Changelog
v1.6.1 (2026-05-23)
🐛 Bug Fixes
- prevent concurrent map read/write when marking jobs stale (#160)
- suspension during state selection (#161)
v1.6.0 (2026-05-22)
✨ Features
- override-until and cli support (#155)
- new TagSelector for AWS-related executors (#151)
- add cache package and revamp notif ratelimit (#146)
- implement caching for notification state (#145)
- support ratelimiter with RPM (#142)
- introduce ratelimiter for notification sink (#140)
- make timestamp output to be local aware (#135)
- revamp data model with Strategy pattern (#131)
- handle pending state on exec/rds (#129)
- introduce cycle id tracker in restore point for idempotency handling (#128)
- cli restore subcommand with stale info (#127)
- refactor restore data flow on runner (#126)
- using keyedworker pool for notification handling (#121)
- sink slack progression on delivery mode of thread (#115)
- sink slack with custom delivery mode (#114)
- sink slack with json format for blocks support (#112)
- add new event for per target execution progress notification (#107)
- implement HibernateNotification CRD feature (#96)
- add Connector context to the notification payload (#102)
- unifying validation webhook path (#97)
- notification lifecycle processor and status callback (#98)
- cmd: add notification subcommand (#101)
- executor: propagate success execution message (#100)
🐛 Bug Fixes
- idempotency handling for deletion-based executor (Karpenter) (#154)
- add poke helper for updated timer (#152)
- worker deadline and idle timer (#150)
- use bigger retry attempts to align with send timeout (#149)
- use LinearJitterBackoff and add retry max capacity (#148)
- ensure idempotency and consistency in ratelimiter (#144)
- implement ratelimit in transport level instead of operation (#143)
- extend retry and timeout on dispatch during Send notification operation (#139)
- ensure auth validation is coming from runner SA (#138)
- [exec/workloadscaler] idempotency on shutdown (#133)
- clarify between capturedAt and reportedAt on restore (#134)
- job should be marked stale on terminal state (#132)
- ensure correct data format passed for execution state (#130)
- [exec/ec2] ensure restore point captures all instances regardless of state (#123)
- inconsistent state handling (#122)
- provide clear information on logs command (#119)
- EC2 executor tolerate on missing instance (#118)
- race condition between execution progress and completed event (#117)
- internal cmd error handler (#110)
- adjust notification lifecycle (#106)
- execution summary also record on error phase (#103)
- fix and refactor planner especially DAG for reverse (#95)
🛠️ Code Refactoring
- restore manager single initialization (#136)
- executor ec2 on wakeup flow to be idempotent (#120)
- outcome messaging with domain-friendly message that include fix to consider stale resource during restore (#116)
- slack auto layout (#113)
- refine execution message on running and failure recovery (#105)
🧹 Miscellaneous
- remove logger from timerset (notify) activity (#153)
- adjust log format on ratelimiter (#147)
- minor correctness (#125)
- cosmetic change like adding TODO marker (#109)
v1.5.0 (2026-03-27)
✨ Features
- removing old controller as well as --legacy-controller flag (#82)
- new subcommand for override and restart action (#79)
- introduce override state handling (#78)
- runner: guarantee restore point on no-op shutdown + pipeline tests (#76)
🐛 Bug Fixes
- the IN attribute in preview should be relative to user time (#81)
- executor: empty restore point is considered no-op (#75)
- executor: ignore notfound error on List API (#70)
- scheduler: advance next event times past exception window boundaries (#87)
🛠️ Code Refactoring
- DAG execution with depedency check (#86)
- enhance provider reconciler (#83)
- scheduler: add support for multi schedule exception (#85)
🧹 Miscellaneous
- rename internal patch to a clearer name and avoid using Error logs for handler terminal (#80)
- refine recovery log (#69)
v1.4.1 (2026-03-09)
🐛 Bug Fixes
- executor: ignore notfound error on List API (#70)
