1 | initial version |
1/ with "sage -n jupyter --ip='*'" is it possible to have 16 or so students connecting to the same server and working on similar worksheets (as it seemed to work rather smoothly for sagenb)?
Jupyter is meant to use unix accounts as user accounts. So, theoretically, you are supposed to create 16 unix users on your system. Having access to such an account implies that users can do many things (they can execure arbitrary Python code on your system with classical user rights, this include looking at various files).
2/ if not, is there another way to have sage use jupyter notebook for multi users?
It is possible to just use only one guest user, and let all students connect to that account. I did that already, with each student usng her own dedcated subirectory (pre-filled with worksheets). It works well as long as the students are trustworthy, and will not play with the other student's directory.
3/ if not, is an install of jupyterhub or sagemathcloud reasonnable with no particular sysadmin skill?
You should give a try and see if you are able to do that, since "having no particular sysadmin skil" really depends on who is speaking... Regarding jupyterhub, there is plenty of docs, and step-by-step guides out there.
4/ if not is there anything else I should try?
Apparenty, the "isolation by container" (e.g. docker) is a current trend : one docker instance per user (it is a bit heavy though).
5/ if not, do some of you have experience of sagenb under sage 8.3 usability?
I would not recommend to start working with sagenb, since it is already deprecated, and might not pass the Python3 migration, so you are likely to invest a lot of time for a non-viable mid-term solution.
Installing sagemath on all computers is not an option for me at the moment.
Let me witness sucessful use of Sage Debian Live (see https://sagedebianlive.metelu.net/). This is an easy-to-replicate live USB with Sage (and all optional packages, and LaTeX, and R, etc) installed, that ca be run without having to install anything on the computers. [disclaimer: i am the developper of SDL, though i got external reports of such experience].
If students can not have their own key, it is possible to use such a key to serve public notebooks as in 2/, without taking the risk of exposing you own private data. I might develop a jupyterhub (with Sage) automatic deployment icon in a future version of SDL (but it is currently not on the top of my todo list).