NFS Options for Private Cloud

Cloudera Machine Learning on Private Cloud requires a Network File System (NFS) server for storing project files and folders.

On ECS, NFS is part of the overall installation, and no additional setup steps are required.

You can use an NFS server that is external to the cluster, such as a NetApp Filer appliance. In this case, you must manually create a directory for each workspace. In this case, the NFS server must be configured before deploying the first CML workspace in the cluster. One important limitation is that CML does not support using shared volumes for storing project files.

Storage provisioner change On OCP, CephFS is used as the underlying storage provisioner for any new internal workspace on PVC 1.5.0. A storage class named "ocs-storagecluster-cephfs" with csi driver set to "openshift-storage.cephfs.csi.ceph.com" must exist in the cluster for new internal workspaces to get provisioned. Each workspace will have separate 1 TB internal storage.

On ECS, any new internal workspace on 1.5.0 will use Longhorn as the underlying storage provisioner. A storage class named "longhorn" with csi driver set to "driver.longhorn.io" must exist in the cluster for new internal workspaces to get provisioned. Each workspace will have separate 1 TB internal storage.

On either ECS or OCP, internal workspaces running on PVC 1.4.0/1.4.1 use NFS server provisioner as the storage provisioner. These workspaces when upgraded to 1.5.0 will continue to run with the same NFS server Provisioner. However, NFS server provisioner is deprecated now and will not be supported in 1.5.1 release.

Existing workspaces in 1.4.0/1.4.1 can be upgraded to 1.5.0 from PVC UI. After this, you can do one of the following:
  • Migrate the 1.5.0 upgraded workspace from NFS server provisioner to Longhorn (ECS) / Cephfs (OCP) if you want to continue using the same workspace in PVC 1.5.1 as well
  • Create a new 1.5.0 workspace and migrate the existing workloads to that before 1.5.1 release.