We're running 11.2.4.1 with Jupyter Notebook server on podman.
We need several python libraries in their current versions (e.g. pycaret, sweetviz, ydata-profiling, pygwalker, dataprep, ...) but seem to run into problems with some rather old packages distributed in the original image.
- python is at 3.11.1
- OS of original image is ubuntu 18.4 LTS, released 4/2018
- spyder is 3.3.1, released 8/2018
Updating python packages runs into conflicts
- When we upgrade all packages before installing new ones the server_instance image does no longer start
- specifying specific versions for certain packages helps, e.g. spyder==3.3.1 and pixiedust==1.1.17, also jinja2==3.0.3 or traitlets==5.9.0
- with a longer list in requirements.txt pip takes some hours to complete, downloading several versions of several packages for requirements resolution with older packages
Questions:
- Is there a possibility to use a more current variant of the server_instance (any updates in the pipeline ?)
- Which packages need to stay on a specific versions for the image not to break on startup
Thanks for help
------------------------------
Thomas Strehl
------------------------------