Docker镜像解析出Dockerfile
目录
一、针对问题
二、解析镜像的Dockerfile
2.1 下载github的解析工具
2.2 解析出Dockerfile文件
2.3 针对官方镜像解析出的Dockfile进行分析
2.4 通过平台的创建镜像功能制作镜像（需要联网）
一、针对问题
在制作适配A100等新的GPU卡镜像时，由于一些pytorch、tensorflow等框架可能还不支持该GPU卡，自己制作的镜像，可能无法使用A100 GPU卡。在使用NVIDIA官方NGC提供的框架镜像时，官方在镜像里面已经配置jupyter等组件，配置方式和我们不一样，导致镜像导入AI平台时，无法使用jupyter。即使我们重新配置，也无法生效，因为配置方式不一样， 为了提高效率和节省时间，针对类似现象，提供一种解决该问题的参考方式。
以NGC官方tensorflow镜像为例：
docker pull nvcr.io/nvidia/tensorflow:20.09-tf2-py3镜像官方说明链接：
TensorFlow Release Notes :: NVIDIA Deep Learning Frameworks Documentation
二、解析镜像的Dockerfile
通过Dockerfile制作的镜像，解析出Dockerfile, 该方式不适用于所有镜像，只针对Dockerfile制作的镜像，而且解析出的Dockerfile仅供参考，和实际的原Dockerfile文件是有差别的。不过可以用来了解，Dockerfile安装了哪些组件，以及对镜像进行了哪些操作和配置，然后进行有针对性的修改，解决适配AIStation平台的相关问题。
2.1 下载github的解析工具
$ cd /home/wjy
$ git clone https://github.com/lukapeschke/dockerfile-from-image.git
$ cd dockerfile-from-image
$ docker build --rm -t lukapeschke/dockerfile-from-image .
查看buid完成的镜像
2.2 解析出Dockerfile文件
注意：
docker run --rm -v '/var/run/docker.sock:/var/run/docker.sock' lukapeschke/dockerfile-from-image 示例：
docker run --rm -v '/var/run/docker.sock:/var/run/docker.sock' lukapeschke/dockerfile-from-image > /home/wjy/cuda11-dockerfile/Dockerfile查看解析出的Dockerfile:
cat /home/wjy/cuda11-dockerfile/DockerfileFROM nvcr.io/nvidia/tensorflow:20.09-tf2-py3
ADD file:5c125b7f411566e9daa738d8cb851098f36197810f06488c2609074296f294b2 in /
RUN /bin/sh -c [ -z "$(apt-get indextargets)" ]
RUN /bin/sh -c set -xe \
&& echo '#!/bin/sh' > /usr/sbin/policy-rc.d \
&& echo 'exit 101' >> /usr/sbin/policy-rc.d \
&& chmod +x /usr/sbin/policy-rc.d \
&& dpkg-divert --local --rename --add /sbin/initctl \
&& cp -a /usr/sbin/policy-rc.d /sbin/initctl \
&& sed -i 's/^exit.*/exit 0/' /sbin/initctl \
&& echo 'force-unsafe-io' > /etc/dpkg/dpkg.cfg.d/docker-apt-speedup \
&& echo 'DPkg::Post-Invoke { "rm -f /var/cache/apt/archives/*.deb /var/cache/apt/archives/partial/*.deb /var/cache/apt/*.bin || true"; };' > /etc/apt/apt.conf.d/docker-clean \
&& echo 'APT::Update::Post-Invoke { "rm -f /var/cache/apt/archives/*.deb /var/cache/apt/archives/partial/*.deb /var/cache/apt/*.bin || true"; };' >> /etc/apt/apt.conf.d/docker-clean \
&& echo 'Dir::Cache::pkgcache ""; Dir::Cache::srcpkgcache "";' >> /etc/apt/apt.conf.d/docker-clean \
&& echo 'Acquire::Languages "none";' > /etc/apt/apt.conf.d/docker-no-languages \
&& echo 'Acquire::GzipIndexes "true"; Acquire::CompressionTypes::Order:: "gz";' > /etc/apt/apt.conf.d/docker-gzip-indexes \
&& echo 'Apt::AutoRemove::SuggestsImportant "false";' > /etc/apt/apt.conf.d/docker-autoremove-suggests
RUN /bin/sh -c mkdir -p /run/systemd \
&& echo 'docker' > /run/systemd/container
CMD ["/bin/bash"]
RUN RUN /bin/sh -c export DEBIAN_FRONTEND=noninteractive \
&& apt-get update \
&& apt-get install -y --no-install-recommends apt-utils build-essential ca-certificates curl patch wget jq gnupg libtcmalloc-minimal4 \
