[Release] ScyllaDB 2025.4.9

Release Notes
, ,
1

The ScyllaDB team is pleased to announce the release of ScyllaDB 2025.4.9, a production-ready patch release for ScyllaDB 2025.4 Feature Release.

Note there is a new Long Term Support (LTS) release 2026.1. You are welcome to upgrade to it for the latest and greatest features.

More information on ScyllaDB Version Support Policy is available here.

Related Links

Bug Fixes

Administration

  • Local system table creation could experience synchronization issues during concurrent operations. The synchronization handling was improved for internal table creation, improving stability during these operations.
    scylladb#29861

Monitoring

  • Diagnostic logging for internal Barrier and drain operations provided limited detail for troubleshooting. Additional logging was added for barrier and drain operations, improving visibility into node state during internal maintenance activities.
    scylladb#29801

  • The system.size_estimates table could produce incomplete aggregation of SSTable size reports across all shards on a node, resulting in incomplete table size estimates. This is resolved by ensuring metrics from every shard are accurately aggregated, providing cluster administrators with complete and reliable per-node table size estimates.
    scylladb#29799

Performance

  • A potentially long-running operation in the trie_writer::complete_until_depth() function could occasionally block shard progress during background tasks. A preemption point was introduced into this function, allowing the shard to yield during long background activities like compaction. This improves latency and overall node responsiveness. scylladb#29837

Reliability

  • An issue could affect a node on shard 0 when concurrent incremental repair operations were running during a scale-out procedure. This is resolved by synchronizing the incremental repair logic with the tablet split mechanism, ensuring stable cluster operations during concurrent scaling and repair activities.
    scylladb#26041, scylladb#27844

  • SSTable attachment during the streaming process could be incomplete in the consumer, sometimes leaving sealed SSTable files consuming disk space. The fix ensures that these sealed files are properly cleaned up when this occurs, preventing unnecessary disk usage and improving storage management during streaming operations.
    scylladb#27414

  • Prepared statement invalidation could lead to stability issues in the CQL layer, potentially affecting CQL service stability. This is resolved by improving internal resource handling during prepared statement invalidation. This change eliminates a source of instability and enhances the reliability of the CQL interface. scylladb#29812