INTEL_CVSDK_DIR=/opt/intel/openvino
ENV OpenCV_DIR=/opt/intel/openvino/opencv/cmake
ENV TBB_DIR=/opt/intel/openvino/deployment_tools/inference_engine/external/tbb/cmake
# The next one will be whatever your working directory is
ENV PWD=/workspace
ENV InferenceEngine_DIR=/opt/intel/openvino/deployment_tools/inference_engine/share
ENV ngraph_DIR=/opt/intel/openvino/deployment_tools/ngraph/cmake
ENV SHLVL=1
ENV PYTHONPATH=/opt/intel/openvino/python/python3.6:/opt/intel/openvino/python/python3:/opt/intel/openvino/deployment_tools/tools/post_training_optimization_toolkit:/opt/intel/openvino/deployment_tools/open_model_zoo/tools/accuracy_checker:/opt/intel/openvino/deployment_tools/model_optimizer
ENV HDDL_INSTALL_DIR=/opt/intel/openvino/deployment_tools/inference_engine/external/hddl
ENV _=/usr/bin/printenv