What risingwavelabs/risingwave shipped
Written by FoxPlug from the repository's own releases; not affiliated with Risingwave. Paste your own repository in the box below and FoxPlug writes the same weekly update for your product. This page is an automatic summary of the releases, pull requests and commits in github.com/risingwavelabs/risingwave; Risingwave did not write it and does not use or endorse FoxPlug. Every line links to the change it describes.
Get a weekly update like this for your product, free
Week of October 5, 2026
What shipped
- Iceberg V3 can now be maintained as an independent job in the meta service. Pull request #26981
- Pulsar sources can use native Avro schemas, inferring columns and decoding messages written with different schema versions. Pull request #26347
- A bounded-memory sort buffer for the locality provider was added to mitigate join amplification. Pull request #26616
- Table caches can now be warmed up explicitly. Pull request #27233
- Iceberg ADLS file IO configuration is shared. Pull request #27407
- Variant access paths are now prebuilt to speed up expression evaluation. Pull request #26530
- Upstream rows of finished vnodes are forwarded after a backfill reschedule. Pull request #27384
- Sinks keep the old primary key delete when an update changes the sink primary key, so stale rows are not retained. Pull request #27417
- SQL Server CDC now validates table name casing. Pull request #27411
- Time travel is rejected in streaming jobs. Pull request #27423
Why it matters
This week brings new capabilities for Pulsar and Iceberg users, and a set of fixes that affect correctness in everyday streaming work: sink output when a primary key changes, recovery during CDC startup, and planner behavior on indexes. Users running MySQL and SQL Server CDC should see more reliable recovery and schema handling. The bounded-memory sort buffer and explicit cache warm-up address resource behavior under load.
Changelog entry
- Reduce the default CDC heartbeat interval to 60 seconds. Pull request #27453
- Synchronize catalog versions with frontend notifications. Pull request #27408
- Compare MySQL schema history positions by binlog coordinates. Pull request #27405
- Cancel CDC engine startup when the reader is dropped. Pull request #27418
- Reject ambiguous correlated column references in the binder. Pull request #27242
- Support DROP CONNECTOR on tables with more than one internal table. Pull request #27363
- Validate SQL Server table name casing in CDC. Pull request #27411
- Fill secrets before validating altered sink properties. Pull request #27183
- Reject time travel in streaming jobs. Pull request #27423
- Keep the old primary key delete when an update changes the sink primary key. Pull request #27417
- Maintain Iceberg V3 as an independent job. Pull request #26981
- Support native Pulsar Avro schemas. Pull request #26347
This week: Iceberg V3 as an independent job, native Pulsar Avro schemas, a bounded-memory sort buffer for the locality provider, explicit table cache warm-up, and fixes across CDC recovery, sink output on primary key changes, and streaming time travel.
This week in the project: Iceberg V3 can be maintained as an independent job in the meta service. Pulsar sources now support native Avro schemas, so they can infer columns and decode messages written with different schema versions. A bounded-memory sort buffer for the locality provider was added to mitigate join amplification, and table caches can now be warmed up explicitly. Iceberg ADLS file IO configuration is shared, and variant access paths are prebuilt. On the fix side, sinks keep the old primary key delete when an update changes the sink primary key, upstream rows of finished vnodes are forwarded after a backfill reschedule, SQL Server CDC validates table name casing, and time travel is rejected in streaming jobs.
Week of September 28, 2026
What shipped
- Upgraded Foyer cache library to 0.22.6 with recovery backport to improve storage performance. Pull request #27334
- Fixed object store metadata error handling during listing operations to properly propagate missing modification times. Pull request #27266
- Force-close PostgreSQL connections on executor shutdown to prevent blocked threads and release replication slots. Pull request #27190
- Force-close SQL Server connections on CDC shutdown to ensure clean resource cleanup. Pull request #27186
- Backported CDC streaming readiness and SQL Server shutdown fixes to improve changefeed reliability. Pull request #27358
- Set rest.auth.type explicitly for Iceberg JNI REST catalog integration. Pull request #27221
- Cherry-picked object store metadata error propagation fix to release-3.0 branch. Pull request #27339
- Refactored metadata layer to unify independent checkpoint job handling. Pull request #26980
Why it matters
This week focused on CDC source stability and resource cleanup, with improvements to PostgreSQL and SQL Server connection handling. Storage performance was enhanced through a Foyer upgrade, and object store listing reliability was fixed to properly handle metadata errors.
Changelog entry
- Perf(storage): upgrade foyer to 0.22.6 with recovery backport Pull request #27334
- Fix(object-store): propagate metadata errors during listing Pull request #27266
- Fix(cdc/postgres): force-close PG connections on stop Pull request #27190
- Fix(cdc): force-close SQL Server connections on shutdown Pull request #27186
- Fix(cdc): backport streaming readiness and SQL Server shutdown fixes Pull request #27358
- Fix(iceberg): set rest.auth.type explicitly for JNI REST catalog Pull request #27221
- Refactor(meta): unify independent checkpoint job handling Pull request #26980
Weekly digest: improved CDC connection shutdown for PostgreSQL and SQL Server, upgraded storage cache library, fixed object store metadata error handling, and refactored checkpoint job coordination.