v1.4.0 (2026-03-09)
✨ Features
- controller: Async Phase-Driven Reconciler (#60)
🐛 Bug Fixes
- schedule evaluation off the evaluation for suspend exception (#61)
- cmd: list and preview subcommand should consider exception (#62)
🧹 Miscellaneous
- controller: remove unused metrics and better labeling (#66)
- controller: integrate metrics to the controller (#65)
v1.3.2 (2026-03-09)
🐛 Bug Fixes
- executor: ignore notfound error on List API (#70)
v1.3.1 (2026-03-09)
🐛 Bug Fixes
- schedule evaluation off the evaluation for suspend exception (#61)
- cmd: list and preview subcommand should consider exception (#62)
v1.3.0 (2026-02-27)
✨ Features
- cmd: kubectl-hibernator more subcommand (#51)
🐛 Bug Fixes
- recovery attempt handling and status updater with exclusion support (#58)
- executor: handle notfound and alreadyexists error (#54)
🛠️ Code Refactoring
- move validationwebhook to internal (#57)
v1.2.1 (2026-02-26)
🐛 Bug Fixes
- executor: handle notfound and alreadyexists error (#54)
v1.2.0 (2026-02-25)
✨ Features
- cmd: kubectl-hibernator prototyping (#47)
🐛 Bug Fixes
- internal: handling RDS operation start and stop (#43)
- schedule: add validation for same window (#35)
- scheduler: proper schedule boundary handling (#44)
- webhook: 1-minute wakeup warn on validation webhook (#46)
🧹 Miscellaneous
- pipe all output to stderr for sync-version
v1.1.3 (2026-02-20)
🐛 Bug Fixes
- webhook: 1-minute wakeup warn on validation webhook (#46)
v1.1.2 (2026-02-20)
🐛 Bug Fixes
- internal: handling RDS operation start and stop (#43)
- scheduler: proper schedule boundary handling (#44)
🧹 Miscellaneous
- pipe all output to stderr for sync-version
v1.1.1 (2026-02-14)
🐛 Bug Fixes
- schedule: add validation for same window (#35) (#36)
v1.1.0 (2026-02-13)
✨ Features
- api: new semantic for ExecutionOperationSummary status
- scheduler: add default schedule buffer to 1-minute
- scheduler: support use case for full-day operation
🧹 Miscellaneous
- using make test instead of test-unit
- fail-fast on failing unit test
- bump README.md (for release)
v1.0.0 (2026-02-12)
✨ Features
- readiness check for controller, streaming, and scheduler
- support multi-platform build
- fixed runner SA and EKS token signing
- complete the failing scenario, including the restore management
- initial helm chart release
- runner and tests (#1)
- controller: improve schedule exception lifecycle and deterministic selection
- executor: enrich RDS Parameters to support more usecases
- executor: add logger parameter to executors for streaming operations
- executor: introduce waitConfig semantic for wait on complete scene
- executor: instead of concrete, change to interface based client
- restore: implement quality-aware data preservation with IsLive tracking
- schedule: implement ScheduleException CRD (RFC-0003)
- streaming: implement DualWriteSink with async log streaming and immediate-send streaming clients
🐛 Bug Fixes
- use target reference on handle error recovery phase reset
- reconciler infinite loop due to redundant watch from scheduleexception controller
- streaming client didnt stop gracefully during network failure
- workloadselector now adopt kubernetes LabelSelector
- schedule: fix schedule to not shift the next day, also add time/tzdata for time zone awareness
🛠️ Code Refactoring
- set architectural decision about error handling
- error handling between control plane and data plane (runner)
- await process, rename ID to Id for aws related executor
- refine e2e, replace with more clarity
- decided with simple target naming convention
- most likely refactor the pattern and refine E2E
- api: rename waitConfig to awaitCompletion for clearer intention
- streaming: decouple RestoreManager from controller
- streaming: emit logs with execution context and cache metadata
🧹 Miscellaneous
- core readiness check
- default value for image
- bump actions version
- fix lint
- add docker build for specific component
- move controller package to each context, add more unit test coverage
- bump dependencies
- Major project reorganization, documentation, and implementation updates