Changes since v26.02.0
Trident
Fixes:
- Fixed an issue where iSCSI device resize would fail if available paths and portals did not match.
- Fixed an intermittent issue where
blkidmisidentifies the filesystem type for LUKS devices preventing new volume creations. - Fixed an issue for GCNV SAN where FLEX UNIFIED ZONAL pools were not able to be used. Now Zonal works as well as Regional.
- Kubernetes: Fixed indentation in tridentactionmirrorupdates k8s CRD YAML (Issue #1120).
- Fixed REST qtree asynchronous delete API call (Issue #1121).
- Enhanced AWS ARN handling to support special regions.
- Kubernetes: Fixed an issue preventing KubeVirt VMs using dataVolumes from failing over using Trident Automated Workload Failover.
- Fixed an issue in the GCNV NAS driver for UNIFIED pools by applying UNIFIED-specific naming and validation and improving fallback volume lookup when InternalID is absent.
- Fixed NVMe/TCP namespace race condition during concurrent volume creation (Issue #1089).
- Fixed controller startup log to correctly report concurrency driver status.
- Fixed handling of suspended backends in concurrent Trident.
- Fixed ControllerPublish to use volume config file system type to avoid using incorrect default file system type.
- Fixed an issue where LUKS publish and stage operations at scale could block Trident controller. Users are now expected to track LUKS passphrases used by volumes.
- Fixed an issue where iSCSI multipath partitions could show up as ghost devices, preventing CSI NodeUnstage operations from succeeding.
- Kubernetes: Fixed an error when attempting to delete an invalid TridentMirrorRelationship CR with trident concurrent core.
- Fixed concurrent clone snapshot when snapshots are executed in the same second.