This project provides a Python-based CLI tool to create and use custom images in Cloud Pak for Data, specifically for:
- Watson Studio
- Jupyter notebook and Jupyter Lab environments in Python
- Rstudio environments
- Watson Machine Learning
- Python online deployment environments
It streamlines the process of creating a custom image for Cloud Pak for Data runtimes by wrapping the key steps into CLI commands, covering:
- gather the needed information such as base image names
- build one custom image per base image, in batch
- push each custom image to a target container registry
- register a custom image with the corresponding Cloud Pak for Data service
Doc: https://ibm.github.io/CPD-Custom-Image-CLI/
0 - For 4.5.x, only the latest version of doc is supported (both WS and WML) because I haven't found a way to grab historical refreshes. As of 4.5.2 (2022 Aug release) you can always use --cpd-version 4.5.2 even if you are on CPD 4.5.1 or 4.5.0, because the doc is used to get the compatible runtime names which fortunately hasn't changed in these three versions. Once the runtime files are located, the actual image name the tool extracts and uses reflect your system version, regardless of what you put in --cpd-version.
1 - The current release is implemented and tested only for Cloud Pak for Data 4.0.x. If you are on 3.5.x, it does not apply because the way to fetch base images is changed in 4.0.x. If you are on 4.5.x, this CLI may work.
2 - Gathering information and registering custom images are specific to Cloud Pak for Data, while building and pushing the custom image can be easily extended to support arbitrary situation.
3 - You need to know the key to access the Cloud Pak for Data base image repository (cp.icr.io), or have access to the OpenShift backend of any 4.0.x Cloud Pak for Data cluster you are building a custom image for from where you will be able to extract the secrets according to the Getting Started page in this doc.
4 - R Jupyter environment is not tested or supported by this CLI as usually R users prefer Rstudio. Contact the author if there is a need.
Watson Studio Doc:
Watson Machine Learning Doc:
DCO is suggested to be used. See here for details on how to do it.
@jkafrouni