Releases: kubernetes-sigs/cluster-api
Releases · kubernetes-sigs/cluster-api
v1.3.10
👌 Kubernetes version support
- Management Cluster: v1.20.x -> v1.26.x [1]
- Workload Cluster: v1.18.x -> v1.26.x
[1] When using Cluster API with the experimental CLUSTER_TOPOLOGY feature, the Kubernetes version of the management cluster must be >= 1.22.0.
More information about version support can be found here
Changes since v1.3.9
📈 Overview
- 6 new commits merged
- 2 bugs fixed 🐛
🐛 Bug Fixes
- Dependency: Bump to docker v24.0.5-0.20230714235725-36e9e796c6fc (#9046)
- KCP: Requeue KCP object if ControlPlaneComponentsHealthyCondition is not yet true (#9034)
🌱 Others
- ClusterCacheTracker: ClusterCacheTracker: ensure Get/List calls are not getting stuck when apiserver is unreachable (#9033)
- Dependency: Bump google.golang.org/grpc to v1.55.0 (#8971)
- Dependency: Bump docker to v24.0.5 (#9067)
- Dependency: Change tilt debug base image to golang (#9075)
Thanks to all our contributors! 😊
v1.5.0-rc.1
🚨 This is a RELEASE CANDIDATE. Use it only for testing purposes. If you find any bugs, file an issue.
⚠️ Deprecation warning
The API version v1alpha3 is no longer served in this release. API version v1alpha4 is deprecated and will be removed in v1.6.
Changes since v1.5.0-rc.0
📈 Overview
- 1 new commits merged
🌱 Others
- util: Move
internal.labelstoformatpackage for use by providers (#9006)
Reference to v1.5.0-rc.0 release notes
Changes since release v1.4.0
👌 Kubernetes version support
- Management Cluster: v1.24.x -> v1.27.x
- Workload Cluster: v1.22.x -> v1.27.x
More information about version support can be found here
Changes since v1.4.0
📈 Overview
- 324 new commits merged
- 4 breaking changes
⚠️ - 19 feature additions ✨
- 62 bugs fixed 🐛
⚠️ Breaking Changes
- CAPD: Use k8s v1.27.0 in quickstart docs and CAPD (#8518)
- clusterctl: Bump to CR v0.15.0 (#8007)
- clusterctl: Follow XDG Directory standard for config/data/... files (#6913)
- clusterctl: Stop serving v1alpha3 API types (#8549)
✨ New Features
- CAPBK: Use caching read for bootstrap config owner (#8867)
- CAPIM: Enable Kubernetes upgrades in CAPIM (#8938)
- ClusterClass: Add webhook warning for missing ClusterClass (#8746)
- ClusterClass: add support or concurrent MD upgrades in classy clusters (#8432)
- clusterctl: Add MachinePool Machine support in clusterctl discovery (#8836)
- clusterctl: Add in memory provider (#8799)
- clusterctl: Add support for add-on providers in clusterctl (#8472)
- clusterctl: Introduce possibility to create clusterctl plugins (#8957)
- clusterctl: feat: accept resource mutators in Move operation (#7966)
- Devtools: hack/observability bump kube-state-metrics chart to get version v2.9.2 (#8979)
- Devtools: hack/observability: Add Parca, Bump observability tools, Add Grafana scale dashboard (#8754)
- Devtools: hack/observability: improve dashboards (#8964)
- e2e: ci: collect debug information about leftover processes (#8734)
- IPAM: Implements Getter interface for IPAddressClaim object (#8374)
- KCP/Core/CAPD: Add flags for configuring rate limits (#8579)
- Machine: Enable configuring the kubernetes rest client timeout for draining nodes (#8917)
- MachinePool: Add MachinePool Machine implementation in core CAPI components (#8828)
- MachineSet: MS preflight checks to improve cluster stability (#8595)
- util: Move mergeMap to util (#8377)
🐛 Bug Fixes
- API: Ensure ownerReference apiVersions are always up to date (#8256)
- API: all: only set finalizers if deletionTimestamp is not set (#8949)
- CAPBK: ignition: start kubeadm after network.target (#8772)
- CAPBK: set uninitialized taint only on worker nodes (#8358)
- CAPD: Add kind mapper (#8880)
- CAPD: CAPD: Implement watch filter (#8789)
- CAPD: CAPD: change the haproxy entrypoint to prevent getting stopped immediately after start (#8685)
- CAPD: Delegate CAPD port selection to the container runtime (#8642)
- CAPD: test/capd: fix kind mapper entry for v1.25.11 (#8914)
- CAPD: test/e2e fix fail-swap-on=false flag not being part of kind images anymore (#8767)
- CAPIM: fix cluster deletion in the in-memory API server (#8818)
- CAPIM: fix inmemory provider docker build (#8822)
- CAPIM: test/e2e/in-memory: set providerID after VM is provisioned (#8879)
