0.4.0

Highlights

bluetape4k-leader 0.4.0 improves the leader-election adoption story for the bluetape4k-dependencies 1.3.x train. It adds runnable backend examples, documents backend promotion gates and benchmark evidence, fixes Kubernetes/Fabric8 runtime compatibility, and rebuilds diagrams so lock-state and scheduling flows are understandable from the README.
Related: #487 #488 #483 #484 #485 #498

Backend examples and scheduling flows

• Added DynamoDB scheduled export and ZooKeeper scheduler examples.
• Made the existing Ktor, Spring AOP, batch scheduler, cache warmer, migration gate, and leader-trigger examples easier to navigate with scenario-level diagrams.
• Clarified lock-state routes and semantic colors so readers can distinguish active leadership, standby state, ownership transfer, and backend-specific lock behavior.
  Related: #487 #488 #492 #493 #496 #505 #506

Backend evidence and promotion rules

• Documented preview backend promotion gates so users can tell which backends are stable, preview, or benchmark-backed.
• Captured MongoDB suspend benchmark evidence and auto-extend backend benchmark guidance.
• Restored leader-spring-boot coverage above the release threshold.
  Related: #483 #484 #485 #500

Runtime and compatibility fixes

• Restored Kubernetes example runtime compatibility and isolated the Fabric8 Vert.x 4 runtime needed by K3s tests.
• Replaced deprecated leader APIs and kept compile-test warnings clean.
• Warmed snapshot test dependencies and hardened Central snapshot metadata retries in scheduled workflows.
  Related: #498 #481 #527 #528 #469 #471 #473 #475 #477 #479

Release-train alignment

• Aligned the release line with bluetape4k-bom 1.11.0, refreshed common/OpenTelemetry catalog aliases, and prepared the repo for the coordinated dependencies train.
• Kept published dependency coordinates current in README material.
  Related: #501 #503 #510 #525

Documentation and diagrams

• Rebuilt source-backed diagrams for core, Lettuce, Redisson, Exposed JDBC/R2DBC, Hazelcast, MongoDB, ZooKeeper, etcd, Kubernetes Lease, Consul, Ktor scheduling, Spring Boot, Micrometer, batch scheduler, cache warmer, and Prometheus dashboard flows.
• Added reader-first module architecture and scenario diagrams instead of leaving users to infer behavior from implementation links.
  Related: #493 #495 #496 #505 #506 #507 #543

Contributors

Thanks to everyone who contributed to this release.

• Sunghyouk Bae sunghyouk.bae@gmail.com
• debop sunghyouk.bae@gmail.com
• dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Full Changelog: 0.3.1...0.4.0

0.3.1

Changed

• Updated the bluetape4k-exposed test helper line to 1.10.0 and imported
  the bluetape4k-exposed-bom as an implementation-scoped platform in the
  publishable Exposed leader modules.
• Updated the default bluetape4k dependencies catalog ref to
  catalog/2026-06-01-00 for release workflow alignment.

---

Contributors

• debop sunghyouk.bae@gmail.com

Full Changelog: 0.3.0...0.3.1

0.3.0

Added

• Added Kubernetes Lease per-slot group election support with Nightly-backed
  coverage for the safe slot contract.
• Added framework-neutral leader event callback handles for callers that need
  lifecycle events without binding to Spring or Ktor.
• Added runnable adoption examples for etcd reconcilers, Consul maintenance
  windows, strategic election, virtual-thread leader jobs, and Redisson-backed
  long-running leader jobs.
• Added a runnable virtual-thread leader runner example for high-concurrency
  leader-only maintenance work. (#426)
• Added a runnable Redisson watchdog example for long-running leader-only jobs
  protected by bluetape4k lease auto-extension. (#425)

Changed

• Opened the 0.3.0 development line after publishing 0.2.2.
• Updated public README dependency snippets to the 0.3.0 release coordinate.
• Refreshed README architecture, benchmark, and example workflow diagrams with
  reproducible rendering evidence.
• Aligned the Ktor example and Ktor-facing consumer guidance with shared
  bluetape4k Ktor health modules.
• Made preview backend release gate evidence explicit in the Nightly summary
  documentation and workflow surface. (#417)
• Updated the bluetape4k-exposed test helper reference from 1.9.2-SNAPSHOT
  to the published 1.9.2 release.
• Aligned the AWS SDK BOM with the central bluetape4k-dependencies catalog.
• Consumed the bluetape4k-projects 1.10.0 BOM line.