&& curl -fsSL https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub | apt-key add - \
&& echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 /" > /etc/apt/sources.list.d/cuda.list \
&& rm -rf /var/lib/apt/lists/* # buildkit
RUN ARG CUDA_VERSION
RUN ARG CUDA_DRIVER_VERSION
RUN ENV CUDA_VERSION=11.0.221 CUDA_DRIVER_VERSION=450.51.06 CUDA_CACHE_DISABLE=1
RUN RUN |2 CUDA_VERSION=11.0.221 CUDA_DRIVER_VERSION=450.51.06 /bin/sh -c /nvidia/build-scripts/installCUDA.sh # buildkit
RUN COPY cudaCheck /tmp/cudaCheck # buildkit
RUN RUN |2 CUDA_VERSION=11.0.221 CUDA_DRIVER_VERSION=450.51.06 /bin/sh -c cp /tmp/cudaCheck/$(uname -m)/sh-wrap /bin/sh-wrap # buildkit
RUN RUN |2 CUDA_VERSION=11.0.221 CUDA_DRIVER_VERSION=450.51.06 /bin/sh -c cp /tmp/cudaCheck/$(uname -m)/cudaCheck /usr/local/bin/ # buildkit
RUN RUN |2 CUDA_VERSION=11.0.221 CUDA_DRIVER_VERSION=450.51.06 /bin/sh -c rm -rf /tmp/cudaCheck # buildkit
RUN COPY cudaCheck/shinit_v2 /etc/shinit_v2 # buildkit
RUN COPY cudaCheck/startup_scripts.patch /tmp # buildkit
RUN COPY singularity /.singularity.d # buildkit
RUN RUN |2 CUDA_VERSION=11.0.221 CUDA_DRIVER_VERSION=450.51.06 /bin/sh -c patch -p0 < /tmp/startup_scripts.patch \
&& rm -f /tmp/startup_scripts.patch \
&& ln -sf /.singularity.d/libs /usr/local/lib/singularity \
&& ln -sf sh-wrap /bin/sh # buildkit
RUN ENV _CUDA_COMPAT_PATH=/usr/local/cuda/compat ENV=/etc/shinit_v2 BASH_ENV=/etc/bash.bashrc NVIDIA_REQUIRE_CUDA=cuda>=9.0
RUN LABEL com.nvidia.volumes.needed=nvidia_driver com.nvidia.cuda.version=9.0
RUN ARG NCCL_VERSION
RUN ARG CUBLAS_VERSION
RUN ARG CUFFT_VERSION
RUN ARG CURAND_VERSION
RUN ARG CUSPARSE_VERSION
RUN ARG CUSOLVER_VERSION
RUN ARG NPP_VERSION
RUN ARG NVJPEG_VERSION
RUN ARG CUDNN_VERSION
RUN ARG TRT_VERSION
RUN ARG TRTOSS_VERSION
RUN ARG NSIGHT_SYSTEMS_VERSION
RUN ARG NSIGHT_COMPUTE_VERSION
RUN ENV NCCL_VERSION=2.7.8 CUBLAS_VERSION=11.2.0.252 CUFFT_VERSION=10.2.1.245 CURAND_VERSION=10.2.1.245 CUSPARSE_VERSION=11.1.1.245 CUSOLVER_VERSION=10.6.0.245 NPP_VERSION=11.1.0.245 NVJPEG_VERSION=11.1.1.245 CUDNN_VERSION=8.0.4.12 TRT_VERSION=7.1.3.4 TRTOSS_VERSION=20.09 NSIGHT_SYSTEMS_VERSION=2020.3.2.6 NSIGHT_COMPUTE_VERSION=2020.1.2.4
RUN RUN |15 CUDA_VERSION=11.0.221 CUDA_DRIVER_VERSION=450.51.06 NCCL_VERSION=2.7.8 CUBLAS_VERSION=11.2.0.252 CUFFT_VERSION=10.2.1.245 CURAND_VERSION=10.2.1.245 CUSPARSE_VERSION=11.1.1.245 CUSOLVER_VERSION=10.6.0.245 NPP_VERSION=11.1.0.245 NVJPEG_VERSION=11.1.1.245 CUDNN_VERSION=8.0.4.12 TRT_VERSION=7.1.3.4 TRTOSS_VERSION=20.09 NSIGHT_SYSTEMS_VERSION=2020.3.2.6 NSIGHT_COMPUTE_VERSION=2020.1.2.4 /bin/sh -c /nvidia/build-scripts/installNCCL.sh \
&& /nvidia/build-scripts/installLIBS.sh \
&& /nvidia/build-scripts/installCUDNN.sh \
&& /nvidia/build-scripts/installTRT.sh \
&& /nvidia/build-scripts/installNSYS.sh \
&& /nvidia/build-scripts/installNCU.sh # buildkit
