We are happy to announce that Charmed Kubeflow 1.8 is now available in Beta. Kubeflow is a foundational part of the MLOps ecosystem that has been evolving over the years. With Charmed Kubeflow 1.8, users will benefit from the ability to run serverless workloads and perform model inference regardless of the machine learning framework they use.
We’re looking for data scientists, ML engineers and developers to take the Beta release for a drive and share their feedback! Our blog is available if you want to read more.
Join us live: tech talk on Charmed Kubeflow 1.8
By the way, if you’ve got no plans tonight, Canonical’s MLOps will host a live stream about Charmed Kubeflow 1.7 Beta. Together with Noha Ihab, we will continue the tradition that started with the previous releases.
The latest release: Kubeflow 1.8 and how our distribution handles it
Key features covered in Charmed Kubeflow 1.8
The differences between the upstream release and Canonical’s Charmed Kubeflow
Please be mindful that this is not a stable version, so there is always a risk that something might go wrong. Save your work to proceed with caution. If you encounter any difficulties, Canonical’s MLOps team is here to hear your feedback and help you out. Since this is a Beta version, Canonical does not recommend running or upgrading it on any production environment.
If you would like to talk more about generative AI, Large language models (LLMs) and how to leverage open source MLOps for your projects, meet us around the globe during Canonical AI Roadshow.
Canonical AI Roadshow is a series of events that will highlight generative AI use cases powered by open source software. The roadshow will take us around the globe between mid-September and mid-November to talk about the latest innovations from the industry, demo some of the most exciting use cases from various industries, including financial service, telco and oil and gas, answer questions about AI, MLOps, Big data and more… We will stop in:
Charmed Kubeflow 1.8 beta has the following known issues:
kubeflow-profiles unit stuck in error and kubeflow-dashboard unit in waiting status
If the kubeflow-profiles unit stays in error status with the message hook failed: "install", this is possibly due to a conflict with istio when applying the authorization policy manifest. As a consequence, you will also see kubeflow-dashboard unit in waiting status. To check that this issue is affecting you, run:
This is a temporary issue due to the version of kubeflow-profiles charm that is used in 1.8/beta. Updating the version of Profiles is pending on #309. The fix is expected in a few days.
Seldon is not functioning correctly
When trying to run a Seldon deployment, it will fail due to incompatibility with k8s v1.25, this is because we haven’t yet updated Seldon from v1.15.
Associated issue
#211 - This issue will be fixed soon and released in 1.8/stable.
Katib experiments in Failed status
If you create a Katib experiment, you might see it in Failed status. This is because it fails to report the experiment results. If you run into this issue, you will see the experiment pods in Error status.
Associated issue
#108 - This issue is being tackled as part of the 1.8/stable release.