- CI: Always build Kind images for upgrade tests (#8859)
- CI: Ensure ALWAYS_BUILD_KIND_IMAGES is defaulted correctly in ci script (#8873)
- CI: verifier: use pull_request instead of pull_request_target (#8635)
- ClusterCacheTracker: ClusterCacheTracker: Stop pod caching when checking workload cluster (#8850)
- ClusterClass: Drop uid from error messages (#8333)
- ClusterClass: Validate variables in defaulting webhook (#8332)
- ClusterClass: avoid errors when MHC and upgrade occur together in classy clusters (#8464)
- ClusterClass: topology controller should avoid unnecessary rollouts during upgrades (#8628)
- ClusterClass: update TopologyReconciled condition on cluster deletion (#8422)
- clusterctl: Add soft ownership from clusters to ClusterResourceSetBinding (#8318)
- clusterctl: Ensure all GitHub releases are fetched when searching provider versions (#8240)
- clusterctl: Return error on infra cluster and control plane discovery (#8604)
- clusterctl: Use local kustomize version in create-local-repository.py (#8414)
- clusterctl: clusterctl: return early if release for latest tag does not exist yet (#8253)
- clusterctl: handle nil pointer in clusterctl describe (#8724)
- ClusterResourceSet: Make sure the Kubernetes API Server service already created on remote cluster before applying ClusterResourceSets (#8048)
- ClusterResourceSet: ClusterResourceSet: continue applying when apply for a single cluster failed (#8611)
- Dependency: Update kindnetd and kindest/haproxy (#8469)
- Devtools: Update clean-charts make target for new paths (#8904)
- Devtools: tilt: fix build_engine setting when using podman (#8972)
- Devtools: tilt: set CR logger in tilt-prepare, fix allowed contexts (#8919)
- e2e: Adjust machinepool helper e2e timeout (#8739)
- e2e: Fix length check in MD helper (#8352)
- e2e: Ignore cert annotation in rollout test (#8761)
- e2e: Increase timeout for clusterclass rollout test (#8897)
- e2e: Increased assertClusterObjects timeout (#8758)
- e2e: Pin kindest/node images to known good versions in clusterctl upgrade tests (#8860)
- e2e: Remove non-functioning ginkgo.no-color flag (#8703)
- e2e: Revert "Add calico as e2e test CNI" (#8575)
- e2e: test/e2e check for machines being ready after provisioning on Runtime SDK test (#8625)
- e2e: test/e2e: use topology cluster-template for clusterctl upgrade mgmt cluster (#8311)
- e2e: test/framework fix docker pod log collector (#8634)
- IPAM: ipam: fix gateway being required for IPAddress (#8506)
- KCP: Allow machine rollout if cert reconcile fails (#8711)
- KCP: Ensure nil-pointer check in KCP syncMachines (#8481)
- KCP: Prevent KCP to create many private keys for each reconcile (#8617)
- KCP: enable updates from CloudInit (mostly ubuntu) based KCPs to Ignition (flatcar) based KCPs (#8339)
- Machine: bug fix to set node interruptible label (#8297)
- Machine: fix node label propagation (#8427)
- MachinePool: Add node watcher to MachinePool controller (#8443)
- MachinePool: Fix MachinePool node taint patching (#8462)
- MachinePool: MachinePool: always patch owned conditions, fix GetTypedPhase, doc fixes (#8627)
- MachinePool: Update MachinePool bootstrap dataSecretName when bootstrap config changes (#8667)
- Testing: Catch failure when fetching component URLs (#8455)
- Testing: Fix flake in machine phases test (#8370)
- Testing: Fix flaky wait in CRS test (#8334)
- Testing: Use CreateAndWait in MP ScaleFromZero test (#8927)
- Testing: internal/machine-controller fix flakes in phases test (#8576)
- Testing: machine-controller: fix phase tests race condition in tests on lastUpdated field (#8484)
🌱 Others
- API: Deprecate v1alpha3 and v1alpha4 in CRDs (#8687)
- CAPBK: replace reflect deepequal in bootstrap kubeadm (#8266)
- CAPD: Automatically set kubelet args for capd (#8881)
- CAPD: CAPD: only ignore necessary kubeadm preflight errors (#7911)
- CAPIM: Add startup timeout to the in memory provider (#8831)
- CAPIM: Add watch to in-memory server multiplexer (#8851)
- CAPIM: CAPIM: Enable update for coreDNS and kube-proxy (#8899)
- CAPIM: Fixups for watch in in-memory apiServer (#8898)
- CAPIM: Improve etcd management in CAPIM (#8906)