RUN LABEL com.nvidia.nccl.version=2.7.8 com.nvidia.cublas.version=11.2.0.252 com.nvidia.cufft.version=10.2.1.245 com.nvidia.curand.version=10.2.1.245 com.nvidia.cusparse.version=11.1.1.245 com.nvidia.cusolver.version=10.6.0.245 com.nvidia.npp.version=11.1.0.245 com.nvidia.nvjpeg.version=11.1.1.245 com.nvidia.cudnn.version=8.0.4.12 com.nvidia.tensorrt.version=7.1.3.4 com.nvidia.tensorrtoss.version=20.09 com.nvidia.nsightsystems.version=2020.3.2.6 com.nvidia.nsightcompute.version=2020.1.2.4
RUN ARG DALI_VERSION
RUN ARG DALI_BUILD
RUN ARG DLPROF_VERSION
RUN ENV DALI_VERSION=0.25.1 DALI_BUILD=1612461 DLPROF_VERSION=20.09
RUN ADD docs.tgz / # buildkit
RUN RUN |18 CUDA_VERSION=11.0.221 CUDA_DRIVER_VERSION=450.51.06 NCCL_VERSION=2.7.8 CUBLAS_VERSION=11.2.0.252 CUFFT_VERSION=10.2.1.245 CURAND_VERSION=10.2.1.245 CUSPARSE_VERSION=11.1.1.245 CUSOLVER_VERSION=10.6.0.245 NPP_VERSION=11.1.0.245 NVJPEG_VERSION=11.1.1.245 CUDNN_VERSION=8.0.4.12 TRT_VERSION=7.1.3.4 TRTOSS_VERSION=20.09 NSIGHT_SYSTEMS_VERSION=2020.3.2.6 NSIGHT_COMPUTE_VERSION=2020.1.2.4 DALI_VERSION=0.25.1 DALI_BUILD=1612461 DLPROF_VERSION=20.09 /bin/sh -c echo "/usr/local/nvidia/lib" >> /etc/ld.so.conf.d/nvidia.conf \
&& echo "/usr/local/nvidia/lib64" >> /etc/ld.so.conf.d/nvidia.conf # buildkit
RUN ENV PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin LD_LIBRARY_PATH=/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64 NVIDIA_VISIBLE_DEVICES=all NVIDIA_DRIVER_CAPABILITIES=compute,utility,video
RUN COPY deviceQuery/deviceQuery /usr/local/bin/ # buildkit
RUN COPY deviceQuery/checkSMVER.sh /usr/local/bin/ # buildkit
RUN RUN /bin/sh -c export DEBIAN_FRONTEND=noninteractive \
&& apt-get update \
&& apt-get install -y --no-install-recommends git libglib2.0-0 less libnl-route-3-200 libnuma-dev libnuma1 libpmi2-0-dev nano numactl openssh-client vim \
&& rm -rf /var/lib/apt/lists/* # buildkit
RUN COPY mellanox /opt/mellanox # buildkit
RUN ARG MOFED_VERSION=4.6-1.0.1
RUN ENV MOFED_VERSION=4.6-1.0.1 IBV_DRIVERS=/usr/lib/libibverbs/libmlx5
RUN RUN |1 MOFED_VERSION=4.6-1.0.1 /bin/sh -c dpkg -i /opt/mellanox/DEBS/${MOFED_VERSION}/*_$(dpkg --print-architecture).deb \
&& ln -sf /opt/mellanox/change_mofed_version.sh /usr/local/bin/ \
&& cd /usr/bin \
&& chmod a+w ibv_* \
&& cd /usr/lib \
&& touch libmlx5.so.1.0.0 \
&& chmod -R a+w libibverbs* libmlx5.so.1.0.0 \
&& ln -sf libmlx5.so.1.0.0 libmlx5.so.1 \
&& chmod a+w libmlx5.so.1 \
&& rm -f /etc/libibverbs.d/* # buildkit
RUN COPY openpmix-pmi1.patch /tmp # buildkit
RUN ARG OPENUCX_VERSION=1.6.1
RUN ARG OPENMPI_VERSION=3.1.6
RUN ENV OPENUCX_VERSION=1.6.1 OPENMPI_VERSION=3.1.6
RUN RUN |3 MOFED_VERSION=4.6-1.0.1 OPENUCX_VERSION=1.6.1 OPENMPI_VERSION=3.1.6 /bin/sh -c wget -q -O - https://github.com/openucx/ucx/releases/download/v${OPENUCX_VERSION}/ucx-${OPENUCX_VERSION}.tar.gz | tar -xzf - \
&& cd ucx-${OPENUCX_VERSION} \