Fixed

• Applied central catalog-managed Netty 4.1, Protobuf, Fabric8, and Vert.x 4
  dependency overrides for Dependabot security alerts. (#389)
• Forced Gradle plugin classpath MySQL and Protobuf dependencies onto the
  central catalog lines to clear plugin-classpath Dependabot alerts. (#389)
• Removed the unused root apply false Exposed plugin declaration so GitHub
  dependency submission no longer reports its stale plugin POM transitives. (#389)
• Removed the unused Exposed migration plugin from examples/migration-gate;
  the example uses runtime Exposed APIs directly and does not need plugin tasks. (#389)
• Kept Spring group stream support explicitly rejected until per-slot group lease
  extension semantics are defined.

Performance

• Added Redis lease-extension benchmark evidence for Lettuce and Redisson.
• Added SQL backend and preview backend benchmark evidence to make release
  recommendations measurement-backed.

---

Contributors

• Sunghyouk Bae sunghyouk.bae@gmail.com
• debop sunghyouk.bae@gmail.com

Full Changelog: 0.2.2...0.3.0

0.2.2

Added

• Expanded DynamoDB leader election edge-case coverage. (#366, PR #368)
• Added preview backend state snapshot contract coverage for Consul, DynamoDB, and etcd. (#374, PR #381)
• Added KDoc for DynamoDB leader factory classes and extension functions. (#365, PR #369)

Changed

• Updated public README dependency snippets to the 0.2.2 release coordinate. (#375)
• Made the preview backend release gate explicit for Consul, DynamoDB Local, etcd, and Kubernetes Lease; K3s runtime coverage remains in Nightly full. (#376, PR #382)
• Shared the DynamoDB Local test launcher through bluetape4k-testcontainers to remove duplicated container wiring. (#367, PR #370)

Fixed

• Bound Consul blocking wait and acquire cleanup paths to the configured request timeout budget. (#372, PR #379)
• Bound etcd cleanup waits to configured timeout budgets. (#373, PR #380)

---

Contributors

• debop sunghyouk.bae@gmail.com

Full Changelog: 0.2.1...0.2.2

0.2.1

Release 0.2.1

See CHANGELOG.md for details.

Contributors

• debop sunghyouk.bae@gmail.com

Full Changelog: 0.2.0...0.2.1

0.2.0

Added

• leader-k8s Kubernetes Lease backend with K3s integration coverage. (#335)
• leader-etcd etcd v3 backend with blocking, async, and suspend election paths. (#227)
• leader-consul Consul KV/session backend with blocking and suspend leader election. (#345)
• leader-dynamodb AWS DynamoDB backend for conditional-write based leader election. (#228)
• Per-lock StateFlow<LeaderState> support for coroutine-native state observation. (#222)
• DB server-time based lock acquisition support for Exposed JDBC/R2DBC backends. (#39)
• Runnable adoption examples for Kubernetes operator leadership and leader-coordinated rate limiting. (#231, #229)
• K3s Lease integration example coverage. (#248)

Changed

• Replaced internal runBlocking bridge delegates with a pure-suspend ExtendDelegate contract. (#271)
• Updated bluetape4k-exposed test helper references to 1.9.0.
• Updated public README dependency snippets to the 0.2.0 release coordinate.

Fixed

• Reworked Exposed JDBC cancellation handling to rethrow CancellationException instead of swallowing it through runCatching. (#304)
• Replaced wall-clock wait deadlines with monotonic timeout budgets in Exposed JDBC, MongoDB, and Lettuce acquisition loops. (#305, #306, #308, #309)
• Fixed Kubernetes Lease annotation conversion to avoid mutating the original Lease instance in-place. (#347)
• Added missing Kubernetes Lease suspend elector K3s scenarios. (#349)
• Synchronized Korean README content with observability and Ktor management route documentation. (#287)
• Expanded Ktor management registry KDoc and translated WebhookPoller public documentation/comments to English. (#288, #348)

Removed

• Removed all APIs deprecated before the 0.1.0 GA line. (#269)

---

Contributors

• Sunghyouk Bae sunghyouk.bae@gmail.com
• debop sunghyouk.bae@gmail.com
• dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Full Changelog: 0.1.0...0.2.0

0.1.0

First public release of bluetape4k-leader. All APIs are stable unless noted as experimental.

Breaking Changes

• runIfLeader() never throws on lock contention: returns null instead of propagating
  backend exceptions when the lock is not acquired — aligns with ShedLock skip-on-contention
  semantics. CancellationException and InterruptedException are still rethrown. (PR #15)

  // Null means "not elected" or "action threw" — check logs
  val result = leaderElector.runIfLeader("job") { riskyWork() }
  
  // To preserve exception propagation, wrap in the action:
  leaderElector.runIfLeader("job") {
      try { riskyWork() } catch (e: MyException) { handleError(e); throw e }
  }

• leader-exposed-jdbc: ExposedJdbcLeaderElector.runIfLeader() now returns null instead
  of rethrowing action exceptions. CancellationException and InterruptedException are still
  rethrown. (issue #50)

• leader-exposed-r2dbc: Same change for ExposedR2DbcSuspendLeaderElector.runIfLeader().
  CancellationException is still rethrown.

