docker提交镜像到阿里ack整体流程

本笔记为阿里云天池龙珠计划Docker训练营的学习内容，链接为：https://tianchi.aliyun.com/specials/activity/promotion/aicampdocker；

➜  testv0928 mkdir docker_test01
➜  testv0928 cd docker_test01 

使用ack账号登陆docker
➜  docker_test01 docker login --username=**** registry.cn-hangzhou.aliyuncs.com
  Password: 
  Login Succeeded
  Logging in with your password grants your terminal complete access to your account. 
  For better security, log in with a limited-privilege personal access token. Learn more at https://docs.docker.com/go/access-tokens/
➜  docker_test01 vim math.py
➜  docker_test01 vim run.sh
➜  docker_test01 cat math.py
  import torch
  device = torch.device("cuda")
  a = torch.randn(3, 3)
  b = torch.randn(3, 3)
  a = a.to(device)
  b = b.to(device)
  c = torch.matmul(a,b)
  print(c)
➜  docker_test01 cat run.sh
  #bin/bash
  #打印GPU信息
  nvidia-smi
  #执行math.py
  python3 math.py
➜  docker_test01 vim Dockerfile
➜  docker_test01 cat Dockerfile 
  # Base Images
  ## 从天池基础镜像构建(from的base img 根据自己的需要更换，建议使用天池open list镜像链接：https://tianchi.aliyun.com/forum/postDetail?postId=67720)
  FROM registry.cn-shanghai.aliyuncs.com/tcc-public/pytorch:1.4-cuda10.1-py3
  
  ##安装依赖包,pip包请在requirements.txt添加
  #RUN pip install --no-cache-dir -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple
  
  ## 把当前文件夹里的文件构建到镜像的//workspace目录下,并设置为默认工作目录
  ADD math.py /workspace
  ADD run.sh /workspace
  WORKDIR /workspace
  
  ## 镜像启动后统一执行 sh run.sh
  CMD ["sh", "run.sh"]
➜  docker_test01 docker build -t registry.cn-hangzhou.aliyuncs.com/ly_22_test/docker_test01:01
  "docker build" requires exactly 1 argument.
  See 'docker build --help'.
  
  Usage:  docker build [OPTIONS] PATH | URL | -
  
  Build an image from a Dockerfile
➜  docker_test01 ls                                                                           
    Dockerfile math.py    run.sh
➜  docker_test01 ll
  total 24