Ubuntu 16.04 Linux update to linux-image-4.4.0-137-generic screws up conda virtual environments?


#1

Hello,

Anyone have comments experience in this?

Steps:
(1) Updated Ubuntu 16.04 Linux to linux-image-4.4.0-137-generic
(sudo apt-get install update & sudo apt-get install upgrade)

(2) While running Jupyter notebook (Python 3.6) after the update, a conda virtual environment stopped working.

(3) The same Jupyter notebook ran fine on Ubuntu 16.04 Linux Un-updated machine.

Environment managers such as conda are supposed to make the “Virtual Environment” impervious to OS and other updates.

Other than NOT updating OS, any suggestions to avoid this problem?

Thank you!


#2

Have you tried updating conda? Maybe there is a new version of conda or some of its dependencies for the new kernel you installed.


#3

If there any binary/library dependencies, you will need to update your virtual environment as well.

The real question is… what got updated when you ran that upgrade?


#4

What kind of error messages do you get? :slight_smile:


#5

Thanks, will first try updating conda.

Dumb Question: Is running Virtual Environment (conda, virtualenv, pipenv) inside Virtual Box a solution to these problems?


#6

(1) Deactivated Virtual Environment
(2) Updated Ubuntu 16.04 OS
(3) After the update, activated Virtual Environment
(4) Ran “conda env update -f environment.yml”


#7

While running the Jupyter notebook, the only Python module that gives error is mapping library “Basemap”, it’s built on Matplotlib; prior to the update Basemap drew maps without problems.


#8

It really depends. If you want to keep things stable, then a docker container (if that’s an option), could be a nice solution. You’d still have the same issues while running within a VM, unless the entire thing is running “natively” within the VM. (Without any of the virtual aspects.)


#9

Thanks, I think it’s time to move to a docker container!