&& ./configure --prefix=/usr/local/ucx --enable-mt \
&& make -j"$(nproc)" install \
&& cd .. \
&& rm -rf ucx-${OPENUCX_VERSION} \
&& echo "/usr/local/ucx/lib" >> /etc/ld.so.conf.d/openucx.conf \
&& wget -q -O - https://www.open-mpi.org/software/ompi/v$(echo "${OPENMPI_VERSION}" | cut -d . -f 1-2)/downloads/openmpi-${OPENMPI_VERSION}.tar.gz | tar -xzf - \
&& cd openmpi-${OPENMPI_VERSION} \
&& ln -sf /usr/include/slurm-wlm /usr/include/slurm \
&& ( cd opal/mca/pmix/pmix2x/pmix \
&& patch -p1 < /tmp/openpmix-pmi1.patch \
&& rm -f /tmp/openpmix-pmi1.patch ) \
&& ./configure --enable-orterun-prefix-by-default --with-verbs --with-pmi --with-pmix=internal --with-ucx=/usr/local/ucx --with-ucx-libdir=/usr/local/ucx/lib --enable-mca-no-build=btl-uct --prefix=/usr/local/mpi --disable-getpwuid \
&& make -j"$(nproc)" install \
&& cd .. \
&& rm -rf openmpi-${OPENMPI_VERSION} \
&& echo "/usr/local/mpi/lib" >> /etc/ld.so.conf.d/openmpi.conf \
&& rm -f /usr/lib/libibverbs.so /usr/lib/libibverbs.a \
&& ldconfig # buildkit
RUN ENV PATH=/usr/local/mpi/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin
RUN COPY singularity /.singularity.d # buildkit
RUN COPY cuda-*.patch /tmp # buildkit
RUN RUN |3 MOFED_VERSION=4.6-1.0.1 OPENUCX_VERSION=1.6.1 OPENMPI_VERSION=3.1.6 /bin/sh -c export DEVEL=1 BASE=0 \
&& /nvidia/build-scripts/installNCU.sh \
&& /nvidia/build-scripts/installCUDA.sh \
&& /nvidia/build-scripts/installNCCL.sh \
&& /nvidia/build-scripts/installLIBS.sh \
&& /nvidia/build-scripts/installCUDNN.sh \
&& /nvidia/build-scripts/installTRT.sh \
&& /nvidia/build-scripts/installNSYS.sh \
&& if [ -f "/tmp/cuda-${_CUDA_VERSION_MAJMIN}.patch" ]; then patch -p0 < /tmp/cuda-${_CUDA_VERSION_MAJMIN}.patch; fi \
&& if [ ! -f /usr/include/x86_64-linux-gnu/cudnn_backend_v8.h ]; then cp /nvidia/build-scripts/cudnn_backend.h /usr/include/x86_64-linux-gnu/cudnn_backend_v8.h; ln -s /usr/include/x86_64-linux-gnu/cudnn_backend_v8.h /usr/include/x86_64-linux-gnu/cudnn_backend.h; fi \
&& rm -f /tmp/cuda-*.patch # buildkit
RUN ENV LIBRARY_PATH=/usr/local/cuda/lib64/stubs:
RUN ENV LD_LIBRARY_PATH=/usr/local/cuda/extras/CUPTI/lib64:/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
RUN ARG TENSORFLOW_VERSION
RUN ENV TENSORFLOW_VERSION=2.3.0
RUN LABEL com.nvidia.tensorflow.version=2.3.0
RUN ARG NVIDIA_TENSORFLOW_VERSION
RUN ENV NVIDIA_TENSORFLOW_VERSION=20.09-tf2
RUN ARG PYVER=3.6
RUN ENV PYVER=3.6
RUN RUN |3 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 /bin/sh -c PYSFX=`[ "$PYVER" != "2.7" ] \
&& echo "$PYVER" | cut -c1-1 || echo ""` \
&& DISTUTILS=`[ "$PYVER" == "3.6" ] \
&& echo "python3-distutils" || echo""` \
&& apt-get update \
&& apt-get install -y --no-install-recommends cmake pkg-config python$PYVER python$PYVER-dev python$PYSFX-pip $DISTUTILS rsync swig unzip zip zlib1g-dev libgoogle-glog-dev libjsoncpp-dev \
