clusterctl 1.12.0 డౌన్‌లోడ్

👌 Kubernetes version support
- Management Cluster: v1.31.x -> v1.34.x
- Workload Cluster: v1.29.x -> v1.34.x
More information about version support can be found here
Highlights
- Bumped to Go 1.24, controller-runtime v0.22, k8s.io/* v0.34, controller-gen v0.19
- In-place updates
- Cluster API introduces support for update extensions allowing users to execute changes on existing machines without deleting the machines and creating a new one.
- Both KCP and Machine deployments support in-place updates based on the new update extension
- Can be enabled with the new InPlaceUpdates feature gate
- More details can be found in the proposal
- Chained upgrades
- Clusters using managed topologies can now upgrade by more than one minor Kubernetes version by performing chained and efficient upgrades.
- Upgrade plan will be computed using Kubernetes version from the ClusterClass or by calling a new runtimeExtension (also defined in the ClusterClass)
- Existing lifecycle hooks have been adapted to the new workflow, new lifecycle hooks have been introduced to allow granular control of the upgrade steps.
- More details can be found in the proposal
- MachineHealthCheck: Add support for checking Machine conditions (#12827)
- Machine: First part of the Taint propagation proposal (Machine/MachineSet/MachineDeployment) (#12936, #12966)
- Feature tracking issue
- Can be enabled with the new MachineTaintPropagation feature gate
- KCP/CABPK: Add EncryptionAlgorithm field to Kubeadmconfig to support keys generated with RSA-3072, RSA-4096, ECDSA-P256, ECDSA-P384 (#12859)
- Introduce ReconcilerRateLimiting feature (#13006)
- Can be enabled with the new ReconcilerRateLimiting feature gate
- Rate-limits all reconcilers to at most 1 request per second
Note: Cluster API is only supported on conformant Kubernetes Clusters and contract-relevant provider resources (e.g. InfraCluster) have to be implemented as CRDs (i.e. not via an aggregated apiserver).
Other changes
- clusterctl: Add conditions filter for clusterctl describe (#12991)
- clusterctl: clusterctl move blocks when Cluster or ClusterClass is paused (#12786)
- KCP: Enable websocket dialer with fallback to spdy (for communication with etcd) (#12902)
- Runtime SDK: Add defensive response status checking in runtime client (#12898)
- Improved logging across several controllers
- Improved e2e test coverage, e.g.: CAPD: Add scale from/to 0 support for CAPD (#12572)
- New providers in clusterctl: HCP (#12800) control plane provider, Metal3 IPAM provider (#12756), metal-stack infrastructure provider (#12925)
Deprecation and Removals Warning
- ClusterResourceSet: Remove deprecated ClusterResourceSet feature gate (#12950)
- Reminder: v1alpha3 & v1alpha4 will be removed in CAPI v1.13 (they are already not served since a while)
- Reminder: v1beta1 is on track to be unserved in CAPI v1.14
- Reminder: Provider should start implementing the v1beta2 contract ASAP.
For additional details for providers, please take a look at Cluster API v1.11 compared to v1.12.
Changes since v1.11.0
📈 Overview
- 298 new commits merged
- 5 breaking changes ⚠️
- 42 feature additions ✨
- 42 bugs fixed 🐛
⚠️ Breaking Changes
- CI: Improve KAL config docs for forbidding OpenAPI defaulting (#12869)
- clusterctl: Block move when Cluster or ClusterClass is paused (#12786)
- Dependency: Bump to controller-runtime v0.22 & controller-tools v0.19 (#12634)
- Runtime SDK: Improve chained upgrade observability (#12973)
- Runtime SDK: Make the AfterClusterUpgrade hook blocking (#12984)
✨ New Features
- API/Machine/MachineSet/MachineDeployment: Machine related API changes, conversion and feature gate (#12936)
- CABPK: Add EncryptionAlgorithm to Kubeadmconfig (#12859)
- CAPD: Add scale from/to 0 support for CAPD (#12572)
- CI: Bump autoscaler to a9cb59fdd (#12643)
- CI: Bump Kubernetes in tests to v1.34.0 and claim support for v1.34 (#12699)
- ClusterCache/KCP: Deprecate GetClientCertificatePrivateKey and stop using it in KCP (#12846)
- ClusterCache: Add GetUncachedClient() (#12803)
- ClusterClass: Add .spec.upgrade.external.generateUpgradePlanExtension field to ClusterClass (#12809)
- ClusterClass: Add types and hook for GenerateUpgradePlan (#12823)
- ClusterClass: Additional validation in Cluster/ClusterClass webhook for chained upgrades (#12816)
- ClusterClass: Call GenerateUpgradePlanRequest Runtime Extension (#12903)
- ClusterClass: Implement core logic for chained upgrades (#12726)
- clusterctl: Add conditions filter for clusterctl describe (#12991)
- Control-plane: Add new control-plane provider HCP (#12800)
- e2e: Bump Kubernetes version used for testing to v1.34.0-rc.1 (#12623)
- e2e: Bump Kubernetes version used for testing to v1.34.0-rc.2 (#12658)
- e2e: Bump Kubernetes version used for testing to v1.35.0-beta.0 (#13029)
- e2e: Change RuntimeSDK e2e test ClusterClass to use GenerateUpgradePlan extension (#12955)
- e2e: Implement e2e test for in-place updates (#12938)
