A repository for Open Data Hub components Kustomize manifests.
- Website: https://opendatahub.io
- Documentation: https://opendatahub.io/docs.html
- Mailing lists: https://opendatahub.io/community.html
- Community meetings: https://gitlab.com/opendatahub/opendatahub-community
Open Data Hub is an end-to-end AI/ML platform on top of OpenShift Container Platform that provides a core set of integrated components to support end end-to-end MLOps workflow for Data Scientists and Engineers. The components currently available as part of the ODH Core deployment are:
Previous versions of ODH relied on JupyterHub for managing the lifecycle of Jupyter notebook pods. Starting with Open Data Hub v1.4, we will be relying on ODH Notebook Controller for controlling the lifecycle of user Juptyer notebook pods with ODH Dashboard as the frontend UI. When ODH v1.5 is released, we will be moving jupyterhub-odh to our ODH Contrib organization and officially ending long term support.
Any components that were removed with the update to ODH 1.4 have been relocated to the ODH Contrib organization under the odh-contrib-manifests repo. You can reference the odh-contrib kfdef as a reference on how to deploy any of the odh-contrib-manifests components
| Manifest Version | ODH Dashboard | ODH Notebook Controller | ODH Notebook Images | Data Science Pipelines | ModelMesh |
|---|---|---|---|---|---|
| master | v2.5.2 | v1.6 | v1.3.1 | v1.2.1 | v0.9.0 |
| v1.4.1 | v2.5.2 | v1.6 | v1.3.1 | v1.2.1 | v0.9.0 |
| v1.4.0 | v2.2.1 | v1.6 | N/A | v1.2.1 | v0.9.0 |
We are relying on Kustomize v3, kfctl and Open Data Hub Operator for deployment.
The two ways to deploy are:
- Following Getting Started guide using a KFDef from this repository as the custom resource.
- Using
kfctland follow the documentation at Kubeflow.org. The only change is to use this repository instead of Kubeflow manifests.
To submit issues please file a GitHub issue in odh-manifests