&& rm -rf /var/lib/apt/lists/* # buildkit
RUN ENV PYTHONIOENCODING=utf-8 LC_ALL=C.UTF-8
RUN RUN |3 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 /bin/sh -c rm -f /usr/bin/python \
&& rm -f /usr/bin/python`echo $PYVER | cut -c1-1` \
&& ln -s /usr/bin/python$PYVER /usr/bin/python \
&& ln -s /usr/bin/python$PYVER /usr/bin/python`echo $PYVER | cut -c1-1` # buildkit
RUN RUN |3 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 /bin/sh -c mkdir -p /usr/lib/x86_64-linux-gnu/include/ \
&& ln -s /usr/include/cudnn.h /usr/lib/x86_64-linux-gnu/include/cudnn.h # buildkit
RUN RUN |3 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 /bin/sh -c curl -O https://bootstrap.pypa.io/get-pip.py \
&& python get-pip.py \
&& rm get-pip.py # buildkit
RUN RUN |3 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 /bin/sh -c pip install --no-cache-dir --upgrade setuptools==49.6.0 numpy==$(test ${PYVER%.*} -eq 2 \
&& echo 1.16.5 || echo 1.17.3) pexpect==4.7.0 psutil==5.7.0 nltk==3.4.5 future==0.18.2 jupyterlab==$(test ${PYVER%.*} -eq 2 \
&& echo 0.33.12 || echo 1.2.14) notebook==$(test ${PYVER%.*} -eq 2 \
&& echo 5.7.8 || echo 6.0.3) mock==3.0.5 portpicker==1.3.1 h5py==2.10.0 scipy==$(test ${PYVER%.*} -eq 2 \
&& echo 1.2.2 || echo 1.4.1) keras_preprocessing==1.1.1 keras_applications==1.0.8 # buildkit
RUN ENV BAZELRC=/root/.bazelrc
RUN RUN |3 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 /bin/sh -c echo "startup --batch" >> $BAZELRC \
&& echo "build --spawn_strategy=standalone --genrule_strategy=standalone" >> $BAZELRC \
&& echo "startup --max_idle_secs=0" >> $BAZELRC # buildkit
RUN ARG BAZEL_VERSION=2.0.0
RUN RUN |4 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 /bin/sh -c mkdir /bazel \
&& cd /bazel \
&& curl -fSsL -O https://github.com/bazelbuild/bazel/releases/download/$BAZEL_VERSION/bazel-$BAZEL_VERSION-installer-linux-x86_64.sh \
&& curl -fSsL -o /bazel/LICENSE.txt https://raw.githubusercontent.com/bazelbuild/bazel/master/LICENSE \
&& bash ./bazel-$BAZEL_VERSION-installer-linux-x86_64.sh \
&& rm -rf /bazel # buildkit
RUN WORKDIR /opt/tensorflow
RUN COPY mkl-dnn.patch . # buildkit
RUN RUN |4 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 /bin/sh -c MKL_DNN_COMMIT=733fc908874c71a5285043931a1cf80aa923165c \
&& curl -fSsL -O https://mirror.bazel.build/github.com/intel/mkl-dnn/archive/${MKL_DNN_COMMIT}.tar.gz \
&& tar -xf ${MKL_DNN_COMMIT}.tar.gz \
&& rm -f ${MKL_DNN_COMMIT}.tar.gz \
&& cd mkl-dnn-${MKL_DNN_COMMIT} \
&& patch -p0 < /opt/tensorflow/mkl-dnn.patch \
&& cd .. \
&& tar --mtime='1970-01-01' -cf mkl-dnn-${MKL_DNN_COMMIT}-patched.tar mkl-dnn-${MKL_DNN_COMMIT}/ \
&& rm -rf mkl-dnn-${MKL_DNN_COMMIT}/ \
&& gzip -n mkl-dnn-${MKL_DNN_COMMIT}-patched.tar # buildkit
RUN RUN |4 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 /bin/sh -c mkdir -p /opt/dlprof \
&& cp /nvidia/opt/dlprof/bin/dlprof /opt/dlprof/ \
