The HPE CSI Driver for Kubernetes leverages Hewlett Packard Enterprise storage platforms to provide scalable and persistent storage for stateful applications.
- Upstream Kubernetes version >= 1.18
- Most Kubernetes distributions are supported
- Recent Ubuntu, SLES, CentOS or RHEL compute nodes connected to their respective official package repositories
- Helm 3 (Version >= 3.2.0 required)
Depending on which Container Storage Provider (CSP) is being used, other prerequisites and requirements may apply, such as storage platform OS and features.
The following table lists the configurable parameters of the chart and their default values.
| Parameter | Description | Default |
|---|---|---|
| disable.nimble | Disable HPE Nimble Storage CSP Service. |
false |
| disable.primera | Disable HPE Primera (and 3PAR) CSP Service. |
false |
| disable.alletra6000 | Disable HPE Alletra 6000 CSP Service. |
false |
| disable.alletra9000 | Disable HPE Alletra 9000 CSP Service. |
false |
| disableNodeConformance | Disable automatic installation of iSCSI/Multipath Packages. | false |
| disableNodeGetVolumeStats | Disable NodeGetVolumeStats call to CSI driver. | false |
| imagePullPolicy | Image pull policy (Always, IfNotPresent, Never). |
IfNotPresent |
| iscsi.chapUser | Username for iSCSI CHAP authentication. | "" |
| iscsi.chapPassword | Password for iSCSI CHAP authentication. | "" |
| logLevel | Log level. Can be one of info, debug, trace, warn and error. |
info |
| registry | Registry to pull HPE CSI Driver container images from. | quay.io |
| kubeletRootDir | The kubelet root directory path. | /var/lib/kubelet |
| controller.labels | Additional labels for HPE CSI Driver controller Pods. | {} |
| controller.nodeSelector | Node labels for HPE CSI Driver controller Pods assignment. | {} |
| controller.affinity | Affinity rules for the HPE CSI Driver controller Pods. | {} |
| controller.tolerations | Node taints to tolerate for the HPE CSI Driver controller Pods. | [] |
| csp.labels | Additional labels for CSP Pods. | {} |
| csp.nodeSelector | Node labels for CSP Pods assignment. | {} |
| csp.affinity | Affinity rules for the CSP Pods. | {} |
| csp.tolerations | Node taints to tolerate for the CSP Pods. | [] |
| node.labels | Additional labels for HPE CSI Driver node Pods. | {} |
| node.nodeSelector | Node labels for HPE CSI Driver node Pods assignment. | {} |
| node.affinity | Affinity rules for the HPE CSI Driver node Pods. | {} |
| node.tolerations | Node taints to tolerate for the HPE CSI Driver node Pods. | [] |
It's recommended to create a values.yaml file from the corresponding release of the chart and edit it to fit the environment the chart is being deployed to. Download and edit a sample file.
These are the bare minimum required parameters for a successful deployment to an iSCSI environment if CHAP authentication is required.
iscsi:
chapUser: "<username>"
chapPassword: "<password>"
Tweak any additional parameters to suit the environment or as prescribed by HPE.
To install the chart with the name my-hpe-csi-driver:
Add HPE helm repo:
helm repo add hpe-storage https://hpe-storage.github.io/co-deployments/
helm repo update
Install the latest chart:
kubectl create ns hpe-storage
helm install my-hpe-csi-driver hpe-storage/hpe-csi-driver -n hpe-storage -f myvalues.yaml
Note: myvalues.yaml is optional if no parameters are overridden from defaults. Also pay attention to what the latest version of the chart is. If it's labeled with prerelease and a "beta" tag, add --version X.Y.Z to install a "stable" chart.
Due to the helm limitation to not support upgrade of CRDs between different chart versions, helm chart upgrade is not supported. Our recommendation is to uninstall the existing chart and install the chart with the desired version. CRDs will be preserved between uninstall and install.
Before version 2.0.0 is uninstalled, the following CRDs needs to be updated.
Important: If there are HPE Alletra 9000, Primera or 3PAR Remote Copy Groups configured on the cluster, follow the next steps before uninstallation.
This step is only necessary if there are HPE Alletra 9000, Primera or 3PAR Remote Copy Groups configured on the cluster. If there are none, proceed to the next step.
Change kubectl context into the Namespace where the HPE CSI Driver is installed. The most common is "hpe-storage".
kubectl config set-context --current --namespace=hpe-storage
Create the Job using the below commands, which will modify the "rcg-info" record to the new key "RCGCreatedByCSP".
kubectl apply -f https://raw.githubusercontent.com/hpe-storage/co-deployments/master/yaml/rcg-info/v1.0.0/convert-rcg-info.yaml
Completion of job status can be verified using the below command.
kubectl wait --for=condition=complete --timeout=600s job/primera3par-rcg-info
Continue to update the CRDs followed by uninstalling the chart.
Before reinstallation of the driver, apply the new CRDs.
kubectl apply -f https://raw.githubusercontent.com/hpe-storage/co-deployments/master/helm/charts/hpe-csi-driver/crds/hpevolumeinfos_v2_crd.yaml
kubectl apply -f https://raw.githubusercontent.com/hpe-storage/co-deployments/master/helm/charts/hpe-csi-driver/crds/hpevolumegroupinfos_v2_crd.yaml
kubectl apply -f https://raw.githubusercontent.com/hpe-storage/co-deployments/master/helm/charts/hpe-csi-driver/crds/snapshotgroupinfos_v2_crd.yaml
kubectl apply -f https://raw.githubusercontent.com/hpe-storage/co-deployments/master/helm/charts/hpe-csi-driver/crds/hpereplicated_deviceinfo_v2_crd.yaml
To uninstall the my-hpe-csi-driver chart:
helm uninstall my-hpe-csi-driver -n hpe-storage
Note: Due to a limitation in Helm, CRDs are not deleted as part of the chart uninstall.
In some cases it's more practical to provide the local configuration via the helm CLI directly. Specify each parameter using the --set key=value[,key=value] argument to helm install. These will take precedence over entries in values.yaml. For example:
helm install my-hpe-csi-driver hpe-storage/hpe-csi-driver -n hpe-storage \
--set iscsi.chapUsername=admin \
--set iscsi.chapPassword=xxxxxxxx
Enable dynamic provisioning of persistent storage by creating a StorageClass API object that references a Secret which maps to a supported HPE primary storage backend. Refer to the HPE CSI Driver for Kubernetes documentation on HPE Storage Container Orchestration Documentation. Also, it's helpful to be familiar with persistent storage concepts in Kubernetes prior to deploying stateful workloads.
The HPE CSI Driver for Kubernetes Helm chart is fully supported by HPE.
Formal support statements for each HPE supported CSP is available on SCOD. Use this facility for formal support of your HPE storage products, including the Helm chart.
Please file any issues, questions or feature requests you may have here (do not use this facility for support inquiries of your HPE storage product, see SCOD for support). You may also join our Slack community to chat with HPE folks close to this project. We hang out in #NimbleStorage, #3par-primera, and #Kubernetes. Sign up at slack.hpedev.io and login at hpedev.slack.com
We value all feedback and contributions. If you find any issues or want to contribute, please feel free to open an issue or file a PR. More details in CONTRIBUTING.md
This is open source software licensed using the Apache License 2.0. Please see LICENSE for details.