Since AI_Station uses a container to build the environment, and only provides image upload and download functions, it is not easy to provide network functions, so it is necessary to pull the image locally on the platform, install some necessary software, and then package it as an image and upload it back. Therefore, you need to build the docker environment locally, as follows:

Install GPG certificate

curl -fsSL http://mirrors.aliyun.com/docker-ce/linux/ubuntu/gpg | sudo apt-key add -

Write software source information

sudo add-apt-repository "deb [arch=amd64] http://mirrors.aliyun.com/docker-ce/linux/ubuntu $(lsb_release -cs) stable"

renew

sudo apt-get -y update

Install Docker-CE

Install the latest version:

sudo apt-get -y install docker-ce

Or install the specified version in this article:

Find the version of Docker-CE

apt-cache madison docker-ce

(The second column marked in red is the version number)

Install the specified version: (The version installed here is 5:18.09.9_3-0ubuntu-bionic)

sudo apt-get -y install docker-ce=5:18.09.9~3-0~ubuntu-bionic

Check the installed docker version:

sudo docker version

Pull the mirrors commonly used on the platform to the local (the platform's ftp download is used here)

Use the load method to load the image locally:

docker load -i ******.tar

sudo docker load -i 14.14.14.100_5000_other_ubuntu18.04-python3.7.5-openssh7.6-jupyterlab1.2.3-inspur_latest.tar

Check out the local mirror:

sudo docker images

You can see that all three images on the platform have been imported.