What scylladb/scylladb shipped
Generated by FoxPlug from public activity — not affiliated with Scylladb.
Week of September 14, 2026
What shipped
- Implemented inter-node tablet migration for strongly consistent tables by modifying raft group configuration during tablet migration stages. Pull request #29446
- Extended tablet tests to run against local, S3, and GCS storage to catch regressions in tablet migration, split/merge, truncate, and LWT across different storage backends. Pull request #30569
- Decoupled statement_restrictions to allow different statement types (SELECT, UPDATE, DELETE, CREATE MATERIALIZED VIEW, INSERT) to use different parameter combinations. Pull request #31632
- Added ExportTableToPointInTime command parsing and validation for Alternator with placeholder status returns. Pull request #30222
- Fixed secondary index reads to re-validate indexed predicates on base rows to prevent returning rows that violate WHERE clauses. Pull request #30683
- Converted shard repair task to the generic task implementation, continuing effort to replace per-operation task subclasses. Pull request #31551
- Added metrics to key cache and eagerly add keys to ID cache for encryption operations. Pull request #31547
- Added validation of compaction sub-options for materialized views and indexes to prevent node crashes during schema commit. Pull request #31616
- Fixed table size estimation logic in vnodes-to-tablets migration to correctly estimate on-disk size and determine appropriate tablet counts. Pull request #31658
- Enforced streaming concurrency cap for RF-change rebuilds in the tablet load balancer to prevent exceeding capacity of source replicas. Pull request #31425
Why it matters
This week includes significant improvements to tablet-aware restore, strongly consistent table migration, and storage backend testing. Multiple fixes address crashes and correctness issues in view building, secondary index reads, and backup operations across different storage systems. These changes improve stability and correctness for users running tablets with various replication factors and storage configurations.
Changelog entry
- Strong consistency: modify raft group config during tablet migration to synchronize with new tablet replica set Pull request #29446
- Test: run tablet tests over local, S3 and GCS storage for comprehensive backend coverage Pull request #30569
- CQL3: decouple statement_restrictions users to allow different statement types to use different parameter combinations Pull request #31632
- CQL3: re-validate indexed predicate on base rows read via secondary index Pull request #30683
- Repair: convert shard repair task into generic task implementation Pull request #31551
- Encryption: add metrics to key cache and eagerly add keys to ID cache Pull request #31547
- CQL3: validate compaction sub-options for views and indexes Pull request #31616
- Vnodes-to-tablets migration: fix table size estimation logic for correct tablet count calculation Pull request #31658
- Alternator: add ExportTableToPointInTime command parsing and validation Pull request #30222
- Tablet load balancer: enforce streaming concurrency cap for RF-change rebuilds Pull request #31425
Week 40: tablet-aware restore improvements, strongly consistent table migration, extended storage testing for tablets, secondary index predicate validation, and generic task conversion. Multiple stability and correctness fixes.
Week 40 shipped significant tablet improvements: inter-node migration for strongly consistent tables, tablet test coverage for S3/GCS storage, and tablet-aware restore fixes. Also improved secondary index correctness, extended view building stability, and continued generic task migration. Multiple backports address crashes in backup operations, streaming concurrency, and collection element handling.