We have a shared Jupyterhub instance for researchers as a service. It is available from a that is also pointing to instances dedicated to teaching and individual researcher instances. This portal and those instances can also be accessed from outside of UC. Request for access should be sent to the eResearch support team via the button below and stating that you want to use the eResearch jupyterhub in the "Other information" box. Other optional fields can be left blank.
Access
Once your access request has been approved you can head to and click on the link called "eResearch_hub".聽 You will land on a login page where you can use your UC username and password.
Using Jupyterhub
The hub is based on the latest release of anaconda on an ubuntu VM instance. The supporting server has 32 vCPUs and 94 GB of RAM.
Python
The Python kernel is available using the standard JupyterLab interface and terminal.
There is ample space for you to install the packages you want in your home directory. However, if the install proves difficult or you think a package would be beneficial to many users you can send a request for the package to be installed instance wide. In case of installation difficulty, you can also request assistance just for yourself.
Python packages can be normally installed in your user space by entering the command:
pip install --user the_package_I_want
from the JupyterLab terminal. As mentioned earlier do not hesitate to contact eResearch staff in case of difficulties.
Julia
Julia is also installed but we are unable to install a jupyter kernel usable by every user. If you want to use Julia in notebook you will need to install the IJulia kernel yourself. To do so follow these instructions:
- Login in Jupyterhub
- Open a JupyterLab terminal (under other in the JupyterLab launcher)
- Start Julia by typing "julia" and then enter
- Once at the Julia prompt type the following
- Using Pkg
- 笔办驳.补诲诲(鈥泪闯耻濒颈补鈥)
- Quit Julia and the terminal
- Logout of the hub and login again
Note that Julia packages cannot be shared system wide (this is a limitation of current packaging technology in Julia) so the assistance will only be for private installs in that case.
File Transfer
For large files the JupyterLab download interface may not work. Files can be transferred from another machine on the campus network via direct access. Either on the command line with scp, sftp, or the like. Or with graphical interfaces such as FileZilla (in the case of a windows client). In this case the address of the machine is
132.181.102.4
and if you are using FileZilla, do not forget to set the port to 22 (sftp). Use your normal user code and password when prompted.
Individual Instances
If you outgrow the shared hub, or your compute is big enough to disrupt other users, you can request (or being offered) your own jupyterhub instance on the RCC. We can make it feature matching with the shared hub and you can customise it to your heart content.
