The ScyllaDB team announces ScyllaDB Open Source 5.4.6, a bugfix release of the ScyllaDB 5.4 stable branch. ScyllaDB Open Source 5.4.6, like all past and future 5.x.y releases, is backward compatible and supports rolling upgrades.
Users are encouraged to upgrade to 5.4.6.
Related links:
- ScyllaDB Open Source 5.4.0
- Get ScyllaDB Open Source - AWS AMI, GCP Docker, binary packages and unified installer
- ScyllaDB Web Installer for Linux (all releases)
- Upgrade from ScyllaDB Open Source 5.x.y to Scylla Open Source 5.x.z
- Upgrade from ScyllaDB Open Source 5.2.x to 5.4.y
- Please let us know if you encounter an issue.
Issue fixed in this release:
- Stability: a rare race condition in gossiper causes nodes to fail observing updated application states of other nodes in gossiper. This as flaky tests, but not as far as we know in production, #15393, #15602, #16668, #16902, #17493, #18118
- Stability: database: view_update_generator is unplugged before writes are drained spamming the log when stopping it. #16001
- Stability: Drop bloom filters from memory, if their total memory consumption is above the configured limit. #17747
- Stability: Materialized View (MV) flow control doesn’t account small view updates properly, and can lead to OOM #17364
- Stability: reclaim_timer::report() is marked noexcept, but its implementation throws #18099
- Stability: repair: memory accounting does not include keys #16710
- Stability: The estimated_partitions is overestimated for repair #18140
- Stability: A rare case of use-after-free in reverse-read workloads #18075
- Tracing: Large partition/row/cell warnings should not print keys contents #18041