Try the new Charmed Kubeflow 1.11-ubuntu2 candidate!

We’re pleased to share that the Charmed Kubeflow 1.11-ubuntu2 candidate release is now available for testing!

Kubeflow remains a central pillar of Canonical’s MLOps portfolio, keeping up to date with the latest developments across the open-source ecosystem. This candidate release (1.11-ubuntu2) delivers significant new capabilities, major component updates, improved service mesh flexibility, and enhanced infrastructure modularity.

What’s New in 1.11-ubuntu2?

  • Spark Integration. This new release promotes the integration between Kubeflow and Spark solution as stable, streamlining big data and data processing workflows alongside your ML pipelines.

  • Service Mesh & Ambient Support. We now support integration with Ambient mesh. This will secure and isolate your deployments, representing a substantial step forward in hardening Kubeflow deployments.

  • Core Component Upgrades. The new release also features important upgrade of the following components

    • Kubeflow Pipelines (KFP) upgraded to version 2.16.

    • KServe upgraded to version 0.17

      This brings the latest security patches, bug fixes and new features into our Charmed Kubeflow distribution.

  • Preview Modular Terraform Architecture. This release also provides a preview for a major refactor of our Terraform modules that deliver cleaner separation, modularized components, and flexible integrations across the platform and products provided by Canonical. In the current version we both support the historical Terraform modules (provided in the 1.11 release for backward compatibility) and the new Terraform modules that will become the standard way of deploying Charmed Kubeflow for future major track releases.

Note that this version of Charmed Kubeflow discontinues testing on some legacy architectures. Specifically Kubernetes clusters below version 1.34 running on Ubuntu 20.04 and on NVIDIA Tesla V100 GPUs

See our draft release notes for further details on component changes and fixes.

Let’s build it together!

This candidate release gives you early access to cutting-edge MLOps innovations, along with an opportunity to influence the future of Charmed Kubeflow. Your feedback is crucial to improving Charmed Kubeflow:

Please note: Candidate releases are intended for testing and validation, not yet for production use. We will allow some time for community testing and feedback before promoting these artifacts to stable.

1 Like