[RELEASE] ScyllaDB 2025.1.10

Release Notes
, ,
1

The ScyllaDB team announces ScyllaDB 2025.1.10, a bug-fix production-ready patch release for ScyllaDB 2025.1 LTS Release.

Note that there is a new Short-Term Support (STS) Feature release, version 2025.3. You are welcome to upgrade to it for the latest and greatest features, or stay on the 2025.1 track for long-term support.

Related Links

Bug Fixes

The following issues are fixed in this release.

Authentication

  • Issue: CQL connections were accepted in the main scheduling group and remained there if authentication was not enabled.
    Fix: Calls to update the scheduling group are now made for non-authenticated connections. (scylladb#26040, scylladb#26581)

  • Issue: Service levels on Raft experienced local reads with timeouts from authentication tables.
    Fix: A long timeout is now set for authentication queries during the Service Level cache update. scylladb#25290

CDC (Change Data Capture)

  • Issue: Critical errors were observed due to a malformed_sstable_exception where a column was missing in the current schema for the cdc_log table.
    Fix: The column drop timestamp is now set in the future to prevent recreating a column too soon. scylladb#26340

Documentation

  • Issue: Documentation was missing explicit support for Debian 12.
    Fix: Support for Debian 12 has been added to the documentation. scylladb#26640

  • Issue: The documentation was missing information on the --list-active-releases option for the Web Installer.
    Fix: The documentation has been updated to include --list-active-releases for the Web Installer. scylladb#26688

Repair

  • Issue: An incorrect order of parameters was being logged in the repair log.
    Fix: The order of uuid and nodes_down in the repair log has been corrected. scylladb#26536

  • Issue: The node operations progress metric was not always resetting to 100% when the operation was complete.
    Fix: Node operations progress is now always reset to 100% upon completion. scylladb#26193

Operations

  • Issue: Maintenance mode was broken, causing test_maintenance_mode to fail.
    Fix: The system now falls back to the default scheduling group when using the maintenance socket, resolving the issue. scylladb#26816

  • Issue: The command to opt out a node from automatic cleanup was missing.
    Fix: Support to allow a node to opt out of automatic cleanup has been added. scylladb#26866

Reliability

  • Issue: Batchlog replay was using Consistency Level (CL)=ALL.
    Fix: Batches are now sent with CL=EACH_QUORUM during batchlog replay. scylladb#25432

Tooling/Automation

  • Issue: The scylla-sstable tool was printing token values in sharding_metadata as bytes instead of numbers.
    Fix: scylla-sstable now correctly dumps sharding_metadata and prints token values as numbers. scylladb#26982