v0.22.0

Features

• support k8s-1.25, remove support for k8s-1.20 (#150) #150 (Ellis Tarn)

Bug Fixes

• remove machine from webhook-gen (#149) #149 (Ellis Tarn)

Tests

• Add a cacheSyncingClient for testing that allows for custom FieldIndexers (#145) #145 (Jonathan Innis)

Chores

• Migrate Machine to v1alpha5 API (#143) #143 (Jonathan Innis)
• move drift controller to karpenter-core (#151) #151 (Nick Tran)
• Create state.Node interface methods to access cluster state details (#152) #152 (Jonathan Innis)

Commits

• 15a2378: Fix status not updating when the spec has an update (#136) (Jonathan Innis) #136
• 913efba: Short requeue to check for emptiness rather than using an observer (#139) (Jonathan Innis) #139
• 1fd0f13: Fix README build status badge (#147) (Jonathan Innis) #147
• 78364b2: Fix MachineNotFound custom error type checking (#148) (Jonathan Innis) #148

v0.21.0

Bug Fixes

• Retrigger provisioning until all pods are scheduled (#130) #130 (Ellis Tarn)
• Fix bug where metric gauges are not deleted on pod/provisioner delete (#133) #133 (Jonathan Innis)

Chores

• add in nodeToMachine function (#124) #124 (Nick Tran)
• reduce log spam (#129) #129 (Todd Neal)
• Add github issue templates (#135) #135 (Nick Tran)
• Register the Machine CR with the scheme (#125) (Jonathan Innis) #125
• rename env-variable PORT to WEBHOOK_PORT (#128) (Serg Kondrashov) #128
• exclude karpenter.sh_machines.yaml (#126) (Ryan Maleki) #126
• Use a machine generated from a node for the termination loop (#132) (Jonathan Innis) #132
• Remove event filters and move logic back into reconcile (#134) (Jonathan Innis) #134

v0.0.2

Features

• Migrate settings to knative injection watcher (#17) #17 (Jonathan Innis)
• Combine webhooks and controllers into a single binary (#29) #29 (Ellis Tarn)
• multi-node consolidation (#72) #72 (Todd Neal)
• allow node annotations to be set by the provisioner (#87) #87 (Ellis Tarn)
• added drift deprovisioner (#93) #93 (Nick Tran)
• add more deprovisioning logs (#109) #109 (Todd Neal)
• add karpenter-core and karpenter-core-crd helm charts (#113) #113 (Ryan Maleki)

Bug Fixes

• Inject settings after the batcher completes waiting (#31) #31 (Jonathan Innis)
• move emptiness tests to deprovisioning and deduplicate deletion logic (#52) #52 (Nick Tran)
• Wait on .status.Capacity and .status.Allocatable in Initialization (#69) #69 (Jonathan Innis)
• track pending pods when consolidating (#79) #79 (Todd Neal)
• update randomdelay to match new version of Ginkgo (#81) #81 (Todd Neal)
• Reduced webhook grace period for faster restarts (#92) #92 (Ellis Tarn)
• Removed spurious error logging (#91) #91 (Ellis Tarn)
• Fix bugs/PDB do not considered namespace (#89) #89 (Clicia Scarlet)
• add our volume selector zone requirement to all terms (#104) #104 (Todd Neal)
• ignore pods with ephemeral volumes using an invalid storage class name (#106) #106 (Todd Neal)
• fix file perm and remove extra labels (#117) #117 (Ryan Maleki)
• fix CRD symlink (#119) #119 (Ryan Maleki)

Tests

• add a discovery label to all test objects for cleanup (#27) #27 (Ellis Tarn)
• trigger CI based off of github env var CI (#58) #58 (Ellis Tarn)
• ensure any waiters are reset between tests (#105) #105 (Todd Neal)

Chores

• remove accidental dependency on aws-go-sdk (#5) #5 (Ellis Tarn)
• migrate scheduling libraries (#6) #6 (Ellis Tarn)
• migrate scheduling libraries (#7) #7 (Ellis Tarn)
• promote spot/ondemand to become a neutral concept (#9) #9 (Ellis Tarn)
• deprecate vendor specific resource names (#12) #12 (Ellis Tarn)
• migrate utils (#13) #13 (Ellis Tarn)
• migrate controllers, operator, etc (#15) #15 (Ellis Tarn)
• migrate webhooks (#16) #16 (Ellis Tarn)
• Added NOTICE and linting (#18) #18 (Ellis Tarn)
• Create a singleton controller that registers as a runnable with the manager (#23) #23 (Jonathan Innis)
• remove non generic ptr libraries (#41) #41 (Ellis Tarn)
• renamed webhooks to be specific to api group (#42) #42 (Ellis Tarn)
• merge Consolidation and Emptiness into Deprovisioning (#43) #43 (Nick Tran)
• use client.IgnoreNotFound (#55) #55 (Ellis Tarn)
• add better error messages (#60) #60 (Todd Neal)
• merge Expiration into deprovisioning controller (#59) #59 (Nick Tran)
• Implement Structured Logging (#68) #68 (Amanuel Engeda)
• simplified deprovisioning interface (#71) #71 (Nick Tran)
• Logs Lower Case Starting Letter (#75) #75 (Amanuel Engeda)
• Convert InstanceType interface to struct (#83) #83 (Jonathan Innis)
• Typed Controller for Read/Write and some controller cleanup (#51) #51 (Jonathan Innis)
• removed unused settingstore (#94) #94 (Ellis Tarn)
• dynamically update cluster state, rather than waiting for sync (#100) #100 (Ellis Tarn)
• remove K8s 1.20 from CI (#107) #107 (Todd Neal)
• Create a Machine spec in memory (#102) #102 (Jonathan Innis)
• remove unused code (#110) #110 (Todd Neal)
• provide a better error message for Karpenter replicas (#108) [#108](https://github.com/aws/karpenter-core/pul...

v0.0.1

fix: presubmits (#4)