I suppose you have:
- A Linux/Unix based Workstation
- A static LAN IP, like 184.108.40.206 (usually provided by CampusNet Service Center)
- conda environment (Python 3.4 or greater)
On Debian/Ubuntu, issue
sudo apt-get install npm nodejs-legacyto install.
- Jupyter Notebook (usually it is preinstalled in conda root environment)
You can generate a default configuration file with
jupyterhub --generate-config, then edit it following the instructions on official site. Here, if you are going to deploy JupyterHub for a class of students or your research group, my config file can be a good reference:
Save this script as
jupyterhub_config.py in your home directory, then read the comments in lines carefully.
sudo chmod 444 ~/jupyterhub_config.py to make config file read only. Then save next script as
jupyterhub.service in your home directory and modify it following comments in lines.
After you complete modification, issue following commands to install service:
Have the JupyterHub start automatically on boot if you like:
If JupyterHub doesn’t seem to start properly, issue
journalctl --boot -u jupyterhub.service, you can view the log data to help figure out what the problem is. If you want to follow the latest logs from JupyterHub you can use
journalctl -f -u jupyterhub.service.
Now you run a multi-user hub for the single-user Jupyter Notebook servers. Visit 220.127.116.11:8080, login with system user and enjoy your work. :)