Changes since v25.10.0
Trident
IMPORTANT: There is a known issue with the csi-snapshotter sidecar. In all versions of Kubernetes VolumeGroupSnapshots v1beta1 will not allow VolumeSnapshots to become ReadyToUse. There are 2 workarounds:
- Delete VolumeGroupSnapshots CRDs, which will disable VolumeGroupSnapshots, and reinstall Trident.
- Install VolumeGroupSnapshots v1beta2 and snapshot-controller v8.4.0 or later, and reinstall Trident. VolumeGroupSnapshots will not function on Kubernetes versions lower than v1.34.
Fixes:
- Kubernetes: Fixed an issue where unpublishing a read-only clone removed export policy rules from the source volume in ONTAP-NAS, ONTAP-NAS-Economy, and GCNV drivers (Issue #1086).
- Kubernetes: Switched kubectl images from deprecated Bitnami to lightweight Alpine-based variants to prevent pull failures following Bitnami's public image deprecation (Issue #1080).
- Kubernetes: Allow clone across different storage classes if both storage classes are pointing to the same backend (Issue #1104).
- Kubernetes: Fixed node prep failures caused by timeouts in cloud environments with network latencies. Increased timeout values for cloud-based installations
- Kubernetes: Fixed an issue in LUN creation that caused the filesystem type attribute to remain unset when the process entered a retry state
- Kubernetes: Fixed REST API volume lookup to ignore volume state, preventing false negatives during volume queries.
- Kubernetes: Improved Trident controller efficiency for ontap-nas-economy driver when used at scale.
- Kubernetes: Set internalID during LUN import in ontap-san-economy driver.
- Kubernetes: Increased Azure Resource Graph query limits to handle more subnets.
- Kubernetes: Improved CSI and ONTAP clone split timeouts to avoid race conditions with some backup applications (Issues #1098, #1100).
- Kubernetes: Fixed suppression of LUKS error messages (Issue #1069).
- Kubernetes: Fixed handling of stale LUKS mappers for both iSCSI and NVMe protocols. Enhanced cleanup logic prevents mount failures from orphaned device mappers.
- Fixed scale limitations for RWX NVMe volumes.
- Updated package "telemetry/opentelemetry-go" to fix CVE-2026-24051.
Enhancements:
- Kubernetes: Added support for Kubernetes 1.35.
- Kubernetes: Concurrency support for ONTAP-NAS (NFS only), ONTAP-SAN (iSCSI, FCP), ONTAP-SAN (NVMe) and GCNV drivers is now generally available (GA), graduated from Tech Preview.
- Kubernetes: Added support for volume autogrow in Trident based on user-defined Trident AutoGrow Policies.
- Kubernetes: Added support for VolumeGroupSnapshot for ontap-san (NVMe/TCP), in addition to existing support for ontap-nas (NFS), ontap-san (iSCSI and FC), and ontap-san-economy.
- Kubernetes: Enhanced Trident node concurrency for higher scalability of node operations for NVMe volumes.
- Kubernetes: Added support for GCNV NAS volume auto-tiering via PVC annotations (tieringPolicy, tieringMinimumCoolingDays), with pool selection and clone inheritance.
- Kubernetes: Added support for GCNV SAN block (iSCSI) volumes via the google-cloud-netapp-volumes-san driver, including provisioning, per-node host group mapping for LUN access, and clone-from-volume for Flex pools.
- Added support for automatic backend configuration for AWS FSxN drivers.
- Added support for different Azure clouds (e.g. Azure Government, Azure China) and custom cloud configuration for azure-netapp-files (ANF) backends (Issue #632).
- Kubernetes: Trident preserves annotation of existing deployment during upgrade (Issue #1004).
Experimental Enhancements:
NOTE: Not for use in production environments.
- [Tech Preview] Added support for concurrency for ONTAP-NAS-Economy and ONTAP-SAN-Economy drivers.
Trident Protect
Fixes:
- Fixed sorting of Kubernetes resources for restore operations.
- Handled non-unique IDs during restores.
- Fixed cluster name in tridentctl-protect output.
- Fixed Kopia restore errors being ignored.
- Fixed app definition when it has no included PV.
- Fixed restoring of 2 PVCs with same name in 2 different namespaces.
Enhancements:
- Ability to define execHooks in CR spec.
- Added data-mover-timeout-sec annotation to schedule CR.
- Allowed destination app name to be specified during restore operations.
- Allowed schedule to create a snapshot immediately rather than wait for first occurrence.
- Schedules are now disabled during IPR operations and re-enabled when complete.