Recommenders: [BUG] conda update conda error - PackageNotInstalledError: Package is not installed in prefix

[roalexan]

Description

I'm getting an error when running the first setup instruction on my Linux (Ubuntu) DSVM.

In which platform does it happen?

Azure Data Science Virtual Machine (Ubuntu)

How do we replicate the issue?

I created a Linux (Ubuntu) DSVM, set up port forwarding, and opened a browser to log into to Jupyter using my VM username/password. I then ran the first step from the command line to setup the dependencies, from the instructions here: https://github.com/Microsoft/Recommenders/blob/master/SETUP.md#dependencies-setup:

conda update conda

This produced the following console output error:

PackageNotInstalledError: Package is not installed in prefix.
prefix: /data/anaconda/envs/py35
package name: conda

Expected behavior (i.e. solution)

Should be no error

Other Comments

I could run the next line, 'conda update anaconda', so it looks like conda is indeed there.

[miguelgfierro]

Maybe this could help https://groups.google.com/a/continuum.io/forum/m/#!topic/anaconda/OZ77eyoFwiw it says to add the name of the base env.

@paulshealy1 any guidance on this?

[paulshealy1]

The _conda_ package is only available in the Anaconda root environment, not all environments. The Ubuntu DSVM has the py35 environment as the default one, and the root environment is py27. This update step should be _conda update conda -n root_ for the Ubuntu DSVM.

[roalexan]

I suggest that this issue is resolved by changing the initial README.md and associated SETUP.md to show the setup steps for each environment - 1) local, 2) Ubuntu DSVM, 3) etc...

[yueguoguo]

Resolved in #563

[yts61]

how was it solved?

[gramhagen]

@yts61 the setup instructions were changed to
conda update conda -n root
I believe you could also use
conda update -n base conda