Jupyter lab - User issues installing packages and plugins. Is there a. workaround?

Hi All,

I have installed Jupyter lab, It is installed as a shared package and loaded as a module for OOD.

  1. The interface works but it looks like because it is installed system wide Users don’t have the perms to install their packages. I don’t see any variables (that work) to set that allows install or running of Jupyter Lab system wide as a module but allow users to save their python stuff and plugins in their home directory? I know I could solve this with a cond env or perhaps a container for users but are there any ways to make the way I have it setup work.

  2. The other thing I would like advise on is, is it a bad thing to let users manage their own install of:
    a. Jupyter plugins?
    b. Python packages?

  3. Is it better to manage the above as a sysadmin centrally or as users. Risks?

Thanks

I think there’s another question around about Jupyter plugins that I’m not so sure on. I don’t know if or how users could supply additional plugins for Jupyter.

For regular python packages they’d like to use in their notebooks is simple though - users should create kernels that hold these packages. Google searching jupyter conda kernel should provide some results for the same.

Thx, Jeff for trying to answer the query. I would love to know how others have implemented this as my approach is somewhat locked down , i.e. users cannot install packages locally. The advantage is that because most of it is central, the core stuff is managed by root. A blend of both is what I am looking for from folks. So Anyone else ??

Users can’t install packages in their own $HOME?