- KCP/CABPK/CI: Bump KAL to pick up latest requiredfields linter, add Min/MaxLength to BootstrapToken (#12563)
- KCP/MachineSet: Refactor BootstrapConfig/InfraMachine managedFields for in-place (#12890)
- KCP: Bump coredns/corefile-migration to v1.0.28 (#12748)
- KCP: Bump coredns/corefile-migration to v1.0.29 (#12862)
- KCP: Bump corefile-migration to v1.0.27 (#12636)
- KCP: Compare ClusterConfiguration via KubeadmConfig instead of annotation on Machine (#12758)
- KCP: Extend rollout logic for in-place updates (#12840)
- KCP: Implement CanUpdateMachine (#12857)
- KCP: Implement trigger in-place update (#12897)
- Machine: Add in-place updates support for machine controller (#12831)
- MachineDeployment: Add in-place to rollout planner (#12865)
- MachineDeployment: Implement CanUpdateMachineSet (#12965)
- MachineHealthCheck: Add support for checking Machine conditions in MachineHealthCheck (#12827)
- Misc: Add inplace updates featuregate (#12755)
- Misc: Improve logs, errors and conditions (#12992)
- Misc: Introduce & use wait for cache utils (#12957)
- Misc: Introduce reconciler rate-limiting and hook caching (#13006)
- Runtime SDK/IPAM/MachinePool: Cleanup exp packages (#12651)
- Runtime SDK: Add in-place update hooks to API (#12343)
- Runtime SDK: Add lifecycle hooks for chained-upgrade (#12878)
- Runtime SDK: Call new lifecycle hooks for chained-upgrades (#12891)
- Runtime SDK: Ensure ExtensionConfig controller can be used outside of the core provider (#12754)
- Runtime SDK: Implement GenerateUpgradePlan handler (#12927)
🐛 Bug Fixes
- API: Only try to convert infraRefs if they are set (#12686)
- API: Register conversion funcs in schemes (#12687)
- CABPK: Always use latest apiVersion when getting owner of KubeadmConfig in CABPK (#12685)
- CAPD: CAPD on rootless podman (#12941)
- CAPD: Fix the format error of healthCheck in test templates (#12787)
- CAPD: Remove finalizers during deletion if ownerRef was never set (#12675)
- CAPD: Run CAPD conversion tests in CI (#12583)
- CAPIM: Eliminate data race in DialContext (#12778)
- ClusterClass/MachinePool: Fix MP error in desired state calculation during Cluster creation (#12607)
- ClusterClass: Do not overwrite global http.DefaultClient TLSConfig (#13063)
- ClusterClass: Ensure holder field path in GeneratePatchRequest is set based on contract (#12684)
- ClusterClass: Fix field paths in ClusterClass compatibility validation errors (#12660)
- ClusterClass: Fix wait for cache in reconcile_state.go (#13032)
- ClusterClass: Stop adding conversion-data annotation to Cluster object (#12719)
- ClusterClass: Stop writing zero values for spec.controlPlaneEndpoint to ControlPlane objects (#12958)
- clusterctl: Removing Ready/Available prefix from STATUS Column (#12729)
- clusterctl: Verify providers need upgrade before applying (#12753)
- Devtools: Fix kube-state-metrics deployment (#13024)
- e2e: Do not require kubetest configration if not needed (#12948)
- e2e: Fix autoscaler e2e test flake (#12613)
- e2e: Fix e2e test issues introduced by chained upgrades (#12766)
- e2e: Fix objects with Changed Resource Versions flake (#12848)
- e2e: Fix upgrade runtimesdk test (#12833)
- e2e: Fix WaitForMachinesReady interval (#13051)
- e2e: Increase reconcile timeout for DockerMachine (#13099)
- e2e: Propagate clusterctl variables for cluster upgrades (#12949)
- KCP: Fix ControlPlaneComponentHealthCheckSeconds validation in KubeadmConfigSpec.Validate (#12609)
- KCP: Fix conversion issue in KubeadmControlPlaneTemplate with rolloutStrategy.type (#12608)
- KCP: Fix KCP KubeadmConfig isJoin detection (#13035)
- KCP: Grant delete permissions to Secrets. (#13097)
- Machine/MachinePool: Fix MachinePool nodeRef UID mismatch after K8s upgrade (#12392)
- Machine/MachineSet: Use MachineSet template values in completeMoveMachine for in-place updates (#13085)
- MachineDeployment: Fix race conditions ScaleDownOldMS (#12812)
- MachineDeployment: Fix race conditions ScaleDownOldMS OnDelete (#12830)
- MachineDeployment: Fix rollout with unavailable machines (#13020)
- Runtime SDK: Export ExtensionConfig webhook (#12598)
- Testing/e2e: Fix unit test flakes, improve clusterctl download error in e2e tests (#13056)
- Testing: Fix flaky TestFuzzyConversion (Cluster) test (#12618)
- Testing: Fix flaky TestReconcileMachinePhases unit test (#12616)
- Testing: Fix flaky TestReconcileState unit test (#12617)
- Testing: Fix KubeadmConfig fuzz test flake (#12679)
- util: Fix a panic in conditions.Delete method if the sources condition list is empty (#13053)
🌱 Others
- API: Relax validation for Machine .status.addresses to maximum of 128 instead of 32 items (#13090)
- Autoscaling: Bump autoscaler in e2e tests to v1.33.1 (#12790)
- CABPK: Migrate from Requeue to RequeueAfter in CABPK (#12988)
- CAPD: Cleanup CAPD exp packages (#12672)
- CAPD: Recreate container if we re-enter reconciliation and it exists but is not running (#12923)
- CI: Add OpenAPI defaulting detection for KubeadmConfig by using forbiddenmarkers (#12851)