Anaconda on Windows subsystem for Linux (WSL) uses the "wrong" anaconda when creating a Jupyter Notebook

Question

I am using Windows subsystem for Linux WSL with the Ubuntu App (Ubuntu 20.04 LTS). I have installed Anaconda (Anaconda3-2020.11-Linux-x86_64) on my Windows 10 Education 1909. I have Jupyter notebook, and can run this in the Firefox on my computer and it seams to be working properly. However when I try to install packages such as:

Ubuntu console: pip install scrapy

Then the Jupyter notebook can not find it.

Jupyter notebook: import scrapy

I am currently working in the base environment, but I believe that Jupyter is actually running python from a different source (I also have Anaconda on my Windows). I confirmed this by running: import sys and sys.version both in the WSL and in the Jupyter notebook.

Jupyter notebook returns: '3.6.6 |Anaconda, Inc.| (default, Oct 9 2018, 12:34:16) \n[GCC 7.3.0]'

WSL returns: '3.8.5 (default, Sep 4 2020, 07:30:14) \n[GCC 7.3.0]' confirming that the "wrong python is used".

I am hesitant to delete my Windows Anaconda since I have my precious environments all set up there and are using them constantly.

The spessific package that forces me to linux can be found at "http://www.nupack.org/downloads" but requires registration for downloads.

I do not have Anaconda or python in my Windows environment variables.

I would be happy If I either would know where to install my packages (as long as they are in Linux), or if someone knows how to force Jupyter to use the Anaconda from WSL.

Answer 1

Thanks to Panagiotis Kanavos I found out that I had both Anaconda3 and Miniconda3 installed and that the WSL command line used the miniconda3 version while Jupiter Notebook used Anaconda3.

There is probably a way of specifying which version to use but for me I simply deleted Miniconda and it now works.