This week's RisingWave updates improve operational stability and performance. CDC sources now properly force-close PostgreSQL and SQL Server connections during shutdown, preventing thread hangs and resource leaks. Storage performance improved via Foyer 0.22.6 upgrade. Object store listing now correctly propagates metadata errors. Additional checkpoint job coordination improvements in the metadata layer.
Week of September 21, 2026
What shipped
- Version 3.1.0 released. Release
- Broadcast temporal join support added to stream processing (cherry-picked to release-3.1). Pull request #27262
- Keyed changelog planning and execution support backported to release branches. Pull request #27253
- Oracle CDC source connector plumbing added as first step of Oracle CDC implementation. Pull request #26903
- Iceberg compaction health metrics exposed in
rw_iceberg_compaction_schedules. Pull request #27226 - Meta fix to preserve connection and secret references during connector alterations backported to release-3.1. Pull request #27200
- Table refresh now completes only after all materialize actors finish and abandons on recovery. Pull request #27041
- Legacy row format source syntax removed from SQL parser. Pull request #27212
- MATCH_RECOGNIZE scan fixed to remain incomplete across eviction rebase. Pull request #27201
- PostgreSQL query function now requires source SELECT privilege (cherry-picked to release-3.1). Pull request #27218
Why it matters
Version 3.1.0 ships with new streaming capabilities including broadcast temporal joins and keyed changelog support. Multiple stability fixes and new connector features expand Risingwave's capabilities for CDC and data integration use cases.
Changelog entry
- Stream: broadcast temporal join support Pull request #27262
- Stream: keyed changelog planning and execution support Pull request #27253
- Meta: preserve connection and secret references during connector alterations Pull request #27200
- Refresh: complete table refresh only after all materialize actors finish Pull request #27041
- Connector: Oracle CDC source plumbing Pull request #26903
- Iceberg: expose compaction health in
rw_iceberg_compaction_schedulesPull request #27226 - Sqlparser: remove legacy row format source syntax Pull request #27212
- Stream: keep MATCH_RECOGNIZE scan incomplete across eviction rebase Pull request #27201
- Stream: derive NOW progress from elapsed time Pull request #27209
- Frontend: enforce source SELECT privilege for
postgres_queryPull request #27218
Risingwave 3.1.0 is out. New: broadcast temporal joins, keyed changelog execution, Oracle CDC plumbing, Iceberg compaction health metrics. Plus stability fixes across streams and refresh handling.
Risingwave 3.1.0 is available. This release adds broadcast temporal join support for stream processing, keyed changelog planning and execution, and Oracle CDC connector plumbing. New Iceberg compaction health metrics are now exposed, and multiple stability improvements cover table refresh handling, MATCH_RECOGNIZE scans, and source privilege enforcement.
Week of September 14, 2026
What shipped
- Released v3.0.4 with fixes for Iceberg, streaming sinks, connectors, and metadata handling Release
- Google Pub/Sub sources and sinks can now use Application Default Credentials for authentication Pull request #26808
- General over-window executor now forwards watermarks to improve streaming query latency Pull request #27050
- Automatic compaction group split and merge is now stabilized in metadata operations Pull request #27044
- Iceberg now supports per-table metadata relations for improved schema introspection Pull request #27106
- SQL Server connectors now preserve composite primary key column order correctly Pull request #27160
- Compaction picker performance improved by reducing scans and memory allocations Pull request #27081
- Query optimizer now prevents unsafe correlated aggregate decorrelation that could drop results Pull request #27065
- Memory accounting no longer drifts due to untracked allocations and chunk retention Pull request #27069
- Stream barrier interval now uses database configuration for consistent recovery timing Pull request #27118
Why it matters
V3.0.4 release brings stability improvements to compaction, streaming recovery, and connector handling. Watermark propagation through over-window operators improves latency for window-based queries. Better authentication support for Google Pub/Sub and fixes for primary key ordering make connectors more reliable.
Changelog entry
- Feat(connector): Google Pub/Sub sources and sinks support Application Default Credentials Pull request #26808
- Feat(stream): Forward watermarks through general over window operators Pull request #27050
- Feat(iceberg): Support per-table metadata relations Pull request #27106
- Feat(stream): Apply backfill rate limit to locality provider drain phase Pull request #27020
- Fix(meta): Stabilize automatic compaction group split and merge Pull request #27044
- Fix(connector): Preserve SQL Server composite primary key order Pull request #27160
- Fix(stream): Preserve CDC events during recovery startup Pull request #27031
- Fix(stream): Use database barrier interval for consistent timing Pull request #27118
- Fix(optimizer): Prevent unsafe correlated aggregate decorrelation Pull request #27065
- Fix(common): Prevent memory accounting drift from untracked allocations Pull request #27069
V3.0.4 shipped with compaction stability, watermark improvements in over-window queries, Google Pub/Sub ADC auth, and connector fixes for SQL Server primary keys.
V3.0.4 released with compaction group stabilization, watermark forwarding through over-window operators for reduced latency, Google Pub/Sub Application Default Credentials support, and SQL Server composite primary key ordering fixes.