&& cp /nvidia/workspace/LICENSE /opt/dlprof/ \
&& cp /nvidia/workspace/README.md /opt/dlprof/ \
&& ln -sf /opt/dlprof/dlprof /usr/local/bin/dlprof # buildkit
RUN COPY tensorboard.patch /tmp/ # buildkit
RUN RUN |4 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 /bin/sh -c pip install --no-cache-dir 'tensorboard>=2.2,<=2.3' \
&& TBFILE=$(find $(find /usr /opt -type d -name tensorboard) -type f -name core_plugin.py) \
&& patch $TBFILE < /tmp/tensorboard.patch \
&& rm -f /tmp/tensorboard.patch # buildkit
RUN ENV NVM_DIR=/usr/local/nvm
RUN RUN |4 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 /bin/sh -c pip install --no-cache-dir git+https://github.com/lspvic/jupyter_tensorboard.git@refs/pull/63/merge \
&& mkdir -p $NVM_DIR \
&& curl -Lo- https://raw.githubusercontent.com/nvm-sh/nvm/v0.35.3/install.sh | bash \
&& source "$NVM_DIR/nvm.sh" \
&& nvm install node \
&& jupyter labextension install jupyterlab_tensorboard \
&& jupyter serverextension enable jupyterlab \
&& pip install --no-cache-dir jupytext \
&& jupyter labextension install jupyterlab-jupytext \
&& ( cd $NVM_DIR/versions/node/$(node -v)/lib/node_modules/npm \
&& ( cd node_modules/libnpx \
&& npm install update-notifier@^4.1.0 --production \
&& npm install yargs@^15.3.1 --production --force \
&& cd ../.. \
&& npm install update-notifier@^4.1.0 --production ) \
&& npm prune --production ) \
&& ( cd /usr/local/share/jupyter/lab/staging \
&& npm prune --production ) \
&& npm cache clean --force \
&& rm -rf /usr/local/share/.cache \
&& echo "source $NVM_DIR/nvm.sh" >> /etc/bash.bashrc \
&& mv /root/.jupyter/jupyter_notebook_config.json /usr/local/etc/jupyter/ # buildkit
RUN COPY jupyter_notebook_config.py /usr/local/etc/jupyter/ # buildkit
RUN ENV JUPYTER_PORT=8888
RUN ENV TENSORBOARD_PORT=6006
RUN ENV TENSORBOARD_DEBUGGER_PORT=6064
RUN EXPOSE map[8888/tcp:{}]
RUN EXPOSE map[6006/tcp:{}]
RUN EXPOSE map[6064/tcp:{}]
RUN COPY tensorflow-source ./tensorflow-source # buildkit
RUN COPY nvbuild.sh nvbuildopts nvbazelcache bazel_build.sh test_grabber.sh ./ # buildkit
RUN COPY tensorflow-logo.jpg /opt/tensorflow/tensorflow-source/tensorflow/docs_src/tensorflow-logo.jpg # buildkit
RUN ARG TFAPI=1
RUN ARG BAZEL_CACHE=
RUN RUN |6 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache /bin/sh -c /opt/tensorflow/nvbuild.sh --triton --testlist --python$PYVER --v$TFAPI $BAZEL_CACHE # buildkit
RUN ENV TF_ADJUST_HUE_FUSED=1 TF_ADJUST_SATURATION_FUSED=1 TF_ENABLE_WINOGRAD_NONFUSED=1 TF_AUTOTUNE_THRESHOLD=2 TF_USE_CUDNN_BATCHNORM_SPATIAL_PERSISTENT=1
RUN COPY horovod-source/ ./horovod-source/ # buildkit
RUN RUN |6 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache /bin/sh -c export HOROVOD_GPU_OPERATIONS=NCCL \
&& export HOROVOD_NCCL_INCLUDE=/usr/include \
&& export HOROVOD_NCCL_LIB=/usr/lib/x86_64-linux-gnu \
&& export HOROVOD_NCCL_LINK=SHARED \
