Skip to content

Latest commit

 

History

History
204 lines (159 loc) · 11.7 KB

CHANGELOG.md

File metadata and controls

204 lines (159 loc) · 11.7 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased

v2.3.1 - 2023-12-14

Changed

  • Improve snapshot synchronization diagnostic logging PR#216

Fixed

  • Consecutive snapshot synchronizations could not synchronize new entity snapshots #215, PR#217

v2.3.0 - 2023-06-19

Added

Changed

  • Rollback deletes only tagged events after the given sequence number #202, PR#203
  • CassandraPersistenceQueries can find the highest sequence number even if deleted partition exists #207, PR#209
  • CassandraPersistenceQueries.currentEventsBefore skips deleted partitions #208, PR#209
  • Rollback preparation fails if required data have been deleted #206, PR#210

Fixed

  • Dead letters about ReplicationRegion$Passivate continue #199, PR#200

v2.2.0 - 2022-012-26

Added

Changed

  • Enhance leader's replication response handling PR#160
  • Change event sourcing log level to debug PR#163
  • ReplicationRegionRaftActorStarter uses its FQCN as its logger name PR178
  • Add diagnostic info to logs of sending replication results PR#179
  • Persist EntitySnapshot as an event PR#184
  • Allow only specific actors defined as a sticky leader to become a leader PR#186, PR#189

Fixed

  • RaftActor might delete committed entries #152 #165 PR#151 PR#166
    ⚠️ This fix adds a new persistent event type. It doesn't allow downgrading after being updated.
  • An entity on a follower could stick at WaitForReplication if the entity has a ProcessCommand in its mailbox #157, PR#158
  • Leader cannot reply to an entity with a ReplicationFailed message in some cases #153, PR#161, #170, PR#171
  • An entity could stick at WaitForReplication when a Raft log entry is truncated by conflict #155, #PR162
  • A RaftAcotor(Leader) could mis-deliver a ReplicationSucceeded message to a different entity 156, #PR162
  • Snapshot synchronization could remove committed log entries that not be included in snapshots #167 #PR168
  • SnapshotStore doesn't reply with SnapshotNotFound sometimes #182, #PR183

v2.1.0 - 2022-03-24

Added

  • Efficient recovery of commit log store, which is on the query side #112

    This change will improve the performance of the recovery on the query side. You should migrate settings described at Migration Guide.

  • Raft actors start automatically after an initialization of ClusterReplication #118

    This feature is enabled only by using typed.ClusterReplication. It is highly recommended that you switch using the typed API since the classic API was deprecated.

  • Raft actors track the progress of the event sourcing #136, PR#137, PR#142.

    This feature ensures that

    • Event Sourcing won't halt even if the event-sourcing store is unavailable for a long period. After the event-sourcing store recovers, Event Sourcing will work again automatically.
    • Compaction won't delete committed events that are not persisted to the event-sourcing store yet.

    It adds new following settings (for more details, please see reference.conf):

    • lerna.akka.entityreplication.raft.eventsourced.committed-log-entries-check-interval
    • lerna.akka.entityreplication.raft.eventsourced.max-append-committed-entries-size
    • lerna.akka.entityreplication.raft.eventsourced.max-append-committed-entries-batch-size

    It deletes the following settings:

    • lerna.akka.entityreplication.raft.eventsourced.commit-log-store.retry.attempts
    • lerna.akka.entityreplication.raft.eventsourced.commit-log-store.retry.delay

    It requires that lerna.akka.entityreplication.raft.compaction.preserve-log-size is less than lerna.akka.entityreplication.raft.compaction.log-size-threshold.

  • Compaction warns if it might not delete enough entries PR#142

Changed

  • Bump up Akka version to 2.6.17 PR#98

    This change will show you deserialization warnings during the rolling update, it's safe to ignore. For more details, see Akka 2.6.16 release note

Fixed

  • TestKit throws "Shard received unexpected message" exception after the entity passivated PR#100

  • ReplicatedEntity can produce illegal snapshot if compaction and receiving new event occur same time #111

  • Starting a follower member later than leader completes a compaction may break ReplicatedLog of the follower #105

  • The Raft leader uses the same previous LogEntryIndex and Term to all batched AppendEntries messages #123

  • Raft Actors doesn't accept a RequestVote(lastLogIndex < log.lastLogIndex, lastLogTerm > log.lastLogTerm) message #125

  • A new event is created even though all past events have not been applied #130

  • InstallSnapshot can miss snapshots to copy PR#128

    ⚠️ This change adds a new persistence event. This might don't allow downgrading after upgrading.

  • Moving a leader during snapshot synchronization can delete committed log entries #133

    ⚠️ This change adds a new persistence event. This might don't allow downgrading after upgrading.

v2.0.0 - 2021-07-16

Breaking Change

  • Change the shard-distribution-strategy to distribute shard (RaftActor) more evenly PR#82

    ⚠️ This change does not allow rolling updates. You have to update your system by stopping the whole cluster.

  • Made internal APIs private

    If you are only using the APIs using in the implementation guide, this change does not affect your application. Otherwise, some APIs may be unavailable. Please see PR#47 to check APIs that will no longer be available.

Added

  • Java11 support
  • Add new typed API based on Akka Typed PR#79
    • This API reduces runtime errors and increases productivity.

Deprecated

  • Untyped (classic) API has been deprecated PR#96

    ⚠️ This API will be removed in the next major version release.

v1.0.0 - 2021-03-29

  • GA release 🚀

v0.1.0 - 2021-01-12

  • Initial release (under development)