• leader-exposed-jdbc / leader-exposed-r2dbc: Elector factories now accept an optional
  historyRecorder parameter (SafeLeaderHistoryRecorder? / SuspendSafeLeaderHistoryRecorder?).
  The previous recordHistory option in election options is superseded.

• LeaderElection / LeaderGroupElection renamed to LeaderElector / LeaderGroupElector
  for consistency across all interfaces. (PR #106, #123, #125)

• Duration APIs: migrated from java.time.Duration to kotlin.time.Duration. (PR #126)

• LeaderElectionEvent.Elected now carries optional leaderId: String? and
  leaseExpiry: Instant? fields (both default to null). Compiled callers that construct
  Elected(lockName) positionally will fail to link against new bytecode; recompilation is
  required (source-compatible).

• Spring Boot 3/4 split consolidated: leader-spring-boot-common, leader-spring-boot3, and
  leader-spring-boot4 merged into a single leader-spring-boot module. (PR #105)

Added

leader-core — Core interfaces and local in-process implementations:

• LeaderElector — blocking single-leader interface
• AsyncLeaderElector — CompletableFuture-based async interface
• VirtualThreadLeaderElector — virtual-thread-per-election interface
• SuspendLeaderElector — Kotlin coroutine suspend interface
• LeaderGroupElector — blocking multi-leader (semaphore) interface
• SuspendLeaderGroupElector — coroutine multi-leader interface
• LeaderElectionOptions(waitTime, leaseTime) — shared option data class
• LeaderGroupElectionOptions(maxLeaders, waitTime, leaseTime) — group option data class
• Local implementations: LocalLeaderElector, LocalLeaderGroupElector,
  LocalSuspendLeaderElector, LocalSuspendLeaderGroupElector, LocalAsyncLeaderElector,
  LocalVirtualThreadLeaderElector
• LockAssert — assertLocked() / assertLocked(lockName) / isLocked() and suspend variants
• LockExtender — extendActiveLock(Duration): Boolean + detailed sealed ExtendOutcome result
  • suspend variants
• LeaderLockHandle sealed class (Real / FailOpen) — explicit lease handle
• LeaderLeaseAutoExtender — periodic background lease renewal with shutdown() / restart()
• ListeningLeaderElector / ListeningLeaderGroupElector — listener-aware decorators with a hot
  events: Flow<LeaderElectionEvent> stream (issue #40, PR #146)
• TenantScopedLeaderElectors — forTenant(tenantId) extension functions for multi-tenant
  lock-name scoping (issue #42)
• Strategic election API (issue #29, #31, #32):
  • CandidateInfo, ElectionStrategy, CandidateScorer interfaces
  • Built-in strategies: FifoElectionStrategy, RandomElectionStrategy, ScoredElectionStrategy
  • Built-in scorers: IdleTimeScorer, SuccessRateScorer, RecentSuccessScorer, WeightedScorer
  • Redis CandidateRegistry (Redisson sorted-set/hash + TTL; Lettuce variant included)
• LeaderSlot audit identity propagation: LeaderSlot(lockName, leaderId) propagates to
  LeaderRunResult.Elected.leaderId (issue #72)

leader-redis-lettuce — Lettuce Redis backend:

• LettuceLeaderElector — blocking, uses SET NX PX via LettuceLock
• LettuceLeaderGroupElector — blocking multi-leader via LettuceSlotTokenGroup (ZSET + Lua TTL)
• LettuceSuspendLeaderElector — coroutine single-leader
• LettuceSuspendLeaderGroupElector — coroutine multi-leader
• LettuceLock, LettuceSuspendLock — Redis distributed lock primitives (self-contained, no bluetape4k-lettuce dependency)

leader-redis-redisson — Redisson Redis backend:

• RedissonLeaderElector — blocking via RLock.tryLock()
• RedissonLeaderGroupElector — blocking multi-leader via RPermitExpirableSemaphore
• RedissonSuspendLeaderElector — coroutine single-leader with PID-seeded Snowflake lock ID
• RedissonSuspendLeaderGroupElector — coroutine multi-leader

leader-exposed-core: Shared Exposed table DDL (LeaderLockTable, LeaderGroupLockTable) for
JDBC and R2DBC backends. (issue #23)

leader-exposed-jdbc — Exposed JDBC blocking backend (issue #21, PR #52):

• ExposedJdbcLeaderElector — blocking single-leader (sync + CompletableFuture async)
• ExposedJdbcLeaderGroupElector — blocking multi-leader
• H2/PostgreSQL/MySQL Testcontainers integration tests

leader-exposed-r2dbc — Exposed R2DBC coroutine backend (issue #22, PR #62):