&& export HOROVOD_WITHOUT_PYTORCH=1 \
&& export HOROVOD_WITHOUT_MXNET=1 \
&& export HOROVOD_WITH_TENSORFLOW=1 \
&& export HOROVOD_WITH_MPI=1 \
&& export HOROVOD_BUILD_ARCH_FLAGS="-march=sandybridge -mtune=broadwell" \
&& cd /opt/tensorflow/horovod-source \
&& ln -s /usr/local/cuda/lib64/stubs/libcuda.so ./libcuda.so.1 \
&& export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PWD \
&& rm -rf ./dist \
&& python setup.py sdist \
&& pip install --no-cache --no-cache-dir dist/horovod-*.tar.gz \
&& python setup.py clean \
&& rm ./libcuda.so.1 # buildkit
RUN RUN |6 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache /bin/sh -c if dpkg --compare-versions ${DALI_VERSION} lt 0.23; then DALI_URL_SUFFIX="/cuda/${CUDA_VERSION%%.*}.0"; else DALI_PKG_SUFFIX="-cuda${CUDA_VERSION%%.*}0"; fi \
&& pip install --no-cache-dir --upgrade --extra-index-url https://developer.download.nvidia.com/compute/redist --extra-index-url http://sqrl/dldata/pip-simple${DALI_URL_SUFFIX:-} --trusted-host sqrl nvidia-dali${DALI_PKG_SUFFIX:-}==${DALI_VERSION} \
&& pip install --no-cache-dir --upgrade --extra-index-url https://developer.download.nvidia.com/compute/redist --extra-index-url http://sqrl/dldata/pip-simple${DALI_URL_SUFFIX:-} --trusted-host sqrl nvidia-dali-tf-plugin${DALI_PKG_SUFFIX:-}==${DALI_VERSION} # buildkit
RUN RUN |6 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache /bin/sh -c pip install --no-cache-dir --upgrade tensorflow-datasets==3.1.0 # buildkit
RUN COPY addons_cuda11.1.patch . # buildkit
RUN COPY tf-addons/ ./tf-addons # buildkit
RUN RUN |6 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache /bin/sh -c cd tf-addons \
&& patch -p1 < ../addons_cuda11.1.patch \
&& TF_NEED_CUDA=1 TF_CUDA_VERSION=$(echo "${CUDA_VERSION}" | cut -d . -f 1-2) TF_CUDNN_VERSION=$(echo "${CUDNN_VERSION}" | cut -d . -f 1) TF_CUDA_COMPUTE_CAPABILITIES="5.2,6.0,6.1,7.0,7.5,8.0" CUDA_TOOLKIT_PATH=/usr/local/cuda CUDNN_INSTALL_PATH=/usr/lib/x86_64-linux-gnu python configure.py \
&& bazel build --enable_runfiles build_pip_pkg \
&& bazel-bin/build_pip_pkg artifacts \
&& pip install artifacts/tensorflow_addons-*.whl \
&& bazel clean --expunge \
&& rm -rf artifacts # buildkit
RUN ARG NVTX_PLUGINS_VERSION=0.1.8
RUN RUN |7 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache NVTX_PLUGINS_VERSION=0.1.8 /bin/sh -c pip install --no-cache-dir --upgrade nvtx-plugins==${NVTX_PLUGINS_VERSION} # buildkit
RUN COPY nvidia_entrypoint.sh /usr/local/bin/ # buildkit
RUN ENTRYPOINT ["/usr/local/bin/nvidia_entrypoint.sh"]
RUN WORKDIR /workspace
RUN RUN |7 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache NVTX_PLUGINS_VERSION=0.1.8 /bin/sh -c chmod -R a+w /workspace # buildkit
RUN COPY NVREADME.md README.md # buildkit
RUN COPY docker-examples docker-examples # buildkit
RUN COPY nvidia-examples /workspace/nvidia-examples # buildkit