- CAPIM: add ClusterClass support for in-memory provider (#8807)
- CAPIM: fix readme for in-memory provider (#8905)
- CAPIM: in-memory provider: add apiserver & etcd metrics (#8962)
- CAPIM: test/e2e/in-memory: enable unit tests (#8886)
- CAPIM: test/e2e/in-memory: improve locking, return errors instead of panic (#8945)
- CAPIM: test/e2e/in-memory: improve performance by fixing locking issues (#8895)
- CAPIM: test/e2e/in-memory: sync in-memory provider ClusterClass with test (#8892)
- CAPIM: test/infra/inmemory: rename controller files (#8830)
- CI: Add area/dependency label for dependabot when bumping gomod (#8410)
- CI: Add dependency label to github actions update (#8845)
- CI: Switch cloudbuild.yaml to e2 machine type (#8428)
- CI: verifier: change back to pull_request_target to make it work again (#8637)
- ClusterCacheTracker: ClusterCacheTracker: improve error when workload cluster is not reachable (#8801)
- ClusterCacheTracker: Deprecate DefaultIndex usage and remove where not needed (#8855)
- ClusterCacheTracker: Use ClusterCacheTracker consistently (instead of NewClusterClient) (#8744)
- ClusterCacheTracker: Use rest config from ClusterCacheTracker consistently (#8894)
- ClusterClass: cluster/topology: use cached Cluster get in Reconcile (#8936)
- ClusterClass: cluster/topology: use cached MD list in get current state (#8922)
- ClusterClass: deprecate rolloutAfter in cluster topology (#8324)
- ClusterClass: upgrading control plane should only be blocked if MD are upgrading (not just rolling out) (#8658)
- clusterctl: Add move annotation on objects for cluster move operation (#8322)
- clusterctl: Bump controller-tools to v0.12 (#8581)
- clusterctl: CAPIM: use consistent spelling for release arti...
v1.5.0-rc.0
🚨 This is a RELEASE CANDIDATE. Use it only for testing purposes. If you find any bugs, file an issue.
⚠️ Deprecation warning
The API version v1alpha3 is no longer served in this release. API version v1alpha4 is deprecated and will be removed in v1.6.
Changes since v1.5.0-beta.1
📈 Overview
- 19 new commits merged
- 7 feature additions ✨
- 1 bugs fixed 🐛
✨ New Features
- CAPIM: Enable Kubernetes upgrades in CAPIM (#8938)
- ClusterClass: Add webhook warning for missing ClusterClass (#8746)
- clusterctl: Add MachinePool Machine support in clusterctl discovery (#8836)
- clusterctl: Introduce possibility to create clusterctl plugins (#8957)
- Devtools: hack/observability bump kube-state-metrics chart to get version v2.9.2 (#8979)
- Devtools: hack/observability: improve dashboards (#8964)
- MachinePool: Add MachinePool Machine implementation in core CAPI components (#8828)
🐛 Bug Fixes
- Devtools: tilt: fix build_engine setting when using podman (#8972)
🌱 Others
- CAPIM: in-memory provider: add apiserver & etcd metrics (#8962)
- Dependency: Bump golang.org/x/oauth2 from 0.9.0 to 0.10.0 (#8985)
- Devtools: Protect well know registry from accidental push during tilt up (#8930)
- Devtools: hack/observability: add apiserver request dashboards (#8978)
- e2e: Add SHA to default KIND image used in e2e tests (#8975)
- e2e: test/e2e: add field to scale test to deploy Clusters in separate namespace (#8963)
- e2e: test/e2e: improve logging for a detected rollout (#8953)
- KCP: KCP: Log the reason of a Machine rollout (#8959)
- Dependency: Update golangci-lint to v1.53.3 (#8796)
- Release: Improve release notes output (#8928)
📖 Additionally, there have been 1 contributions to our documentation and book. (#8958)
Reference to v1.5.0-beta.1 release notes
Changes since release v1.4.0
👌 Kubernetes version support
- Management Cluster: v1.24.x -> v1.27.x
- Workload Cluster: v1.22.x -> v1.27.x
More information about version support can be found here
Changes since v1.4.0
📈 Overview
- 323 new commits merged
- 4 breaking changes
⚠️ - 19 feature additions ✨
- 62 bugs fixed 🐛
⚠️ Breaking Changes
- CAPD: Use k8s v1.27.0 in quickstart docs and CAPD (#8518)
- clusterctl: Bump to CR v0.15.0 (#8007)
- clusterctl: Follow XDG Directory standard for config/data/... files (#6913)
- clusterctl: Stop serving v1alpha3 API types (#8549)
✨ New Features
- CAPBK: Use caching read for bootstrap config owner (#8867)
- CAPIM: Enable Kubernetes upgrades in CAPIM (#8938)
