We are pleased to announce the release of Velero v1.17. This is a feature-rich release that delivers volume group snapshot support, a modernized fs-backup architecture, Windows workload backup/restore, and major scalability improvements for data movers.
Velero 1.17 supports volume group snapshots, a beta feature in Kubernetes, for both CSI snapshot backup and CSI snapshot data movement. This allows snapshots to be taken from multiple volumes at the same point-in-time to achieve write order consistency, which is important for achieving better data consistency when multiple correlated volumes are backed up together.
See the documentation for details.
The fs-backup subsystem has been rebuilt on the micro-service architecture, bringing several benefits:
See the design document for details.
Velero fs-backup now supports backing up and restoring Windows workloads. By leveraging the new micro-service architecture, data mover pods can run on Windows nodes and handle Windows volumes. Together with CSI snapshot data movement for Windows delivered in v1.16, Velero now supports Windows workload backup/restore across all scenarios.
Kubernetes priority classes are now supported across all Velero modules. Users can configure priority classes separately for Velero server, node-agent, data mover pods, and backup repository maintenance jobs.
See the design document for details.
Velero resource policy now supports includeExcludePolicy alongside the existing volumePolicy. This allows users to set include/exclude filters for resources in a resource policy configmap, making these filters reusable across multiple backups.
A new PrepareQueueLength setting in node-agent configuration limits how many data mover pods and volumes are created ahead of available data path quota. This prevents excessive cluster resource consumption, particularly helpful in large-scale environments. This applies to both fs-backup and CSI snapshot data movement.
See the design document for details.
Data movements in all phases now survive node-agent restarts and resume automatically. Orphaned data movements from scenarios like cluster node absence are canceled appropriately after restart.
CSI snapshot data movement restore now has the same node-selection capability as backup. Users can specify which nodes can or cannot run data mover pods for both backup and restore, with per-storage-class configuration for environments where a storage class is not usable by all cluster nodes.
Per the
Velero deprecation policy, backup under the Restic path is removed in v1.17. --uploader-type=restic is no longer a valid installation configuration. Restores from previous Restic-path backups remain supported until v1.19.
Repository maintenance job configurations have been moved from Velero server parameters to a repository maintenance job configmap. The following server parameters are removed: --keep-latest-maintenance-jobs, --maintenance-job-cpu-request, --maintenance-job-mem-request, --maintenance-job-cpu-limit, --maintenance-job-mem-limit.
Thank you to everyone who contributed to this release:
To help you get started, see the documentation.