• Hardware Platform (Jetson / GPU)
GCP VM with NVIDIA T4 GPU
• DeepStream Version
DeepStream 7.1
• JetPack Version (valid for Jetson only)
N/A
• TensorRT Version
10.3.0.26
• NVIDIA GPU Driver Version (valid for GPU only)
550.90.07
• Issue Type
Bug
• How to reproduce the issue?
I am trying to install FFmpeg dependencies inside the deepstream:7.1-gc-triton-develDocker container, but I am encountering issues where FFmpeg-related libraries seem to be missing or broken, despite installing them via apt.
My Dockerfile (attached below) includes ffmpeg and related dependencies, but when I try to use FFmpeg, it either fails due to missing shared libraries or incorrect package versions. I reinstalled specific FFmpeg-related libraries such as libmp3lame0, libxvidcore4, and libflac8, which fixes the issue, but I’m curious how I can get this working without reinstalling?
I suspect there might be compatibility issues with the base deepstream:7.1-gc-triton-devel image.
• Steps to Reproduce
- Build the Docker image using the provided Dockerfile.
- Run
ffmpeg -versioninside the container and observe missing dependencies.
FROM nvcr.io/nvidia/deepstream:7.1-gc-triton-devel
ENV SHELL /bin/bash
ENV CUDA_HOME=/usr/local/cuda
ENV LD_LIBRARY_PATH=$LD_LIBRARY_PATH:${CUDA_HOME}/lib64
ENV PATH=$PATH:$CUDA_HOME/bin
ENV DEBIAN_FRONTEND=noninteractive
SHELL ["/bin/bash", "-c"]
WORKDIR /root
COPY pip.conf /etc/pip.conf
# Install system dependencies
RUN apt-get update && \
apt-get install -y \
clang-format-11 \
libspdlog-dev \
ffmpeg \
libglib2.0-dev \
libjson-glib-dev \
uuid-dev \
libyaml-cpp-dev \
libprotobuf-dev \
protobuf-compiler \
mosquitto \
python3.10 \
python3.10-dev \
python3.10-venv \
python3.10-dbg \
gcovr \
lcov \
python3-gi \
python3-gi-cairo \
libgirepository1.0-dev \
libxml2 \
libxml2-dev \
libxml2-utils \
libmp3lame0 \
libxvidcore4 \
libflac8 \
&& apt clean && rm -rf /var/lib/apt/lists/*
# reinstall corrupted packages for ffmpeg
RUN apt-get update \
&& apt-get install --reinstall -y \
libmp3lame0 \
libxvidcore4 \
libflac8
# Update alternatives to use Python 3.10
RUN update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.10 1
# Install VPI
RUN apt install gnupg \
&& apt-key adv --fetch-key https://repo.download.nvidia.com/jetson/jetson-ota-public.asc \
&& apt install -y software-properties-common \
&& add-apt-repository -y 'deb https://repo.download.nvidia.com/jetson/x86_64/focal r34.1 main' \
&& apt update \
&& apt install -y libnvvpi2 vpi2-dev \
&& apt clean \
&& rm -rf /var/lib/apt/lists/*
# install interpipe
RUN apt-get update \
&& apt-get install -y \
git \
gtk-doc-tools \
autoconf \
automake \
libtool \
&& mkdir -p /root/tmp \
&& cd /root/tmp \
&& git clone https://github.com/RidgeRun/gst-interpipe/ \
&& cd gst-interpipe \
&& git checkout v1.1.8 \
&& ./autogen.sh --libdir /usr/lib/$(uname -m)-linux-gnu/ \
&& make \
&& make install \
&& rm -rf /root/tmp/ \
&& apt clean \
&& rm -rf /var/lib/apt/lists/*
# Install CMake
ARG CMAKE_VERSION=3.27
ARG CMAKE_BUILD=9
RUN mkdir /root/tmp \
&& cd /root/tmp \
&& apt update \
&& apt install -y wget build-essential \
&& wget https://cmake.org/files/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}.${CMAKE_BUILD}.tar.gz \
&& tar -xzvf cmake-${CMAKE_VERSION}.${CMAKE_BUILD}.tar.gz \
&& cd cmake-${CMAKE_VERSION}.${CMAKE_BUILD} \
&& ./bootstrap \
&& make -j$(nproc) \
&& make install \
&& rm -rf /root/tmp \
&& apt clean \
&& rm -rf /var/lib/apt/lists/*
# Install Python packages
RUN pip3 install --upgrade pip && pip3 install --upgrade numpy==1.24.2
# Build CUDA-enabled OpenCV from source
RUN cd /root \
&& git clone -b 4.10.0 https://github.com/opencv/opencv.git \
&& git clone -b 4.10.0 https://github.com/opencv/opencv_contrib.git \
&& cd opencv \
&& mkdir build \
&& cd build \
&& cmake .. \
-DCMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D WITH_CUDA=ON \
-D ENABLE_FAST_MATH=1 \
-D CUDA_FAST_MATH=1 \
-D WITH_CUBLAS=1 \
-D WITH_GSTREAMER=ON \
-D WITH_LIBV4L=ON \
-D INSTALL_PYTHON_EXAMPLES=ON \
-D WITH_FFMPEG=ON \
-D OPENCV_ENABLE_NONFREE=ON \
-D BUILD_opencv_python2=OFF \
-D BUILD_opencv_python3=ON \
-D OPENCV_EXTRA_MODULES_PATH=/root/opencv_contrib/modules \
-DOPENCV_PYTHON3_INSTALL_PATH=/usr/lib/python3/dist-packages \
-DPYTHON_EXECUTABLE=/usr/bin/python3 \
&& make -j$(nproc) \
&& make install \
&& rm -rf /root/opencv \
rm -rf /root/opencv_contrib
RUN python3 -c "import cv2; print(cv2.getBuildInformation())"
# Install Python testing tools
RUN python3 -m pip uninstall -y importlib-metadata && \
python3 -m pip install -U argcomplete \
flake8-blind-except \
flake8-builtins \
flake8-class-newline \
flake8-comprehensions \
flake8-deprecated \
flake8-docstrings \
flake8-import-order \
flake8-quotes \
pytest-repeat \
pytest-rerunfailures \
pytest
# DeepStream environment setup
RUN . /opt/nvidia/deepstream/deepstream-7.1/user_additional_install.sh \
&& . /opt/nvidia/deepstream/deepstream-7.1/install.sh \
&& cd /opt/nvidia/deepstream/deepstream-7.1 \
&& ldconfig
ENV LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/opt/nvidia/deepstream/deepstream-7.1/lib/
ENV GST_PLUGIN_PATH=/opt/nvidia/deepstream/deepstream-7.1/lib/gst-plugins
# Check CUDA and cuDNN versions
RUN nvcc --version && \
whereis cudnn.h && \
cat /usr/include/cudnn_version.h | grep CUDNN_MAJOR -A 2
# Check TensorRT version
RUN dpkg -l | grep nvinfer
COPY entrypoint.sh /entrypoint.sh
RUN chmod +x /entrypoint.sh
ENTRYPOINT ["/entrypoint.sh"]
WORKDIR /root
CMD ["bash"]