- ClusterClass: Add webhook warning for missing ClusterClass (#8746)
- ClusterClass: add support or concurrent MD upgrades in classy clusters (#8432)
- clusterctl: Add MachinePool Machine support in clusterctl discovery (#8836)
- clusterctl: Add in memory provider (#8799)
- clusterctl: Add support for add-on providers in clusterctl (#8472)
- clusterctl: Introduce possibility to create clusterctl plugins (#8957)
- clusterctl: feat: accept resource mutators in Move operation (#7966)
- Devtools: hack/observability bump kube-state-metrics chart to get version v2.9.2 (#8979)
- Devtools: hack/observability: Add Parca, Bump observability tools, Add Grafana scale dashboard (#8754)
- Devtools: hack/observability: improve dashboards (#8964)
- e2e: ci: collect debug information about leftover processes (#8734)
- IPAM: Implements Getter interface for IPAddressClaim object (#8374)
- KCP/Core/CAPD: Add flags for configuring rate limits (#8579)
- Machine: Enable configuring the kubernetes rest client timeout for draining nodes (#8917)
- MachinePool: Add MachinePool Machine implementation in core CAPI components (#8828)
- MachineSet: MS preflight checks to improve cluster stability (#8595)
- util: Move mergeMap to util (#8377)
🐛 Bug Fixes
- API: Ensure ownerReference apiVersions are always up to date (#8256)
- API: all: only set finalizers if deletionTimestamp is not set (#8949)
- CAPBK: ignition: start kubeadm after network.target (#8772)
- CAPBK: set uninitialized taint only on worker nodes (#8358)
- CAPD: Add kind mapper (#8880)
- CAPD: CAPD: Implement watch filter (#8789)
- CAPD: CAPD: change the haproxy entrypoint to prevent getting stopped immediately after start (#8685)
- CAPD: Delegate CAPD port selection to the container runtime (#8642)
- CAPD: test/capd: fix kind mapper entry for v1.25.11 (#8914)
- CAPD: test/e2e fix fail-swap-on=false flag not being part of kind images anymore (#8767)
- CAPIM: fix cluster deletion in the in-memory API server (#8818)
- CAPIM: fix inmemory provider docker build (#8822)
- CAPIM: test/e2e/in-memory: set providerID after VM is provisioned (#8879)
- CI: Always build Kind images for upgrade tests (#8859)
- CI: Ensure ALWAYS_BUILD_KIND_IMAGES is defaulted correctly in ci script (#8873)
- CI: verifier: use pull_request instead of pull_request_target (#8635)
- ClusterCacheTracker: ClusterCacheTracker: Stop pod caching when checking workload cluster (#8850)
- ClusterClass: Drop uid from error messages (#8333)
- ClusterClass: Validate variables in defaulting webhook (#8332)
- ClusterClass: avoid errors when MHC and upgrade occur together in classy clusters (#8464)
- ClusterClass: topology controller should avoid unnecessary rollouts during upgrades (#8628)
- ClusterClass: update TopologyReconciled condition on cluster deletion (#8422)
- clusterctl: Add soft ownership from clusters to ClusterResourceSetBinding (#8318)
- clusterctl: Ensure all GitHub releases are fetched when searching provider versions (#8240)
- clusterctl: Return error on infra cluster and control plane discovery (#8604)
- clusterctl: Use local kustomize version in create-local-repository.py (#8414)
- clusterctl: clusterctl: return early if release for latest tag does not exist yet (#8253)
- clusterctl: handle nil pointer in clusterctl describe (#8724)
- ClusterResourceSet: : Make sure the Kubernetes API Server service already created on remote cluster before applying ClusterResourceSets (#8048)
- ClusterResourceSet: ClusterResourceSet: continue applying when apply for a single cluster failed (#8611)
- Dependency: Update kindnetd and kindest/haproxy (#8469)
- Devtools: Update clean-charts make target for new paths (#8904)
- Devtools: tilt: fix build_engine setting when using podman (#8972)
- Devtools: tilt: set CR logger in tilt-prepare, fix allowed contexts (#8919)
- e2e: Adjust machinepool helper e2e timeout (#8739)
- e2e: Fix length check in MD helper (#8352)
- e2e: Ignore cert annotation in rollout test (#8761)
- e2e: Increase timeout for clusterclass rollout test (#8897)
- e2e: Increased assertClusterObjects timeout (#8758)
- e2e: Pin kindest/node images to known good versions in clusterctl upgrade tests (#8860)
- e2e: Remove non-functioning ginkgo.no-color flag (#8703)
