Dell ObjectScale Application 1.3.x Installation Guide for Red Hat OpenShift

Uninstall ObjectScale from the Red Hat OpenShift cluster

1. From the ObjectScale Portal user interface, delete the SupportAssist Connection, if configured.
   Click Administration > SupportAssist to view the current SupportAssist configuration.
2. Click Administration > Licensing and deactivate any ObjectScale licenses.
3. If you have not yet done so, delete the object stores displayed on the Object Stores page.

   Deleting the object stores cleans up all the allocated storage.

   Wait until all object store resources are removed before continuing. Deleting an object store can take up to a few minutes, depending on the size of the object store.
4. Remove all the ObjectScale application.

   helm uninstall objs -n <OBJECTSCALE_NAMESPACE>

   Wait for all the ObjectScale applications to be removed before proceeding with the next steps.

5. Run helm list --all-namespaces to get the list of installed components for ObjectScale.
6. Uninstall Postgres.

   kubectl delete postgrescluster postgres-ha -n <OBJECTSCALE_NAMESPACE> 
   helm uninstall -n <OBJECTSCALE_NAMESPACE> postgres
   kubectl get crd | grep postgres | awk '{print $1}' | xargs kubectl delete crd

7. Start removing the CSI Bare Metal components by first deleting the PVCs bound for ObjectScale and its components.
   1. List all the PVCs used by ObjectScale:

      kubectl -n <OBJECTSCALE_NAMESPACE> get pvc | grep csi-baremetal-sc-

   2. Delete each ObjectScale PVC one by one until all have been removed:

      kubectl -n <OBJECTSCALE_NAMESPACE> delete pvc <PVC_NAME>

   3. List all the volumes used by ObjectScale:

      kubectl -n <OBJECTSCALE_NAMESPACE> get volumes | grep csi-baremetal-sc-

   4. Delete each ObjectScale volume one by one until all have been removed:

      kubectl -n <OBJECTSCALE_NAMESPACE> delete volumes <VOLUME_NAME>

   5. List all the LVGs used by ObjectScale:

      kubectl -n <OBJECTSCALE_NAMESPACE> get lvgs | grep csi-baremetal-sc-

   6. Delete each ObjectScale LVGs one by one until all have been removed:

      kubectl -n <OBJECTSCALE_NAMESPACE> delete lvgs <LVG_NAME>

   NOTE:Repeat these steps for any other object store namespaces.
8. Delete the csibmnodes resources:

   kubectl -n <CSI_NAMESPACE> delete csibmnodes --all

9. Delete CSI Bare-Metal.

   helm -n <CSI_NAMESPACE> delete csi-baremetal

10. Delete CSI Bare-Metal Operator.

    helm -n <CSI_NAMESPACE> delete csi-baremetal-operator

11. Delete the OpenShift secondary scheduler operator.

    helm -n <SECONDARY_SCHEUDLER_OPERATOR_NAMESPACE> delete openshift-secondary-scheduler-operator

12. Delete CSI Bare-Metal alerts.

    helm -n <CSI_NAMESPACE> delete csi-baremetal-alerts

13. Delete the associated CSI Custom Resource Definitions.

    kubectl delete crd deployments.csi-baremetal.dell.com availablecapacities.csi-baremetal.dell.com availablecapacityreservations.csi-baremetal.dell.com logicalvolumegroups.csi-baremetal.dell.com volumes.csi-baremetal.dell.com drives.csi-baremetal.dell.com nodes.csi-baremetal.dell.com

14. Delete the role that you created for CSI.

    kubectl delete role pod-csi -n <CSI_NAMESPACE>

15. Delete the rolebinding that you created for CSI.

    kubectl delete rolebinding role-binding -n <CSI_NAMESPACE>

16. Also, complete the cleanup of the deployment:
    1. Check that there are no ObjectScale jobs remaining:

       kubectl -n <OBJECTSCALE_NAMESPACE> get jobs

    2. Check that there are no ObjectScale cron jobs remaining:

       kubectl -n <OBJECTSCALE_NAMESPACE> get cronjobs

    3. Check that there are no ObjectScale secrets remaining:

       kubectl -n <OBJECTSCALE_NAMESPACE> get secrets | grep <DOCKER_REGISTRY_SECRET>

       For each returned ObjectScale secret, use:

       kubectl -n <OBJECTSCALE_NAMESPACE> delete secret <SECRET>

17. Check for the following remaining resources in ObjectScale and object store namespaces and carefully delete only the resources that have "objectscale" or the name of a deleted object store:
    CAUTION:Only delete ObjectScale or object store entries.
    1. List any config maps:

       kubectl -n <OBJECTSCALE_NAMESPACE> get configmaps

    2. List any acr:

       kubectl -n <OBJECTSCALE_NAMESPACE> get acr

    3. List any Pravega CRDs:

       kubectl -n <OBJECTSCALE_NAMESPACE> get crd | grep pravega

    4. List any apps:

       kubectl -n <OBJECTSCALE_NAMESPACE> get app

    5. List any service accounts:

       kubectl -n <OBJECTSCALE_NAMESPACE> get sa

    6. List any service procedures:

       kubectl get serviceprocedures -A

    NOTE:Repeat these steps for any other object store namespaces. If you are unable to delete any component or resource, then edit that resource and check for finalizers. Delete that row including finalizers; then the component or resource is automatically deleted.