dockerfile语法

• 常用的关键字：

FROM

RUN

USER

ADD

• 符号：

&&

/

• 切换源：

pip install -i

conda install -c

docker命令

docker run -it image_name

docker build -t iamge_name .

docker tag 【image_id】 【iamge_full_name】
docker images | grep 【key_word】

docker push 【image_full_name】

docker pull 【image_full_name】

注意：

image_full_name ：【server_host]】/ 【project_root】 / 【name】： 【tag】

build成功后可以通过docker run 启动一个 container ，通过 conda list | greep *** 查看安装的库是否有冲突：

root@239b60aff0c6:~# conda list | grep numpy
numpy 1.14.5 py36hcd700cb_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
numpy 1.14.5 <pip>
numpy-base 1.14.5 py36hdbf6ddf_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
root@239b60aff0c6:~# conda list | grep keras
keras 2.2.0 0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
keras-applications 1.0.2 py_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
keras-base 2.2.0 py36_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
keras-gpu 2.2.0 0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
keras-preprocessing 1.0.1 py_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
root@239b60aff0c6:~# conda list | grep mxnet
mxnet-cu92 1.2.0 <pip>
root@239b60aff0c6:~# conda list | grep pytorch
cuda90 1.0 h6433d27_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch
pytorch 0.4.0 py36_cuda9.0.176_cudnn7.1.2_1 [cuda90] https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch
torchvision 0.2.1 py36_1 https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch
root@239b60aff0c6:~# conda list | grep scipy
scipy 1.1.0 <pip>
scipy 1.1.0 py36hfc37229_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
root@239b60aff0c6:~# conda list | grep matl
root@239b60aff0c6:~# conda list | grep matplot
matplotlib 2.2.2 <pip>
root@239b60aff0c6:~# conda list | grep pandas
pandas 0.23.1 <pip>

例子

FROM docker.io/deepintelligent/tensorflow-1.8.0-notebook-gpu
USER root

ADD sources.list /etc/apt/

RUN conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/ &&
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/ &&
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/noarch &&
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/ &&\

conda config --add channels https://mirrors.ustc.edu.cn/anaconda/pkgs/free/ &&\

conda config --add channels https://mirrors.ustc.edu.cn/anaconda/pkgs/main/ &&\

conda config --set show_channel_urls yes
RUN pip install -i https://pypi.tuna.tsinghua.edu.cn/simple mxnet-cu92
RUN conda install keras-gpu
RUN conda install -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/ pytorch torchvision cuda90 -c pytorch

RUN conda config --set ssl_verify no

RUN conda install -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/ caffe2 caffe2-cuda8.0-cudnn7

命令例子

[root@st1-deepintelligent-1] /data4/huineng/gpu$docker tag 01720871f2ae mangan-prod-3.srv.yiran.com/deepintelligent/tensorflow-keras-mxnet-pytorch-gpu:20180710-v1

[root@st1-deepintelligent-1] /data4/huineng/gpu$ docker push ccr.ccs.tencentyun.com/deepintelligent/tensorflow-keras-mxnet-pytorch-gpu:20180710-v1

kubectl describe pod -n=tfworkflow jupyter-huineng

[root@adml9st] /data3$ docker pull ccr.ccs.tencentyun.com/deepintelligent/tensorflow-keras-mxnet-pytorch-gpu:20180710-v1