Table of contents

Manage conda channels as a Watson Studio Local administrator

In the Conda Management panel, a Watson Studio Local administrator can customize a local repository to control which channels that Watson Studio Local users can pull packages from (using conda install). Alternatively, an administrator on the Watson Studio Local cluster machine can add or remove channels manually with the /user-home/_global_/config/conda/.condarc file.

A Watson Studio Local administrator can add a custom channel that Watson Studio Local users can pull packages from by completing the following steps:

  1. In the Admin Console, click the menu icon ( Menu icon) and click Conda management. The Conda Management window opens.
  2. In the Custom Conda Channel field, type in the new channel URL and click Add Channel.

    Conda Management

Connect a JFrog custom conda channel to Watson Studio Local

An administrator can add a JFrog custom conda channel to Watson Studio Local so that users can pull and install packages from JFrog.

Requirements:

  • A JFrog Artifactory server with a Python or R package.
  • A username in URL encode format and a user API key for the JFrog Artifactory server in the following format: https://<URL_ENCODED_USERNAME>:<USER_API_KEY>@<JFrog_Artifactory_Conda_Channel_URL>. You can copy this information from the JFrog Artifactory by clicking the conda channel in the Set Me Up section.

To connect the JFrog custom conda channel to Watson Studio Local:

  1. In the Admin Console, click the menu icon ( Menu icon) and click Conda management. The Conda Management window opens.
  2. In the Custom Conda Channel field, paste in the new JFrog channel URL and click Add Channel.

    Conda Management

  3. If you want Watson Studio Local users to only download packages from JFrog, remove the conda-forge and defaults channels.
  4. In a Jupyter notebook, test the connection by installing a package from the JFrog custom conda channel.

Set .condarc for all images on the Watson Studio Local cluster

To ensure a common conda policy is enforced across the organization, an administrator on the Watson Studio Local cluster machine can manually create a global .condarc file in the /user-home/_global_/config/conda/ directory to specify a common configuration for conda, including the channels that Watson Studio Local users can pull packages from. If the file is present, all images using anaconda (including built in images and custom images) will use the .condarc file. For details, see Administering a multi-user conda installation.

Commonly, an administrator adds local repositories as custom channels for the data scientists to install packages from. In Watson Studio Local, this can be achieved by adding the channel URL to the channels setting in the global .condarc file. If the administrator does not want Watson Studio Local to use the default channels and conda-forge, they can be removed from the .condarc file.

The administrator can also use the conda config command in either a Jupyter notebook or a terminal to modify the global .condarc file. The –file /user-home/_global_/config/conda/.condarc option should be used to make sure the correct .condarc file is updated.

Note that the global .condarc file will affect the behaviors of conda installations in custom images created in the current release, and not in a previous release.

Learn more