[RELEASE] ScyllaDB 2025.1.13

Release Notes
, ,
1

The ScyllaDB team announces ScyllaDB 2025.1.13, a bug-fix production-ready patch release for ScyllaDB 2025.1 LTS 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’s Long Term Support (LTS) policy is available here.

Related Links

Bug Fixes

Authentication

  • Authentication version changes during node startup could affect authentication behavior. The initialization logic was updated to correctly manage authorization versions during startup, ensuring a consistent authentication process for nodes.
    scylladb#1918, scylladb#29831.

  • An issue in LDAP message processing in poll_results() could affect system stability during authentication. Memory handling was corrected to ensure safe processing of LDAP messages and improve authentication stability.
    scylladb#1446, scylladb#29646.

  • ScyllaDB could encounter an assertion when accessing the authorized_prepared_statements_cache after enabling authentication via a restart. The cache logic was updated to handle inserts safely when caching is disabled, maintaining system stability.
    scylladb#1699, scylladb#29803.

Compaction

  • New split compaction groups could be created even when the main compaction group was disabled. The logic was adjusted to prevent these groups from being created under such conditions, optimizing resource usage.
    scylladb#1549, scylladb#29700.

  • An issue during tablet cleanup could affect tasks for ongoing compactions. Tablet cleanup logic was corrected to prevent these issues, ensuring smooth cleanup processes.
    scylladb#26229, scylladb#29639.

CQL

  • Data formatting logic was improved for null handling in CQL data values to ensure consistent data representation.
    scylladb#1350, scylladb#29468.

  • Analysis of WHERE clauses and the processing of CQL function calls exhibited high recursion depth. Limits were introduced to the depth of CQL expressions to improve performance and ensure predictable query execution.
    scylladb#1002, scylladb#1003, scylladb#14472.

  • EverywhereStrategy logic was updated to correctly handle read scenarios, improving data consistency during read operations.
    scylladb#1147, scylladb#29294.

Encryption

  • Encryption coverage was extended to include the system.raft table in system_info_encryption to ensure all system tables are correctly included in encryption processes.
    scylladb#29615.

  • A soft reboot of a node could affect the retrieval of Azure KMS keys due to DNS timing. The scylla_swap_setup dependency on swap.target was adjusted to ensure reliable key retrieval after node reboots.
    scylladb#1257, scylladb#29517.

Monitoring

  • Stale token metadata diagnostics were improved to provide better insight into system state. These updates offer enhanced visibility into metadata, aiding in more efficient troubleshooting.
    scylladb#29734.

Networking

  • Memory accounting for large responses in the transport layer was refined. Improvements were made to better manage memory usage for these large responses, enhancing system stability under load.
    scylladb#1453, scylladb#29590.

Raft

  • Stale snapshot transfers in the Raft module were properly aborted when the term changed, preventing stale data propagation.
    scylladb#862, scylladb#29378.

  • Improved reliability of Raft state transitions by strengthening error management in state-change notifications.
    scylladb#665, scylladb#29724.

  • The node join logic for Group 0 was adjusted to ensure safe initialization during startup.
    scylladb#29419.

Reliability

  • Improved stability during multi-DC operations by ensuring table metadata remains valid throughout request processing.
    scylladb#2137, scylladb#30169.

  • The partition_snapshot_row_cursor was updated to correctly maintain version entries during a reversed maybe_refresh() operation. scylladb#1552, scylladb#29955.

  • The SSTable parse_assert() logic was updated to validate messages, ensuring stability during SSTable processing.
    scylladb#29727.

  • Prepared-statement invalidation in table_helper was made safe for concurrent operations.
    scylladb#1667, scylladb#29844.

  • The reader concurrency semaphore logic was corrected to ensure accurate semaphore management.
    scylladb#1014, scylladb#29869.

  • Single-key read handling was improved for cleaned-up tablet replicas to increase read reliability.
    scylladb#27147.

Repair

  • Dropping a table while a repair was in progress could lead to column family identification issues. The cluster repair logic was updated to proceed smoothly when a table has been dropped, improving the robustness of administrative operations.
    scylladb#568, scylladb#29280.

REST API

  • REST API behavior during shutdown and cross-shard forwarding was improved for smoother node operations.
    scylladb#29696.

Service Levels

  • Service Levels v2 migration was updated to correctly complete on empty legacy tables, ensuring consistent migration processes.
    scylladb#29353.