rancher 部署ipython-notebook.md

发表于 | 分类于 |

对于数据工程师来说ipython/notebook是常用的编写python的工具

如何快速部署

部署ipython-notebook当然不难，不过为了办公方便，在家和在公司都能使用同一个环境。研究了一下在rancher上部署ipyhon。 在docker hub 上查找了一下ipython/notebook很快就找到了镜像。看来一下overview基本上就清楚如何部署。在服务器上直接如下命令即可。

# 拉去镜像
docker pull ipython/notebook

# 启动镜像
docker run -d -p 80:8888 -e "PASSWORD=MakeAPassword" -e "USE_HTTP=1" ipython/notebook

当然在rancher上部署也容器，在启动容器的时候注意配置上PASSWORD=，USE_HTTP=这两个环境变量即可。

访问

https://ip:80

问题

由于镜像安装的python版本是2.7.5 在安装numpy会遇到异常

  Collecting numpy
  Using cached https://files.pythonhosted.org/packages/ac/36/325b27ef698684c38b1fe2e546e2e7ef9cecd7037bcdb35c87efec4356af/numpy-1.17.2.zip
    Complete output from command python setup.py egg_info:
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/tmp/pip-build-ipyhqlik/numpy/setup.py", line 31, in <module>
        raise RuntimeError("Python version >= 3.5 required.")
    RuntimeError: Python version >= 3.5 required. 解决方案

numpy以及scipy的新版本已不再支持python2，但有很多库装的时候依赖numpy和scipy，
会默认装最新的版本，但最新版本又不支持python2，所以记录下支持python2的版本号：
pip install numpy==1.16.1
pip install scipy==1.2.2

更新

由于这个版本所用的python版本为2.7.5。
替换到python3.5.2。
dockerfile内容如下:

# Ubuntu 16.04 with python 3.5.3 with jupyter
# docker build -t nikhilmetrani/python3-jupyter:latest .

FROM ubuntu:16.04
MAINTAINER Nikhil Metrani <Nikhil Metrani>

RUN apt-get update
RUN DEBIAN_FRONTEND=noninteractive \
    apt-get install -y --no-install-recommends apt-utils
RUN DEBIAN_FRONTEND=noninteractive \
    apt-get install -y --no-install-recommends \
            curl \
            xz-utils \
            build-essential \
            libsqlite3-dev \
            libreadline-dev \
            libssl-dev \
            ca-certificates \
            openssl \
        && apt-get clean \
        && rm -rf /var/lib/apt/lists/*

WORKDIR /tmp
RUN curl -O https://www.python.org/ftp/python/3.5.3/Python-3.5.3.tar.xz
RUN tar -xf Python-3.5.3.tar.xz
WORKDIR /tmp/Python-3.5.3
RUN ./configure
RUN make \
    && make install

WORKDIR /
RUN rm -rf /tmp/Python-3.5.3.tar.xz /tmp/Python-3.5.3

RUN pip3 install --no-cache-dir \
        ipykernel \
        jupyter

# Set up our notebook config.
COPY support/jupyter_config.py /root/.jupyter/

# Jupyter has issues with being run directly:
#   https://github.com/ipython/ipython/issues/7062
# We just add a little wrapper script.
COPY support/jupyter_launch.sh /
RUN chmod +x /jupyter_launch.sh

# IPython
EXPOSE 8888

WORKDIR "/home"

CMD ["/jupyter_launch.sh"]

jupyter_launch.sh 内容如下:

jupyter notebook "$@"

jupyter_config.py 内容如下:

import os
from IPython.lib import passwd

c.NotebookApp.ip = '*'
c.NotebookApp.port = int(os.getenv('PORT', 8888))
c.NotebookApp.open_browser = False
c.MultiKernelManager.default_kernel_name = 'python3'
c.NotebookApp.allow_root = True

# sets a password if PASSWORD is set in the environment
if 'PASSWORD' in os.environ:
  c.NotebookApp.password = passwd(os.environ['PASSWORD'])
  del os.environ['PASSWORD']

总结

一些常用的工具通过docker部署之后共享给公司其他人用非常方便。

Life is more than the present.