[RELEASE] ScyllaDB 2026.1.5

The ScyllaDB team is pleased to announce the release of ScyllaDB 2026.1.5, a production-ready patch release for ScyllaDB 2026.1 Long Term Support (LTS) Release.

More information on ScyllaDB Version Support Policy is available here.

Related Links

Bug Fixes

The following issues are fixed in this release.

Compaction

  • The compaction module could remain unexpectedly registered and active during an early node shutdown sequence. The system now ensures that the compaction module is properly released when a node is stopped shortly after startup, guaranteeing a clean shutdown process and preventing potential resource issues.
    scylladb#30091

Installation

  • ScyllaDB was previously unable to boot on certain Non-XFS filesystems when they were mounted with a 4096-byte logical block size. An update to the underlying platform libraries resolves a disk alignment issue, ensuring successful node boot and operation on these filesystems.
    bf743d8

Load Balancing

  • The node drain process could behave unexpectedly when forced capacity-based balancing was enabled. The balancing logic is corrected to function properly during node draining, allowing administrators to reliably remove nodes from the cluster while preserving the desired capacity distribution.
    scylladb#29885

Schema

  • The committed_by_group0 property was not consistently set for all non-system tables, leading to a schema inconsistency. The schema logic is updated to ensure this property is consistently applied, which enhances cluster-wide schema consistency.
    scylladb#30065

Service Levels

  • Service levels could remain in a stalled V1-to-V2 migration state after an upgrade, leaving workload prioritization settings partially migrated. The system now includes self-healing logic that automatically completes any incomplete migration, ensuring service levels operate correctly immediately after the upgrade.
    scylladb#29935

Stability

  • A memory resource lifetime issue was observed during multi-DC replication. The update improves stability during multi-datacenter operations by ensuring consistent table object lifetime management
    scylladb#30074

  • A potential startup stability issue could occur if a node was starting up following an interrupted resharding operation combined with a Materialized View build. An update to the underlying framework resolves this, improving node resilience during complex recovery operations.
    scylladb#29952

Tablets

  • Lightweight Transactions (LWT) or serial statements could lose the TABLETS_ROUTING_V1 payload following an internal coordinator shard restart. ScyllaDB now consistently returns this routing payload, which improves client-side routing accuracy and compatibility following internal state changes.
    scylladb#29963

Vector Search

  • Precision loss was occurring when using decimal or varint types within the filter logic for vector search operations. The updated precision handling ensures that data of these types maintains its full expected accuracy when filtered.
    scylladb#29956