- e2e: Revert "Add calico as e2e test CNI" (#8575)
- e2e: test/e2e check for machines being ready after provisioning on Runtime SDK test (#8625)
- e2e: test/e2e: use topology cluster-template for clusterctl upgrade mgmt cluster (#8311)
- e2e: test/framework fix docker pod log collector (#8634)
- IPAM: ipam: fix gateway being required for IPAddress (#8506)
- KCP: Allow machine rollout if cert reconcile fails (#8711)
- KCP: Ensure nil-pointer check in KCP syncMachines (#8481)
- KCP: Prevent KCP to create many private keys for each reconcile (#8617)
- KCP: enable updates from CloudInit (mostly ubuntu) based KCPs to Ignition (flatcar) based KCPs (#8339)
- Machine: bug fix to set node interruptible label (#8297)
- Machine: fix node label propagation (#8427)
- MachinePool: Add node watcher to MachinePool controller (#8443)
- MachinePool: Fix MachinePool node taint patching (#8462)
- MachinePool: MachinePool: always patch owned conditions, fix GetTypedPhase, doc fixes (#8627)
- MachinePool: Update MachinePool bootstrap dataSecretName when bootstrap config changes (#8667)
- Testing: Catch failure when fetching component URLs (#8455)
- Testing: Fix flake in machine phases test (#8370)
- Testing: Fix flaky wait in CRS test (#8334)
- Testing: Use CreateAndWait in MP ScaleFromZero test (#8927)
- Testing: internal/machine-controller fix flakes in phases test (#8576)
- Testing: machine-controller: fix phase tests race condition in tests on lastUpdated field (#8484)
🌱 Others
- API: Deprecate v1alpha3 and v1alpha4 in CRDs (#8687)
- CAPBK: replace reflect deepequal in bootstrap kubeadm (#8266)
- CAPD: Automatically set kubelet args for capd (#8881)
- CAPD: CAPD: only ignore necessary kubeadm preflight errors (#7911)
- CAPIM: Add startup timeout to the in memory provider (#8831)
- CAPIM: Add watch to in-memory server multiplexer (#8851)
- CAPIM: CAPIM: Enable update for coreDNS and kube-proxy (#8899)
- CAPIM: Fixups for watch in in-memory apiServer (#8898)
- CAPIM: Improve etcd management in CAPIM (#8906)
- CAPIM: add ClusterClass support for in-memory provider (#8807)
- CAPIM: fix readme for in-memory provider (#8905)
- CAPIM: in-memory provider: add apiserver & etcd metrics (#8962)
- CAPIM: test/e2e/in-memory: enable unit tests (#8886)
- CAPIM: test/e2e/in-memory: improve locking, return errors instead of panic (#8945)
- CAPIM: tes...
v1.5.0-beta.1
🚨 This is a BETA RELEASE. Use it only for testing purposes. If you find any bugs, file an issue.
Changes since v1.5.0-beta.0
📈 Overview
- 19 new commits merged
- 1 feature additions ✨
- 1 bugs fixed 🐛
✨ New Features
- Machine: Enable configuring the kubernetes rest client timeout for draining nodes (#8917)
🐛 Bug Fixes
- API: all: only set finalizers if deletionTimestamp is not set (#8949)
🌱 Others
- CAPIM: test/e2e/in-memory: improve locking, return errors instead of panic (#8945)
- ClusterClass: cluster/topology: use cached Cluster get in Reconcile (#8936)
- Devtools: Build golangci-lint locally (#8933)
- Devtools: all: Add flags to enable block profiling (#8934)
- Devtools: hack/observability: add metrics-server chart (#8950)
- Devtools: hack/observability: skip preload images on non-kind clusters (#8931)
- Devtools: hack/observability: use 20 Gi of storage for Prometheus (#8961)
- KCP: KCP: cache secrets between LookupOrGenerate and ensureCertificatesOwnerRef (#8926)
- ClusterCacheTracker/KCP/CAPBK: cache secrets in KCP, CABPK and ClusterCacheTracker (#8940)
- MachineSet: add MachineFinalizer during machine computation (#8463)
- CAPD: Drop cgroup-driver patch for CAPD (#8921)
- e2e: add improvements to scale e2e (#8910)
- e2e: test/e2e: increase apply timeout to 1m (#8941)
📖 Additionally, there have been 4 contributions to our documentation and book. (#8960, #8956, #8955, #8932)
Reference to v1.5.0-beta.0 release notes
More details about the release
👌 Kubernetes version support
- Management Cluster: v1.24.x -> v1.27.x
- Workload Cluster: v1.22.x -> v1.27.x
More information about version support can be found here
Changes since v1.4.0
📈 Overview
- 304 new commits merged
- 4 breaking changes
⚠️ - 12 feature additions ✨
- 61 bugs fixed 🐛
⚠️ Breaking Changes
