What's new since v0.4.0
This is the first release candidate for v1.0.0. We're looking for feedback and real-world testing before the final release.
This release makes the driver more secure, more flexible, and easier to deploy.
Split node & controller modes — The Xen Orchestra API access is now restricted to the CSI controller only. Node daemons
no longer carry XO credentials or secrets; they rely on CCM-provided node metadata instead. Smaller attack surface, cleaner
RBAC.
SR selection with VolumeAttributesClass (Kubernetes ≥ 1.31) — Use the new storageRepositoryId parameter to create a VDI
directly in the Storage Repository of your choice, with validation against the pool and storage type. You can also migrate an
existing volume to another SR within the same pool at runtime, simply by annotating the PVC — csi-resizer picks it up and the
VDI migrates once the pod is stopped. Precedence is now: VAC SR > explicit poolId > topology-aware selection.
Easier tuning — The XO client timeout is a proper driver flag (-xo-client-timeout, default 30s), no longer stored
alongside secrets, and the kxo helper gains a --vdi-name-prefix override.
Helm chart — A new, fully tested Helm chart aligned with the split deployment modes, with toggles to enable or disable
components, plus Helm hook tests for CSI volume provisioning.
All the usual housekeeping: dependency bumps (gRPC, k8s, CI actions) and linter fixes.
Full Changelog: v0.4.0...v1.0.0-rc.1