Ah yes, that's all making a lot of sense, thank you very much.
If I ever make any progress down that path, I'll report back!
On Friday, June 2, 2017 at 8:56:07 AM UTC+10, Matthias Bussonnier wrote:
>
> Well, if you are running code in a kernel then you **need** both your
> data store and computation to be on the same machine.
> Otherwise you have no way of having open('filename') to work.
>
> So if hosing on a server is ok with you, your best chance is a shared
> filesystem where at least user data is mounted on the server.
> That's (IMHO) the least worse situation.
>
> You don't need Jupyter and Python libraries to be on users machines
> these can be system wide on your server(s), but for each user you
> mount their home directory. I would suggest [1] as a read, it's
> starting to show its age, but shows you what can be done.
>
> Any way as soon as you need multiple users you want to look at
> JupyterHub because the single-user jupyter server have this assumption
> : only one user is using it. And if you try to avoid this you will get
> into painful situations.
>
> The things you have to figure out are how you want to architecture
> your shared filesystem, mounting the server on machines or have
> machine being nfs-servers mounting on the main jupyter-hub server.
> --
> M
> 1:
> https://developer.rackspace.com/blog/deploying-jupyterhub-for-education/
>
>
--
You received this message because you are subscribed to the Google Groups
"Project Jupyter" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/jupyter/69a54408-081f-4574-882d-dc39db49b944%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
