Migrating Cloud Pak for Data data from Red Hat OpenShift Version 3.11

If your existing Cloud Pak for Data installation is running on Red Hat® OpenShift® Container Platform Version 3.11, a cluster or project administrator must migrate your installation to a supported version of Red Hat OpenShift Container Platform.

1. Back up your entire Cloud Pak for Data system. See Backing up and restoring your deployment for details.
2. To avoid data inconsistencies, disable the route to the platform to ensure that users do not access Cloud Pak for Data during the migration:
   1. Log in to your Red Hat OpenShift cluster as a cluster administrator:

      oc login OpenShift_URL:port

   2. Run the following command to get the list of routes:

      oc get route

   3. Copy and save the route that you are currently using for your Cloud Pak for Data.
      Important: Ensure that you save this information in a safe place. If you lose or forget the route, you must direct your users to a new URL.
   4. Run the following command to delete the route:

      oc delete route Route_name

   5. If you need to restore the route to the platform, retrieve the information that you saved before you deleted the route and run the following commands:
      1. Run the following command to re-create the route:

         oc create route passthrough Route_name --service=ibm-nginx-svc

      2. Run the following command to annotate the route:

         oc annotate route Route_name haproxy.router.openshift.io/balance='roundrobin'

3. Stop all running jobs on Cloud Pak for Data, including environment runtimes, automated discovery, and data stage jobs.
4. On the node where you plan to run the migration job, ensure you have sufficient CPU, memory, and disk resource to avoid a job failure.
5. Using the cpd-cli export-import command-line interface, export your Cloud Pak for Data control plane data from the OpenShift Version 3.11 cluster. Then, import that data into the new cluster.
   For more information, see Migrating metadata between Cloud Pak for Data installations.
6. For each service, migrate data:

   Service	Steps
   Anaconda Repository for IBM Cloud Pak® for Data	Not supported.
   Analytics Engine Powered by Apache Spark	Not supported.
   Cognos® Dashboards	In the Cognos Dashboards console on the old cluster, open the project and download the dashboard as a JSON file. On the new cluster, create a new dashboard and import the JSON file.
   Data Refinery	Does not use permanent storage, so migration is inapplicable.
   Data Virtualization	Not supported.
   DataStage®	Not supported.
   Db2®	Back up data from the old cluster and restore it to the new cluster. See Backing up and restoring Db2.
   Db2 Data Gate	Not supported.
   Db2 Data Management Console	Not supported.
   Db2 Event Store	Back up data from the old cluster and restore it to the new cluster. See Backing up and restoring Db2 Event Store.
   Db2 Warehouse	Back up data from the old cluster and restore it to the new cluster. See Backing up and restoring Db2 Warehouse.
   Execution Engine for Apache Hadoop	See Exporting and importing projects.
   Guardium® External S-TAP	Not supported.
   Informix® Enterprise Edition	Not supported.
   MongoDB	See Migrating MongoDB Enterprise Advanced data.
   Open Data for Industries	Back up data from the old cluster and restore it to the new cluster. See Backing up and restoring Open Data for Industries.
   SPSS® Modeler	Export data assets from the old cluster by using the Data Asset Export node. Import data assets to the new cluster by using the Data Asset node. When you create a new SPSS Modeler flow, you can import an SPSS Modeler stream (.str) from the old SPSS Modeler to start a migration wizard. The wizard locates the correct data assets when you import the flow into your project. You can also use the same process as Watson Studio to export and import data assets at the project level.
   Watson Knowledge Catalog	Export each root category (include subcategories even though only names are exported, not the content) from the old cluster. Export each asset type (such as term or data class) from the old cluster. Import the assets and categories in reverse order (limit each export file to 1000 items) to the new cluster. Publish all of the assets after each import to the new cluster. Add artifacts back to the subcategories by matching the subcategory IDs.
   Watson Machine Learning	See cpdctl asset export and cpdctl asset import in Managing AI lifecycle with the CPDCTL.
   Watson Machine Learning Accelerator	Not supported.
   Watson OpenScale	Not supported.
   Watson Studio (includes SPSS Modeler and Decision Optimization)	Export the projects from the old cluster. Import the projects to the new cluster. If you enabled Git integration, you can use the CICD pipeline.

   Ensure you also migrate all metadata such as users, privileges, and instance user mapping. If you have service instances, migrate the instance metadata and identify provider data specific to the instance in providers like LDAP.
   During the migration, pay attention to whether:

   • The directory structure changed.
   • The directory privilege or owner changed.
   • The database table schema changed.
   • The database or identify provider version changed.
7. Verify whether your migration succeeded:
   • Test whether a migrated instance user can successfully log in.
   • Test whether a migrated user can read or edit the data that they created (if they have the privilege).
   • Check whether the migrated data is complete and that no data was lost.
   • Check whether the migrated metadata and migrated data can function well.
   • Optional: Check whether Cloud Pak for Data functions well across components.