If you want to run Jupyter Notebook on the cluster and work in this from your desktop the following may help.
As you do not have write access to the software volume you will need to do a local install, please look at the guidelines here for doing a local pip install.
Do not use the standard system version of anaconda. Rather activate one of the cluster versions inside of an interactive job. An example:
sintx --ntasks=4 module load python/miniconda3-py39 pip install --user jupyter exit
Now jupyter is installed, you only need to do the above once. To use it start an interactive job, load the module and start jupyter. NB, do not run jupyter on the head node, use an interactive job!
sintx --ntasks=4 module load python/miniconda3-py39 jupyter notebook --no-browser
Make a note of the published URL to access jupyter, it will most likely be 8888 but may be different if this port is already being used by someone:
http://localhost:8888/?token=3aac4c26e1d82als723f73be5fe0347easf7ea59dbcda
On your desktop activate a port forwarding session in a command prompt:
ssh -N -L localhost:8889:localhost:8888 myhpcusername@srvcnthpc102.uct.ac.za
Ignore the key error, enter yes, then enter your password. No further responses are required, just leave this running.
Here 8888 is the port from the published URL above and srvcnthpc103 is the worker node where you’re running your interactive job (hint: look at the prompt to get the name of the worker node).
Open a browser on your desktop and paste the URL you got from jupyter remembering to change the port number to the listening port on your desktop, 8889:
http://localhost:8889/?token=574e786ada1574f5ec97dbb219dd157b223789519cd3e6fa
You should now see notebooks running in your browser:
If you get an error in your browser make sure that you remembered to change the port number in the URL from 8888 to 8889
Once done you should close down you command line ssh session on your desktop, exit the jupyter notebook in your interactive job with ctrl-c, then type exit to terminate your job.