This is a guest post by Michael Mattsson Principal Tech Marketing Engineer at HPE.

Hewlett Packard Enterprise and Red Hat have over two decades of innovation in our joint alliance partnership spanning a wide variety of infrastructure solutions built on open source software and high performance hardware backed by a customer-focused joint support engagement that ensures an unsurpassed product experience.

We are excited today to formally introduce a new solution as both teams have been heads down delivering the first Operator to achieve Red Hat OpenShift container storage interface (CSI) certification. The upstream HPE CSI Driver for Kubernetes has been repackaged, hardened and delivered by HPE and certified by Red Hat. The HPE CSI Operator for Kubernetes is published in the Red Hat Ecosystem Catalog and can be deployed from the OpenShift OperatorHub. The HPE CSI Driver is a unified CSI driver supporting multiple enterprise block storage (iSCSI and Fibre Channel) platforms through a pluggable architecture. HPE Nimble Storage, HPE Primera and HPE 3PAR are currently supported and certified.

Technology Overview

The HPE CSI Driver is a comprehensive CSI driver currently conforming to version 1.2 of the CSI specification. Data management capabilities such as snapshots and cloning are available through CSI or StorageClass parameter overloading during provisioning to allow Red Hat OpenShift users to immediately take advantage of their HPE storage platforms. CSI native cloning and snapshots is currently available as a “Technology Preview” in Red Hat OpenShift 4.4. Other notable features currently under “Technology Preview” in Red Hat OpenShift is CSI inline ephemeral volumes and expanding CSI volumes. All features are marked GA by the HPE CSI Driver and customer may rest assured they’ve invested in a future proof storage platform for Red Hat OpenShift.


The HPE CSI Driver is designed to support multiple storage backends and a clean separation has been created between CSI and the storage backends. This allows the backend storage teams to focus on making sure they present their platform capabilities in the most efficient way possible. A separate team is dedicated to ensuring all the Kubernetes, OpenShift and CSI minutia conforms to recent standards, conventions and certifications. The interface against the backend storage platforms is called Container Storage Providers (CSP). The CSP follows an open source specification that allows any vendor with a block storage (iSCSI or Fibre Channel) protocol to implement a CSP. Both Nimble and Primera/3PAR CSPs run on the Kubernetes cluster as ephemeral API gateways but the CSP may run embedded on the storage platform itself (or elsewhere) as long as it allows communication from the cluster and follow the CSP specification.


The architecture also allows building extensions either to CSI itself using traditional sidecars or extend functionality by other means. Any such extension will become available for existing CSPs automatically. HPE recently announced a beta of the NFS Server Provisioner feature in version 1.2.0 of the HPE CSI Driver. It allows the CSI driver to create ReadWriteMany or ReadOnlyMany (RWX/ROX) PersistentVolumeClaims by transparently spinning up a NFS server on the Kubernetes cluster served by a ReadWriteOnce (RWO) volume. This lends itself well for certain use cases where shared storage between Pods within the same Namespace is required which isn’t possible with strict block storage, which only supports RWO. Expect more innovations like these from HPE in the near future!

Get Started!

Installing the HPE CSI Operator may be performed either via the OpenShift Console or the OpenShift CLI. For the purpose of demonstrating the installation process, we’ll use oc. How to use the OpenShift Console to install the CSI operator is available on the HPE Storage Container Orchestrator Documentation portal (SCOD).


A preview of what the certified HPE CSI Operator for Kubernetes looks like in the OpenShift Console.

The CSI driver requires privileged access. For the driver to work correctly a custom SecurityContextConstraint needs to be applied. But, first let’s create a new Project where we would deploy the CSI operator.

oc new-project hpe-csi-driver --display-name="HPE CSI Driver for Kubernetes"

The next set of YAML stanzas may be created by running oc create -f- and paste the YAML into the terminal. Hit CTRL-D on a new line to apply it.

Create the custom SCC.

kind: SecurityContextConstraints
name: hpe-csi-scc
allowHostDirVolumePlugin: true
allowHostIPC: true
allowHostNetwork: true
allowHostPID: true
allowHostPorts: true
allowPrivilegeEscalation: true
allowPrivilegedContainer: true
- '*'
defaultAddCapabilities: []
type: RunAsAny
groups: []
readOnlyRootFilesystem: false
requiredDropCapabilities: []
type: RunAsAny
type: RunAsAny
type: RunAsAny
- system:serviceaccount:hpe-csi-driver:hpe-csi-controller-sa
- system:serviceaccount:hpe-csi-driver:hpe-csi-node-sa
- system:serviceaccount:hpe-csi-driver:hpe-csp-sa
- system:serviceaccount:hpe-csi-driver:hpe-csi-operator-sa
- system:serviceaccount:hpe-nfs:hpe-csi-nfs-sa
- '*'

Next, an OperatorGroup and Subscription needs to be created.

kind: OperatorGroup
name: hpe-csi-driver-for-kubernetes
namespace: hpe-csi-driver
- hpe-csi-driver
kind: Subscription
name: hpe-csi-operator
namespace: hpe-csi-driver
channel: stable
name: hpe-csi-operator
source: certified-operators
sourceNamespace: openshift-marketplace

Watch the CSI operator being rolled out.

oc rollout status deploy/hpe-csi-driver-operator -n hpe-csi-driver

The next step involves creating a HPECSIDriver. This is where you need to make conscious choice if you want to use HPE Nimble Storage (nimble) or HPE Primera/HPE 3PAR (primera3par), which share the same backendType.

kind: HPECSIDriver
name: csi-driver
namespace: hpe-csi-driver
backendType: nimble
imagePullPolicy: IfNotPresent
logLevel: info
create: true
password: admin
servicePort: '8080'
username: admin
allowVolumeExpansion: true
create: false
defaultClass: false
name: hpe-standard
accessProtocol: iscsi
fsType: xfs
volumeDescription: Volume created by the HPE CSI Driver for Kubernetes

Note: The backend parameter refers to the IP address of the management interface on the storage system.

The next step would be to create a StorageClass to allow dynamic PersistentVolume provisioning from PersistentVolumeClaims. Below is a basic StorageClass that will be marked as “default” on the cluster.

kind: StorageClass
annotations: "true"
name: hpe-storageclass
parameters: xfs nimble-secret kube-system nimble-secret kube-system nimble-secret kube-system nimble-secret kube-system nimble-secret kube-system
description: "Volume created by the HPE CSI Driver for Kubernetes”
fsMode: "0770"
reclaimPolicy: Delete
allowVolumeExpansion: true

The StorageClass parameters are unique for the different CSPs. These are documented on SCOD for HPE Nimble Storage and HPE Primera/HPE 3PAR.

Next steps could be to deploy any application from OperatorHub requiring persistent storage, such as MariaDB or MongoDB.

Learn More

The Red Hat OpenShift certified HPE CSI Operator for Kubernetes is available now. Current supported platforms are HPE Nimble Storage, HPE Primera and HPE 3PAR. Sign up to the HPE DEV Slack community at (or login at if you already have signed up) to chat with the HPE staff, partners and customers. Also stay informed via our announcements and updates, we hang out in #kubernetes, #nimblestorage and #3par-primera.


Storage, How-tos, Operators

< Back to the blog