To which CSI driver Isilon/PowerScale version does this FAQ apply? This FAQ applies to version 1.3 of the CSI Driver for Isilon/PowerScale available here : https://github.com/dell/csi-powerscale/tree/v1.3.0
What's new with version 1.3 ?
Supports Kubernetes version 1.17, 1.18, 1.19
Supports OpenShift 4.3 and 4.4 with both RHEL and RHCOS worker nodes
Supports Docker EE 3.1
Supports volume cloning
Supports CentOS versions 7.8, Red Hat Enterprise Linux 7.8
Which CSI version does the driver conform to? The driver works with upstream Kubernetes1.17, 1.18, 1.19 ; Openshift 4.3 and 4.4 and CSI specification 1.1 with the CSI interface enabled. Additional Kubernetes versions and variants will be qualified over time.
What are the supported features? The following table lists the supported action to manage the lifecycle of a Isilon/PowerScale volume managed with the CSI driver
Action
Supported (version)
Static Provisioning
yes
Dynamic Provisioning
yes
Binding
yes
Retain Reclaiming
yes
Delete Reclaiming
yes
Recycle Reclaiming
no
Expand Persistent Volume
yes
Shrink Persistent Volume
no
Create Snapshot Volume
yes
Create Volume from Snapshot
yes
Delete Snapshot
yes
CSI Volume Cloning
yes
CSI Raw Block Volume
no
CSI ephemeral volumes
no
Is it possible to create volume in different share ? It is possible to create multiple or different StorageClass to access separate access zones.
What are the known limitations? NFS is the only supported protocol. One driver instance can manage one Isilon/PowerScale cluster only. One driver instance can be installed on the Kubernetes cluster.
With the promotion of Volume Snapshot to beta, one significant change is the CSI external-snapshotter sidecar has been split into two controllers, a common snapshot controller and a CSI external-snapshotter sidecar.
The provided install script will :
By default, install of the external-snaphotter for CSI Unity.
Optionally, install the beta snapshot CRD when the option --snapshot-crd is set during the initial installation.
How to uninstall the CSI driver? Like for the installation, you can uninstall the driver with helm : helm delete ISILON_RELEASE [--purge]
Or execute the script: sh csi-uninstall.sh
How to upgrade the CSI driver?
Previously, till version 1.2.0, the driver supported Kubernetes alpha Volume Snapshot. With the addition of support for beta snapshot, the Product Guide recommends to remove volume snapshots, volume snapshot content, and volume snapshot class objects before anything else. Nevertheless, the Kubernetes blog announcing the beta release explains how to manually import the snapshots. That procedure haven't been tested by Dell EMC so feel free to share your experience in the forum
If you come from an installation with helm v2. You have to uninstall then reinstall the driver with helm v3 : * Uninstall the driver using the latest csi-uninstall.sh * Install the driver using the latest csi-install.sh script and the new myvalues.yaml file
Because of the multi-array support the and the secret with the credentials are very different from previous versions ; make sure secret.json is valid.
The version 1.3 is the first to bring the operator support for multi-array. Make sure to use the latest version of dell-csi-operator, before proceeding to the installation.
What are the pre-requisites for CSI driver installation? To check your system complies with the pre-requisites, you can execute the script sh verify.sh
The detailed list of pre-requisites is available in the product guide
Which k8s distributions are supported? The supported versions are upstream Kubernetes 1.17, 1.18, 1.19 ; OpenShift 4.3, 4.4 with RHEL or RHCOS nodes ; Docker EE v3.1
If you need support on any other CSI v1.1 compatible distribution, please use this forum to discuss it.
What about future releases of k8s? In the next release, the objective is to support at least version n-2 of the current Kubernetes version at the moment of the release.
What operating systems are supported for Kubernetes nodes?
The driver was qualified with Red Hat Enterprise Linux and CentOS 7.6, 7.7, 7.8. With Openshift we support both RHEL and RHCOS nodes.
How do I troubleshoot the driver? The driver can be troubleshooting using the usual kubectl commands as any other k8s pod/resources. Most often used commands are : * kubectl get pods -n isilon : gives the status of the controller and drivers on every node * kubectl describe pods isilon-controller-0 -n isilon : provides details on the deployment for the controller * kubectl logs isilon-controller-0 -n isilon -c driver : logs the API calls between the driver and Unisphere
Which driver version is installed? helm list -c ISILON_RELEASE
What version and model of Isilon/PowerScale versions are supported? OneFS 8.1, 8.2 and 9 are supported.
Is volume expansion supported? Yes, volume expansion is not supported by the driver. It is not supported to reduce a volume size.
Why a volume in Isilon/PowerScale can be greater than the Volume size required by the PersistentVolumeClaim ? By default, SmartQuota feature is disabled ; without it, nothing prevents a volume to be greater than its configured size in Kubernetes.
Can I run this driver in a production environment? Yes, the driver is production-grade. Please make sure your environment follows the pre-requisites and Kubernetes' best practices.
What support does Dell EMC provide for the driver built via the GitHub sources?
Dell EMC is fully committed to supporting the driver image on Dockerhub built from the sources hosted on GitHub.
Do I need to have programming skills to use CSI driver? No.
To use the driver, you need to have basic knowledge around Kubernetes administration (How to create a PV, How to use a Volume, etc.).
How can connect Dell EMC storage to Kubernetes running on VMWare? There is no impact on the configuration or the usage of the Isilon/PowerScale driver if your Kubernetes nodes are hosted on a virtualized environment.
Are theer plans for supporting other advanced K8s solutions, like VMW/Pivotal PKS (which uses Ubuntu OS) and VMW Tanzu K8s Grid (TKG, which uses VMW PhotonOS)?
If so, what is the timeframe? I'd appreciate an update.
CoreOS support for the OpenShift worker nodes is currently on the backlog for qualification. Along with OpenShift 4.3, it is something we are trying to target for the next release.
The resizer works well! Glad that was added. I would add that I have it running on 1.17.6. I had no issues with deployment and nothing seems out of the ordinary. Also wanted to add that we use Oracle Linux 7. No issues with that either. But it makes sense since OL7 is 100% binary compatible to RHEL.
Is there a way to add Version object in any way somewhere in the driver so it's easy to find out which version of CSI driver is installed? Perhaps in the default storage class if it's allowed.
This is how i find it but there could be an easier way:
SergeySchindler
6 Posts
10336
0
Posted April 15th, 2020 06:00
Are theer plans for supporting other advanced K8s solutions, like VMW/Pivotal PKS (which uses Ubuntu OS) and VMW Tanzu K8s Grid (TKG, which uses VMW PhotonOS)?
If so, what is the timeframe? I'd appreciate an update.
Regards,
Sergey Schindler
DTS Global Enterprise Architect