RUN RUN |7 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache NVTX_PLUGINS_VERSION=0.1.8 /bin/sh -c BASE=0 DEVEL=0 SAMPLES=0 PYTHON=1 /nvidia/build-scripts/installTRT.sh \
&& mkdir -p /usr/src/tensorrt/python \
&& cd /usr/src/tensorrt/python \
&& BASE=0 DEVEL=0 SAMPLES=0 PYTHON=0 TENSORFLOW=1 FETCHONLY=1 /nvidia/build-scripts/installTRT.sh \
&& mv /usr/src/tensorrt /opt \
&& ln -s /opt/tensorrt /usr/src/tensorrt # buildkit
RUN COPY trt_python_setup.sh /opt/tensorrt/python/python_setup.sh # buildkit
RUN RUN |7 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache NVTX_PLUGINS_VERSION=0.1.8 /bin/sh -c cd /opt/tensorrt/python \
&& ./python_setup.sh # buildkit
RUN ENV PATH=/usr/local/mpi/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/tensorrt/bin
RUN RUN |7 TENSORFLOW_VERSION=2.3.0 NVIDIA_TENSORFLOW_VERSION=20.09-tf2 PYVER=3.6 BAZEL_VERSION=3.1.0 TFAPI=2 BAZEL_CACHE=--bazel-cache NVTX_PLUGINS_VERSION=0.1.8 /bin/sh -c ln -sf ${_CUDA_COMPAT_PATH}/lib.real ${_CUDA_COMPAT_PATH}/lib \
&& echo ${_CUDA_COMPAT_PATH}/lib > /etc/ld.so.conf.d/00-cuda-compat.conf \
&& ldconfig \
&& rm -f ${_CUDA_COMPAT_PATH}/lib # buildkit
RUN ARG NVIDIA_BUILD_ID
RUN ENV NVIDIA_BUILD_ID=16003717
RUN LABEL com.nvidia.build.id=16003717
RUN ARG NVIDIA_BUILD_REF
RUN LABEL com.nvidia.build.ref=bbb7f8edd544619c81635e75dd330bff437db82e2.3 针对官方镜像解析出的Dockfile进行分析
官方镜像的jupyer配置与AIStation平台的配置有区别，无法进行随机分配jupyter端口，以及没有指定外部访问的JUPYTER_HOST等环境变量。导致平台开发环境启动容器后，无法访问jupyter页面。
分析出实际问题后，针对这些问题，修改我们的Dockerfile文件，以NGC官方镜像为基础镜像，制作出AIStation平台需要的镜像。
修改我们的Dockerfile文件，适配AIStation平台：
FROM nvcr.io/nvidia/tensorflow:20.09-tf2-py3
# Install OpenSSH for MPI to communicate between containers
RUN apt-get -o Acquire::Check-Valid-Until=false -o Acquire::Check-Date=false update && apt-get install -y --no-install-recommends openssh-client openssh-server && \
mkdir -p /var/run/sshd
# Allow OpenSSH to talk to containers without asking for confirmation
RUN cat /etc/ssh/ssh_config | grep -v StrictHostKeyChecking > /etc/ssh/ssh_config.new && \
echo " StrictHostKeyChecking no" >> /etc/ssh/ssh_config.new && \
cat /etc/ssh/sshd_config | grep -v PermitRootLogin> /etc/ssh/sshd_config.new && \
echo "PermitRootLogin yes" >> /etc/ssh/sshd_config.new && \
mv /etc/ssh/ssh_config.new /etc/ssh/ssh_config && \
mv /etc/ssh/sshd_config.new /etc/ssh/sshd_config
##jupyter
RUN wget -O /usr/local/etc/jupyter/jupyter_notebook_config.py https://raw.githubusercontent.com/Winowang/jupyter_gpu/master/jupyter_notebook_config.py
RUN wget -O /usr/local/etc/jupyter/custom.js https://raw.githubusercontent.com/Winowang/jupyter_gpu/master/custom.js
## tini
ENV TINI_VERSION v0.18.0
ADD https://github.com/krallin/tini/releases/download/${TINI_VERSION}/tini /tini
RUN chmod +x /tini
WORKDIR "/home"2.4 通过平台的创建镜像功能制作镜像（需要联网）
将我们修改的Dockerfile文件，通过平台的创建功能，将镜像引入平台：
使用该镜像，在开发环境提交任务，然后平台可以正常使用jupyter：
