配置tensorflow docker环境并用ssh连接

配置tensorflow docker环境并用ssh连接

1. 配置并且进入到tensorflow docker中

 nvidia-docker run -itd --name "tf18" -p 6000:8888 -p 6100:22 -v /home/ethan/文档/project/ocr:/root tensorflow/tensorflow:1.8.0-devel-gpu-py3 /bin/bash
 docker exec -it tf18 /bin/bash

2.安装ssh服务

2.1 下载并按照apt-transport-https_1.2.32ubuntu0.2_amd64.deb包

ps: 版本可以不一样

dpkg -i install apt-transport-https_1.2.32ubuntu0.2_amd64.deb 

2.2 换源

echo -e "deb http://mirrors.aliyun.com/ubuntu/ xenial main restricted universe multiverse\ndeb http://mirrors.aliyun.com/ubuntu/ xenial-security main restricted universe multiverse\ndeb http://mirrors.aliyun.com/ubuntu/ xenial-updates main restricted universe multiverse\ndeb http://mirrors.aliyun.com/ubuntu/ xenial-proposed main restricted universe multiverse\ndeb http://mirrors.aliyun.com/ubuntu/ xenial-backports main restricted universe multiverse\ndeb-src http://mirrors.aliyun.com/ubuntu/ xenial main restricted universe multiverse\ndeb-src http://mirrors.aliyun.com/ubuntu/ xenial-security main restricted universe multiverse\ndeb-src http://mirrors.aliyun.com/ubuntu/ xenial-updates main restricted universe multiverse\ndeb-src http://mirrors.aliyun.com/ubuntu/ xenial-proposed main restricted universe multiverse\ndeb-src http://mirrors.aliyun.com/ubuntu/ xenial-backports main restricted universe multiverse\n" >> sources.list
cp source.list /etc/apt/sources.list
apt update

2.3 安装openssh-server, vim

apt install openssh-server
apt install vim

2.4. 修改密码

passwd

2.5 配置ssh

vi /etc/ssh/sshd_config
 Subsystem      sftp    /usr/libexec/openssh/sftp-server

注释掉以下这一行

Subsystem      sftp    /usr/libexec/openssh/sftp-server

后面加上这几行

在这里插入代码片Subsystem       sftp    internal-sftp
PubkeyAuthentication yes #启用公钥私钥配对认证方式
AuthorizedKeysFile .ssh/authorized_keys #公钥文件路径
PermitRootLogin yes #root能使用ssh登录

2.6 启动ssh-server

ps -e | grep ssh
/etc/init.d/ssh start

然后离开容器

3. ssh 客户端链接上docker

查看容器端口映射,这里用的是6100端口作为映射

ethan@ethan-B360M-D3V:~$ docker port tf18
22/tcp -> 0.0.0.0:6100
8888/tcp -> 0.0.0.0:6000

链接ssh

ssh root@127.0.0.1 -p 6100

4. 如果nvcc --version提示没有找到命令，那么需要把cuda添加上环境变量

vim ~/.bashrc

加入cuda的安装地址

export PATH=$PATH:/usr/local/cuda/bin

更新环境变量

source ~/.bashrc

All done~ enjoy!