Version 5.3.X Release Highlights

Version 5.3 of Trilio for Kubernetes introduces major new capabilities as follows

5.3.0

✨ New Features

• In-Place Restore: Trilio for Kubernetes (T4K) now supports In-Place Restore, enabling applications to be restored to a previous known-good state by fully cleaning up existing resources before restoring from backup. This guarantees a clean, consistent recovery with no residual, drifted, or corrupted resources left behind.

  Key capabilities include:

  • Automatic cleanup strategy selection based on backup scope (namespace-level or application-level)

  • Two-phase resource deletion with graceful cleanup followed by automatic force deletion for stuck resources

  • Application-aware cleanup for Helm, Operator, OLM, Virtual Machine, and custom applications

  • Atomic cleanup and restore of infrastructure resources (ServiceAccounts, Secrets, ConfigMaps, RBAC) to preserve cluster stability

  • Built-in safety guardrails such as protected critical namespaces, ownership validation for data components, and automatic exclusion of shared cluster resources (CRDs, StorageClasses, OLM infrastructure)

  • Full visibility into cleanup actions through detailed status tracking and force-deletion warnings in the Restore custom resource

  In-Place Restore is configured using the new cleanupConfig field in the Restore or ClusterRestore custom resource. Refer to the In-Place Restore documentation for configuration details.

• License Configuration via Secret: License management has been enhanced to improve security and flexibility. Instead of specifying the license key directly in the License custom resource, users can now store the license key in a Kubernetes Secret and reference that Secret from the License CR. Refer this

🐞 Bug Fixes

• Windows NFS Target Support: Fixed an issue where targets backed by NFS hosted on Windows servers failed during creation. Added support for Windows NFS locking mechanisms, enabling successful target creation and usage.

5.3.1

KubeVirt / OpenShift Virtualization

• Backups no longer fail or stall when a namespace contains KubeVirt backend storage PVCs (persistent TPM or UEFI state). T4K now recognizes these VM-owned volumes that are not listed on the VM's disk list, excludes them from the data snapshot, and records a warning on the backup so the behavior is visible.

• Restore: No extra handling is required — KubeVirt recreates backend storage when the VM starts; TPM/UEFI persistent state is not restored from backup, which aligns with KubeVirt's current limitations for that storage.

Other fixes

• Resolved intermittent failures when cleaning up snapshots on the target caused cleanup pods to fail.

• Fixed transformation

  • For an entire GVK when some objects lacked the configured JSON path.

  • For backupplan CR, when StorageClass is not available on site cluster.

• Fixed pre-cluster backup validation failing for ClusterBackupPlan resources that use a namespaceSelector after some selected namespaces were removed.