- CAPD: Use k8s v1.27.0 in quickstart docs and CAPD (#8518)
- clusterctl: Bump to CR v0.15.0 (#8007)
- clusterctl: Follow XDG Directory standard for config/data/... files (#6913)
- clusterctl: Stop serving v1alpha3 API types (#8549)
✨ New Features
- CAPBK: Use caching read for bootstrap config owner (#8867)
- ClusterClass: add support or concurrent MD upgrades in classy clusters (#8432)
- Devtools: hack/observability: Add Parca, Bump observability tools, Add Grafana scale dashboard (#8754)
- IPAM: Implements Getter interface for IPAddressClaim object (#8374)
- KCP/Core/CAPD: Add flags for configuring rate limits (#8579)
- Machine: Enable configuring the kubernetes rest client timeout for draining nodes (#8917)
- MachineSet: MS preflight checks to improve cluster stability (#8595)
- clusterctl: Add in memory provider (#8799)
- clusterctl: Add support for add-on providers in clusterctl (#8472)
- clusterctl: feat: accept resource mutators in Move operation (#7966)
- e2e: ci: collect debug information about leftover processes (#8734)
- util: Move mergeMap to util (#8377)
🐛 Bug Fixes
- API: Ensure ownerReference apiVersions are always up to date (#8256)
- API: all: only set finalizers if deletionTimestamp is not set (#8949)
- CAPBK: ignition: start kubeadm after network.target (#8772)
- CAPBK: set uninitialized taint only on worker nodes (#8358)
- CAPD: Add kind mapper (#8880)
- CAPD: CAPD: Implement watch filter (#8789)
- CAPD: CAPD: change the haproxy entrypoint to prevent getting stopped immediately after start (#8685)
- CAPD: Delegate CAPD port selection to the container runtime (#8642)
- CAPD: test/capd: fix kind mapper entry for v1.25.11 (#8914)
- CAPD: test/e2e fix fail-swap-on=false flag not being part of kind images anymore (#8767)
- CAPIM: fix cluster deletion in the in-memory API server (#8818)
- CAPIM: fix inmemory provider docker build (#8822)
- CAPIM: test/e2e/in-memory: set providerID after VM is provisioned (#8879)
- CI: Always build Kind images for upgrade tests (#8859)
- CI: Ensure ALWAYS_BUILD_KIND_IMAGES is defaulted correctly in ci script (#8873)
- CI: verifier: use pull_request instead of pull_request_target (#8635)
- ClusterCacheTracker: ClusterCacheTracker: Stop pod caching when checking workload cluster (#8850)
- ClusterClass: Drop uid from error messages (#8333)
- ClusterClass: Validate variables in defaulting webhook (#8332)
- ClusterClass: avoid errors when MHC and upgrade occur together in classy clusters (#8464)
- ClusterClass: topology controller should avoid unnecessary rollouts during upgrades (#8628)
- ClusterClass: update TopologyReconciled condition on cluster deletion (#8422)
- ClusterResourceSet: : Make sure the Kubernetes API Server service already created on remote cluster before applying ClusterResourceSets (#8048)
- ClusterResourceSet: ClusterResourceSet: continue applying when apply for a single cluster failed (#8611)
- Dependency: Update kindnetd and kindest/haproxy (#8469)
- Devtools: Update clean-charts make target for new paths (#8904)
- Devtools: tilt: set CR logger in tilt-prepare, fix allowed contexts (#8919)
- IPAM: ipam: fix gateway being required for IPAddress (#8506)
- KCP: Allow machine rollout if cert reconcile fails (#8711)
- KCP: Ensure nil-pointer check in KCP syncMachines (#8481)
- KCP: Prevent KCP to create many private keys for each reconcile (#8617)
- KCP: enable updates from CloudInit (mostly ubuntu) based KCPs to Ignition (flatcar) based KCPs (#8339)
- Machine: bug fix to set node interruptible label (#8297)
- Machine: fix node label propagation (#8427)
- MachinePool: Add node watcher to MachinePool controller (#8443)
- MachinePool: Fix MachinePool node taint patching (#8462)
- MachinePool: MachinePool: always patch owned conditions, fix GetTypedPhase, doc fixes (#8627)
- MachinePool: Update MachinePool bootstrap dataSecretName when bootstrap config changes (#8667)
- Testing: Catch failure when fetching component URLs (#8455)
- Testing: Fix flake in machine phases test (#8370)
- Testing: Fix flaky wait in CRS test (#8334)
- Testing: Use CreateAndWait in MP ScaleFromZero test (#8927)
- Testing: internal/machine-controller fix flakes in phases test (#8576)
