[RELEASE] ScyllaDB Manager 3.8.1

Release Notes
, ,
1

The ScyllaDB team announces the release of ScyllaDB Manager 3.8.1, a production-ready patch release of the stable 3.8 branch.

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

This release focuses on bug fixes and compatibility fixes with new ScyllaDB features.

Below are the changes in this release.

Task is always rescheduled in UTC timezone (#4743)

This bug was causing tasks to be rescheduled using the UTC timezone, ignoring the timezone explicitly set with the –timezone flag. Although this did not affect the actual execution interval of the task, it did incorrectly shift the task’s starting time.

Tablet repair task causes general repair task to fail (#4751)

With the ScyllaDB Manager 3.8.0 release, we introduced a dedicated tablet repair task. This feature allows users to repair keyspaces replicated with vnodes and tablets at different frequencies (leveraging ScyllaDB Incremental Tablet Repair feature), as well as repairing tablet keyspaces during cluster topology changes.

The general repair task begins by checking the cluster for any currently running ScyllaDB repair jobs. If any are found, the general repair task fails, assuming the cluster is already busy with repair activity. This check incorrectly caused the general repair task to fail whenever a tablet repair task was already running.

ScyllaDB Manager 3.8.1 resolves this issue on two levels:

  • Scheduler now prevents general and tablet repair tasks from running simultaneously, unless the general repair task is explicitly using –keyspace-replication=vnodes. Running these overlapping repair tasks in parallel is inefficient and could lead to degraded performance.
  • When the general repair task uses –keyspace-replication=vnodes, the initial check for running ScyllaDB repair jobs is now limited only to keyspaces that are replicated with vnodes.

Handle LWT state tables during backup and restore procedures (#4732)

Beginning with ScyllaDB 2025.4 LWT (Lightweight Transactions) state for keyspaces that are replicated with tablets is now kept in tables within the user keyspaces, rather than the previously used system.paxos table. This caused failures in Alternator backup and CQL schema restore procedures, as they weren’t adjusted accordingly.

The LWT state is an internal implementation detail, containing only the information needed to process ongoing LWT queries. Because of this, it has never been included in the backup or restore procedures.

Skip audit keyspace schema restoration (#4748)

When the ScyllaDB Audit feature is used in table mode, an audit keyspace is created. Despite not having the typical system prefix, this is a system keyspace that ScyllaDB creates automatically upon cluster startup. This caused issues when restoring a backed-up schema onto a fresh cluster, as the audit keyspace schema was already present there.

As a temporary solution, we are now skipping the restoration of the audit keyspace schema and relying on ScyllaDB to recreate it upon cluster startup. We plan to deliver a more comprehensive procedure for restoring the audit keyspace schema in the future.

If necessary, the desired audit keyspace schema can be manually recreated using the backed-up schema file.

Other bug fixes and improvements

  • Stopped and disabled task is not stopped (#4736)
  • Known cluster hosts are cleared up on cluster update (#4733)
  • Describe task statuses in managerclient pkg (#4739)

Upgrade to the new release

ScyllaDB customers are encouraged to upgrade to ScyllaDB Manager 3.8.1 in coordination with the ScyllaDB support team.

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

Useful Links:

Scylla Manager 3.8.1 supports the following ScyllaDB releases

  • 2025.4
  • 2025.3
  • 2025.2
  • 2025.1
  • 2024.1

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