[RELEASE] ScyllaDB Manager 3.5.0

The ScyllaDB team announces the release of ScyllaDB Manager 3.5.0, a production-ready minor release of the stable 3.5 branch.

ScyllaDB Manager is a centralized cluster administration and recurrent tasks automation tool.

This release focuses on supporting the ScyllaDB 2025.1 release, but it also contains a few important changes to the ScyllaDB Manager license, container image, and new features.

Below are the major changes in this release.

License Change

ScyllaDB Manager license changed to AGPL (#4229).
As ScyllaDB is moving to source available license, ScyllaDB Manager is changing its license to AGPL. See our blog post for a broader explanation and effective scope of this change.

OpenShift

ScyllaDB Manager container image OpenShift certification (#4242).
As a part of the effort of getting ScyllaDB Operator OpenShift certification, ScyllaDB Manager container images are now passing all the criteria in sections 2.1 (image content requirements) and 2.2 (image policy requirements) of the certification policy guide. Among other changes, ScyllaDB Manager image is now based on the Red Hat ubi9-minimal image.

Leveraging ScyllaDB New Tablets Repair API

ScyllaDB Tablet repair API support (#4188, #4292, #4273).
ScyllaDB 2025.1.0 introduced a new storage_service/tablets/repair API for repairing tablet replicated tables. The old /storage_service/repair_async/{keyspace} API proved to be unsafe when used with tablet tables. Moreover, it required stopping tablet load balancing for the time of repair, which could cause various problems for long running repairs.
As it turned out, previous ScyllaDB Manager releases contained a bug which resulted in not stopping tablet load balancing when using the old repair API. This in combination with using non default tombstone gc mode ‘repair’ could lead to data resurrection. Because of that, all users are encouraged to upgrade to ScyllaDB Manager 3.5 even if they haven’t started using the ScyllaDB 2025.1 yet.

Improve Restore

New --dc-mapping restore task flag (#3829).
This new flag allows for specifying which data center in the restored cluster will be responsible for downloading and streaming the data from which backed up data center. Using this flag in a multi data center scenario allows to save on cross region traffic costs and improve restore speed. This flag works only when the number of data centers in the restored cluster is the same as in the backup. For more information see the sctool documentation.

Don’t allow for other tasks to run during restore (#4045).
Forgetting to unschedule other tasks for the time of running restore was a common source of issues. Now, it is handled by ScyllaDB Manager automatically. For more information see restore task documentation.

Bug Fixes

The ScyllaDB Manager 3.5.0 release also contains a few bug fixes:

  • choose correct prometheus address by consul (#4300)
  • make sctool progress work for restore tasks even when the cluster is no longer available (#4191)

Upgrade to the new release

ScyllaDB customers are encouraged to upgrade to ScyllaDB Manager 3.5.0 in coordination with the ScyllaDB support team.

The new release includes upgrades of both ScyllaDB Manager Server and Agent.

Useful Links:

Scylla Manager 3.5.0 supports the following ScyllaDB releases:

  • 2025.1
  • 2024.2
  • 2024.1

You can install and run Scylla Manager on Kubernetes using Scylla Operator. More here.