- Testing: machine-controller: fix phase tests race condition in tests on lastUpdated field (#8484)
- clusterctl: Add soft ownership from clusters to ClusterResourceSetBinding (#8318)
- clusterctl: Ensure all GitHub releases are fetched when searching provider versions (#8240)
- clusterctl: Return error on infra cluster and control plane discovery (#8604)
- clusterctl: Use local kustomize version in create-local-repository.py (#8414)
- clusterctl: clusterctl: return early if release for latest tag does not exist yet (#8253)
- clusterctl: handle nil pointer in clusterctl describe (#8724)
- e2e: Adjust machinepool helper e2e timeout (#8739)
- e2e: Fix length check in MD helper (#8352)
- e2e: Ignore cert annotation in rollout test (#8761)
- e2e: Increase timeout for clusterclass rollout test (#8897)
- e2e: Increased assertClusterObjects timeout (#8758)
- e2e: Pin kindest/node images to known good versions in clusterctl upgrade tests (#8860)
- e2e: Remove non-functioning ginkgo.no-color flag (#8703)
- e2e: Revert "Add calico as e2e test CNI" (#8575)
- e2e: test/e2e check for machines being ready after provisioning on Runtime SDK test (#8625)
- e2e: test/e2e: use topology cluster-template for clusterctl upgrade mgmt cluster (#8311)
- e2e: test/framework fix docker pod log collector (#8634)
🌱 Others
- API: Deprecate v1alpha3 and v1alpha4 in CRDs (#8687)
- CAPBK: replace reflect deepequal in bootstrap kubeadm (#8266)
- CAPD: Automatically set kubelet args for capd (#8881)
- CAPD: CAPD: only ignore necessary kubeadm preflight errors (#7911)
- CAPIM: Add startup timeout to the in memory provider (#8831)
- CAPIM: Add watch to in-memory server multiplexer (#8851)
- CAPIM: CAPIM: Enable update for coreDNS and kube-proxy (#8899)
- CAPIM: Fixups for watch in in-memory apiServer (#8898)
- CAPIM: Improve etcd management in CAPIM (#8906)
- CAPIM: add ClusterClass support for in-memory provider (#8807)
- CAPIM: fix readme for in-memory provider (#8905)
- CAPIM: test/e2e/in-memory: enable unit tests (#8886)
- CAPIM: test/e2e/in-memory: improve locking, return errors instead of panic (#8945)
- CAPIM: test/e2e/in-memory: improve performance by fixing locking issues (#8895)
- CAPIM: test/e2e/in-memory: sync in-memory provider ClusterClass with test (#8892)
- CAPIM: test/infra/inmemory: rename controller files (#8830)
- CI: Add area/dependency label for dependabot when bumping gomod (#8410)
- CI: Add dependency label to github actions update (#8845)
- CI: Switch cloudbuild.yaml to e2 machine type (#8428)
- CI: verifier: change back to pull_request_target to make it work again (#8637)
- ClusterCacheTracker: ClusterCacheTracker: improve error when workload cluster is not reachable (#8801)
- ClusterCacheTracker: Deprecate DefaultIndex usage and remove where not needed (#8855)
- ClusterCacheTracker: Use ClusterCacheTracker consistently (instead of NewClusterClient) (#8744)
- ClusterCacheTracker: Use rest config from ClusterCacheTracker consistently (#8894)
- ClusterClass: cluster/topology: use cached Cluster get in Reconcile (#8936)
- ClusterClass: cluster/topology: use cached MD list in get current state (#8922)
- ClusterClass: deprecate rolloutAfter in cluster topology (#8324)
- ClusterClass: upgrading control plane should only be blocked if MD are upgrading (not just rolling out) (#8658)
- Core: Cache unstructured in Cluster, MD and MS controller (#8916)
- Core: Remo...
v1.5.0-beta.0
🚨 This is a beta release. Use it only for testing purposes. If you find any bugs, file an issue.
Highlights
Some of the hightlights in this release are:
- MS preflight checks to improve cluster stability (behind feature gate).
- Support for concurrent MachineDeployment upgrades in classy clusters.
- Support for add-on providers in clusterctl.
- Performance improvements when using CAPI at scale, new dev-tools and a new in-memory infrastructure provider for scale testing.
⚠️ Deprecation warning
The API version v1alpha3 is no longer served in this release. API version v1alpha4 is deprecated and will be removed in v1.6.
Cluster API v1.4 compared to v1.5
See the Cluster-API book for an overview over relevant changes between Cluster API v1.4 and v1.5 for maintainers of providers and consumers of our Go API.