• ExposedR2DbcSuspendLeaderElector — coroutine single-leader
• ExposedR2DbcSuspendLeaderGroupElector — coroutine multi-leader
• R2DBC PostgreSQL Testcontainers integration tests

leader-mongodb — MongoDB backend (issue #8, PR #46):

• MongoLeaderElector — blocking, findOneAndUpdate upsert + deleteOne(token) owner-only release
• MongoSuspendLeaderElector — coroutine, Kotlin coroutine driver
• MongoLeaderGroupElector — blocking multi-leader (lockName:slot:N slot model)
• MongoSuspendLeaderGroupElector — coroutine multi-leader (dual-collection design)
• TTL index auto-creation at startup
• 82.4% line coverage (42 tests, Testcontainers MongoDB)

leader-hazelcast — Hazelcast IMap token-lock backend (issue #9):

• HazelcastLeaderElector, HazelcastLeaderGroupElector — blocking single/multi-leader
• HazelcastSuspendLeaderElector, HazelcastSuspendLeaderGroupElector — coroutine variants

leader-zookeeper — ZooKeeper/Curator backend. (PR #138)

leader-micrometer — Micrometer metrics integration:

• MicrometerLeaderElectionListener records leader.election.events with lock.name and event tags (issue #40, PR #146)

leader-spring-boot — Spring Boot 4 auto-configuration + AOP:

• @LeaderElection / @LeaderGroupElection annotations with AspectJ compile-time weaving (CTW)
• Supports suspend, Mono, Flux, and Kotlin Flow return types (#74, #90, #91)
  • @LeaderElection(streamBounded = true) explicit opt-in for bounded streams
• LeaderAnnotationValidatorBeanPostProcessor — startup validation; blocks unsafe return types
  (CompletableFuture, Deferred, etc.) (PR #79)
• LockAssert / LockExtender integration in AOP aspect (issue #79)
• LeaderMetricsHealthIndicator — Spring Boot Actuator health indicator (registered as leaderMetricsHealthIndicator)
• LeaderLeaseAutoExtenderLifecycle — context-lifecycle-aware auto-extender integration
• Backend auto-configurations: Lettuce, Redisson, Exposed JDBC, Exposed R2DBC, MongoDB, Hazelcast
• LeaderProperties — bluetape4k.leader.* configuration properties

leader-ktor — Ktor 3.x integration (issue #37, PR #164):

• LeaderElectionPlugin — createApplicationPlugin DSL, SuspendLeaderElector-based
• Application.leaderScheduled(lockName, period) { } — Spring @Scheduled-style leader-only
  periodic task helper; auto-cancelled on ApplicationStopped

leader-bom — Bill of Materials for consumers. All leader-* modules included; BOM users
do not need to specify individual versions.

examples/ — Runnable example applications (issue #36):

• batch-scheduler — Lettuce Redis periodic batch single-execution (PR #159)
• migration-gate — Exposed JDBC boot-time schema migration gate (PR #160)
• webhook-poller — MongoDB single-instance webhook polling (PR #161)
• cache-warmer — Hazelcast per-partition leader cache warming (PR #162)
• tenant-aggregator — Exposed R2DBC coroutine multi-tenant aggregation (PR #163)
• ktor-app — Ktor 3.x + Lettuce Redis leaderScheduled() demo (PR #166)
• prometheus-dashboard — Spring Boot + Prometheus/Grafana leader metrics dashboard

CI/CD (issue #13, #35, PR #19, #20, #44, #135):

• GitHub Actions build, test, secret-scan, Gradle wrapper validation
• Nightly SNAPSHOT auto-publish (on test success only)
• Parallel test jobs for Lettuce and Redisson backends

Changed

• LeaderElectionAspect / LeaderGroupElectionAspect outer catch narrowed to Exception (not
  Throwable) to let OutOfMemoryError / StackOverflowError propagate.
• LeaderElectionOptions, LeaderGroupElectionOptions, LeaderGroupState validate eagerly in
  init {} (waitTime ≥ 0, leaseTime > 0, maxLeaders ≥ 1). (PR #25)
• ExposedJdbcGroupLock.tryLock() return type changed to Boolean?: null = DB error,
  false = slot contention, true = acquired. (issue #60, PR #63)
• Suspend tests use runBlocking(Dispatchers.IO) instead of runTest for real-IO
  (MongoDB / Testcontainers) tests.
• CI uses paths-filter and retry configuration to reduce unnecessary work and transient failures.
  (PR #135)
• Prometheus export coverage verified through PrometheusServer scrape tests. (PR #144)
• leader-bom NMCP aggregation and Central Snapshots publishing fixed. (PR #140)
• factory create() I/O failures now respect the configured failure mode. (PR #107)
• @ConfigurationProperties added to AOP properties binding where missing. (PR #93)

Fixed

• **Coroutine cancellation...