Jupyter 远程访问
jupyter 远程访问的工作方法是,在本地通过浏览器打开jupyter,但是代码和服务运行在远程集群中。
集群设置
首先需要确保集群中安装有python和jupyter。
在集群 /opt/intel/2019 目录下,已经安装好python3和jupyter。将其添加入环境变量中
export PATH=/opt/intel/2019/intelpython3/bin:${PATH}
随后对python路径进行检查,可以看出此时已经在调用intel python。
$ which python /opt/intel/2019/intelpython3/bin/python
随后用以下命令对 jupyter 进行设置,设置 jupyter 登录密码。
$ jupyter notebook --generate-config $ jupyter notebook password Enter password: **** Verify password: ****
此时在个人目录下就生成了 jupyter 配置文件 jupyter_notebook_config.py
。
对配置文件进行修改,增加以下设置
c.NotebookApp.ip='*' # 星号代表任意ip,这个跟mysql的权限设置一样,所以说知识是互通的 c.NotebookApp.allow_remote_access = True # 允许远程连接 c.NotebookApp.password = u'sha1:xxx' # 复制密码在此 c.NotebookApp.open_browser = False # 不自动打开浏览器 c.NotebookApp.port = 6666 # 指定端口, 访问时使用
注意在设置端口时给定一个较大的整数,22 或 24 等常用端口时不可以的。
此时即可在命令行中启动jupyter notebook服务,会显示以下内容:
$ jupyter notebook [W 14:01:28.051 NotebookApp] Collisions detected in /home/lilongxiang/.jupyter/jupyter_notebook_config.py and /home/lilongxiang/.jupyter/jupyter_notebook_config.json config files. /home/lilongxiang/.jupyter/jupyter_notebook_config.json has higher priority: { "NotebookApp": { "password": "'sha1:llx891209' ignored, using 'sha1:a221858aea95:923e3211599640af335a06052fcea5818f118621'" } } [W 14:01:28.054 NotebookApp] Config option `allow_remote_access` not recognized by `NotebookApp`. Did you mean one of: `allow_credentials, allow_root`? [W 14:01:28.206 NotebookApp] WARNING: The notebook server is listening on all IP addresses and not using encryption. This is not recommended. [I 14:01:28.213 NotebookApp] Serving notebooks from local directory: /home/lilongxiang [I 14:01:28.213 NotebookApp] 0 active kernels [I 14:01:28.213 NotebookApp] The Jupyter Notebook is running at: [I 14:01:28.213 NotebookApp] http://[all ip addresses on your system]:1224/ [I 14:01:28.213 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation). [I 14:01:41.532 NotebookApp] 302 GET / (10.155.11.18) 0.50ms
本地登录
注意上面内容中,提示可以使用任意ip地址登录Notebook服务。在本地网页浏览器中,以集群IP地址代替。如集群IP为10.166.15.60,在本地浏览器中输入
https://10.166.15.60:1224/
随后即可登录集群jupyter notebook服务进行操作。