Kubernetes 1.29: Easily Modify Volume Attributes – A Developer’s Advantage!

The release of Kubernetes 1.29 introduces an exciting new feature: an alpha capability that allows volumes to be modified by changing the volumeAttributesClassName set for a PersistentVolumeClaim (PVC). This feature significantly simplifies the management of volume attributes, as changes can now be made directly through Kubernetes rather than having to manage them through various vendor APIs.

VolumeAttributesClass

The new API group storage.k8s.io/v1alpha1 offers two new types:

VolumeAttributesClass
Represents a specification of mutable volume attributes defined by the CSI driver. This class can be specified during the dynamic provisioning of PersistentVolumeClaims and altered in the PVC specification after provisioning.

ModifyVolumeStatus
Represents the status object of the ControllerModifyVolume operation.

With this alpha feature enabled, the specification of the PersistentVolumeClaim defines the VolumeAttributesClassName used in the PVC. During the volume provisioning, the CreateVolume operation will apply the parameters in the VolumeAttributesClass along with the parameters in the StorageClass.

When a change to the volumeAttributesClassName occurs in the PVC specification, the external-resizer sidecar receives an informer event. Based on the current state of the configuration, the resizer will trigger a CSI ControllerModifyVolume. For more details, see KEP-3751.

How to Use It

If you want to test this feature in its alpha stage, you need to enable the relevant feature gate in both the kube-controller-manager and the kube-apiserver. Use the command line argument --feature-gates:

--feature-gates="...,VolumeAttributesClass=true"

Additionally, it is required that the CSI driver has implemented the ModifyVolume API.

This new functionality opens up exciting possibilities for developers and DevOps teams to simplify and make volume management more efficient. At ayedo, we are proud to support and advance such developments as a partner in the Kubernetes community.

Source: Kubernetes Blog