Open Data Hub logo
DOCS
BLOG
COMMUNITY
GITHUB
GET STARTED

Upgrade Installation(v2.0,2.1->v2.2)

disclaimer

Version 2.2 represents an alpha release, exclusively accessible via the "fast" channel. Subsequent releases will transition to the "fast" channel once the new operator attains greater stability.

For installation steps of the old (version 1.X, stable), see quick installation of the 1.X version.

For installation steps of the old (version 2.0, 2.1, fast), see quick installation of the 2.X version.

Pre-requisites

Below information is only appliable for Open Data Hub Operator v2.0.0 and forth release.

Installing Open Data Hub requires OpenShift Container Platform version 4.10+. All screenshots and instructions are from OpenShift 4.12. For the purposes of this quick start, we used try.openshift.com on AWS.

Tutorials will require an OpenShift cluster with a minimum of 16 CPUS and 32GB of memory across all OpenShift worker nodes.

Installing v2.2 Open Data Hub Operator

The Open Data Hub operator is available for deployment in the OpenShift OperatorHub as a Community Operators. You can install it from the OpenShift web console by following the steps below:

  1. From the OpenShift web console, log in as a user with cluster-admin privileges. For a developer installation from try.openshift.com, the kubeadmin user will work. Log in to OpenShift

  2. On the lefthand bar, from Operators -> OperatorHub,

    • filter for Open Data Hub Operator.
    • select AI/Machine Learning and look for the icon for Open Data Hub Operator.

    OperatorHub

  3. Click Continue in the "Show community Operator" dialog if it pops out. Click the Install button to install the Open Data Hub operator. Install

  4. The subscription creation view will offer a few options including Update Channel, make sure the fast channel is selected. Click Install to deploy the opendatahub operator into the openshift-operators namespace. Select Channel

  5. To view the status of the Open Data Hub operator installation, find the Open Data Hub Operator under Operators -> Installed Operators. It might take a couple of minutes to show, but once the Status field displays Succeeded, you can proceed to create a DataScienceCluster instance Installed Operators

  6. [Optional] To customize the "applications namespace", update the "default" DSCInitialization instance from either "Form view" or "YAML view", see screenshot for example: Select NS

Create a DataScienceCluster instance

  1. Click on the Open Data Hub Operator from Installed Operators page to bring up the details for the version that is currently installed. Open Data Hub Operator2.2

  2. Two ways to create DataScienceCluster instance:

    • Click Create DataScienceCluster button from the top warning dialog DataScienceCluster required(Create a DataScienceCluster instance to use this Operator.)
    • Click tab Data Science Cluster then click Create DataScienceCluster button

    They both lead to a new view called "Create DataScienceCluster". By default, namespace/project opendatahub is used to host all applications.

  3. In the view of "Create DataScienceCluster", user can create DataScienceCluster instance in two ways with components fields.

    • Configure via "Form view":

      1. fill in Name field
      2. in the components section, by clicking > it expands currently supported core components. Check the set of components enabled by default and tick/untick the box in each component section to tailor the selection.

      Create DSC2 v2.3

    • Configure via "YAML view":

      1. write config in YAML format
      2. get detail schema by expanding righthand sidebar Create DSC1 v2.3
      3. read ODH Core components to get the full list of supported components

  4. Click Create button to finalize creation process in seconds.

  5. Verify the installation by viewing the project workload. Click Home then Projects, select "opendatahub" project, in the Workloads tab to view enabled compoenents. These should be running. Verify Status

Dependencies

  • to use "kserve" component, users are required to install two operators via OperatorHub before enable it in DataScienceCluster CR

    1. Red Hat OpenShift Serverless Operator from "Red Hat" catalog.
    2. Red Hat OpenShift Service Mesh Operator from "Red Hat" catalog.

  • to use "datasciencepipeline" component, users are required to install one operator via OperatorHub before enable it in DataScienceCluster CR

    1. Red Hat OpenShift Pipelines Operator from "Red Hat" catalog.

  • to use "distributedworkloads" component, users are required to install one operator via OperatorHub before enable it in DataScienceCluster CR

    1. CodeFlare Operator from "Community" catalog.

  • to use "modelmesh" component, users are required to install one operator via OperatorHub before enable it in DataScienceCluster CR

    1. Prometheus Operator from "Community" catalog.

Limitation

We offer a feature that allows users to configure the namespace for their application. By default, the ODH operator utilizes the opendatahub namespace. However, users have the flexibility to opt for a different namespace of their choice. This can be accomplished by modifying the DSCInitialization instance with the .spec.applicationsNamespace field.

There are two scenarios in which this feature can be utilized:

  • Assigning a New Namespace: Users can set a new namespace using DSCInitialization instance before creating the DataScienceCluster instance.
  • Switching to a New Namespace: Users have the option to switch to a new namespace after resources have already been established in the application's current namespace. It's important to note that in this scenario, only specific resources (such as deployments, configmaps, networkpolicies, roles, rolebindings, secrets etc) will be removed from the old application namespace during cleanup. For namespace-scoped CRD instances, users will be responsible to cleanup themselves.

Upgrade from v2.0/v2.1 to v2.2 and forth release

Cleanup resource in cluster

To upgrade, follow these steps:

  • Disable the component(s) in your DataScienceCluster instance.
  • Delete both the DataScienceCluster instance and DSCInitialization instance.
  • Click "uninstall" Open Data Hub operator.
  • Delete 2 CRD
    • Click Administration then CustomResourceDefinitions, search for DSCInitialization
    • Under Latest version column, if shows v1alpha1, click "Edit" button(3 dots) on the right, and Delete CustomResourceDefinition
    • Repeat the same procedure on DataScienceCluster

After completing these steps, please refer to the installation guide to proceed with a clean installation of the v2.2+ operator.

API change

  • when create or update DataScienceCluster instance, API version has been upgraded from v1alpha1 to v1

    • schema in v1alpha1 API to enabled each component: .spec.components.[component_name].enabled: true to disable each component: .spec.components.[component_name].enabled: false

    • schema in v1 API to enabled each component: .spec.components.[component_name].managementState: Managed to disable each component: .spec.components.[component_name].managementState: Removed

    Example for default DataScienceCluster instance in v2.2

    apiVersion: datasciencecluster.opendatahub.io/v1
kind: DataScienceCluster
metadata:
   name: default
spec:
   components:
      codeflare:
         managementState: Removed
      dashboard:
         managementState: Managed
      datasciencepipelines:
         managementState: Managed
      kserve:
         managementState: Removed
      modelmeshserving:
         managementState: Managed
      ray:
         managementState: Removed
      workbenches:
         managementState: Managed