More details about the release
👌 Kubernetes version support
- Management Cluster: v1.24.x -> v1.27.x
- Workload Cluster: v1.22.x -> v1.27.x
Changes since v1.4.0
📈 Overview
- 285 new commits merged
- 4 breaking changes
⚠️ - 11 feature additions ✨
- 60 bugs fixed 🐛
⚠️ Breaking Changes
- CAPD: Use k8s v1.27.0 in quickstart docs and CAPD (#8518)
- clusterctl: Bump to CR v0.15.0 (#8007)
- clusterctl: Follow XDG Directory standard for config/data/... files (#6913)
- clusterctl: Stop serving v1alpha3 API types (#8549)
✨ New Features
- CAPBK: Use caching read for bootstrap config owner (#8867)
- ClusterClass: add support or concurrent MD upgrades in classy clusters (#8432)
- Devtools: hack/observability: Add Parca, Bump observability tools, Add Grafana scale dashboard (#8754)
- IPAM: Implements Getter interface for IPAddressClaim object (#8374)
- KCP/Core/CAPD: Add flags for configuring rate limits (#8579)
- MachineSet: MS preflight checks to improve cluster stability (#8595)
- clusterctl: Add in memory provider (#8799)
- clusterctl: Add support for add-on providers in clusterctl (#8472)
- clusterctl: feat: accept resource mutators in Move operation (#7966)
- e2e: ci: collect debug information about leftover processes (#8734)
- util: Move mergeMap to util (#8377)
🐛 Bug Fixes
- API: Ensure ownerReference apiVersions are always up to date (#8256)
- CAPBK: ignition: start kubeadm after network.target (#8772)
- CAPBK: set uninitialized taint only on worker nodes (#8358)
- CAPD: Add kind mapper (#8880)
- CAPD: CAPD: Implement watch filter (#8789)
- CAPD: CAPD: change the haproxy entrypoint to prevent getting stopped immediately after start (#8685)
- CAPD: Delegate CAPD port selection to the container runtime (#8642)
- CAPD: test/capd: fix kind mapper entry for v1.25.11 (#8914)
- CAPD: test/e2e fix fail-swap-on=false flag not being part of kind images anymore (#8767)
- CAPIM: fix cluster deletion in the in-memory API server (#8818)
- CAPIM: fix inmemory provider docker build (#8822)
- CAPIM: test/e2e/in-memory: set providerID after VM is provisioned (#8879)
- CI: Always build Kind images for upgrade tests (#8859)
- CI: Ensure ALWAYS_BUILD_KIND_IMAGES is defaulted correctly in ci script (#8873)
- CI: verifier: use pull_request instead of pull_request_target (#8635)
- ClusterCacheTracker: ClusterCacheTracker: Stop pod caching when checking workload cluster (#8850)
- ClusterClass: Drop uid from error messages (#8333)
- ClusterClass: Validate variables in defaulting webhook (#8332)
- ClusterClass: avoid errors when MHC and upgrade occur together in classy clusters (#8464)
- ClusterClass: topology controller should avoid unnecessary rollouts during upgrades (#8628)
- ClusterClass: update TopologyReconciled condition on cluster deletion (#8422)
- ClusterResourceSet: : Make sure the Kubernetes API Server service already created on remote cluster before applying ClusterResourceSets (#8048)
- ClusterResourceSet: ClusterResourceSet: continue applying when apply for a single cluster failed (#8611)
- Dependency: Update kindnetd and kindest/haproxy (#8469)
- Devtools: Update clean-charts make target for new paths (#8904)
- Devtools: tilt: set CR logger in tilt-prepare, fix allowed contexts (#8919)
- IPAM: ipam: fix gateway being required for IPAddress (#8506)
- KCP: Allow machine rollout if cert reconcile fails (#8711)
- KCP: Ensure nil-pointer check in KCP syncMachines (#8481)
- KCP: Prevent KCP to create many private keys for each reconcile (#8617)
- KCP: enable updates from CloudInit (mostly ubuntu) based KCPs to Ignition (flatcar) based KCPs (#8339)
- Machine: bug fix to set node interruptible label (#8297)
- Machine: fix node label propagation (#8427)
- MachinePool: Add node watcher to MachinePool controller (#8443)
- MachinePool: Fix MachinePool node taint patching (#8462)
- MachinePool: MachinePool: always patch owned conditions, fix GetTypedPhase, doc fixes (#8627)
- MachinePool: Update MachinePool bootstrap dataSecretName when bootstrap config changes (#8667)
- Testing: Catch failure when fetching component URLs (#8455)
- Testing: Fix flake in machine phases test (#8370)
- Testing: Fix flaky wait in CRS test (#8334)
- Testing: Use CreateAndWait in MP ScaleFromZero test (#8927)
- Testing: internal/machine-controller fix flakes in phases test (#8576)
- Testing: machine-controller: fix phase tests race condition in tests on lastUpdated field (#8484)
- clusterctl: Add soft ownership from clusters to ClusterResourceSetBinding (#8318)
- clusterctl: Ensure all GitHub releases are fetched when searching provider versions (#8240)
- clusterctl: Return error on infra cluster and control plane discovery (#8604)
- clusterctl: Use local kustomize version in create-local-repository.py (#8414)
- clusterctl: clusterctl: return early if release for latest tag does not exist yet (#8253)
- clusterctl: handle nil pointer in clusterctl describe (#8724)
- e2e: Adjust machinepool helper e2e timeout (#8739)
- e2e: Fix length check in MD helper (#8352)
- e2e: Ignore cert annotation in rollout test (#8761)
- e2e: Increase timeout for clusterclass rollout test (#8897)
- e2e: Increased assertClusterObjects timeout (#8758)
- e2e: Pin kindest/node images to known good versions in clusterctl upgrade tests (#8860)
- e2e: Remove non-functioning ginkgo.no-color flag (#8703)
- e2e: Revert "Add calico as e2e test CNI" (#8575)
- e2e: test/e2e check for machines being ready after provisioning on Runtime SDK test (#8625)
- e2e: test/e2e: use topology cluster-template for clusterctl upgrade mgmt cluster (#8311)
- e2e: test/framework fix docker pod log collector (#8634)
🌱 Others
- API: Deprecate v1alpha3 and v1alpha4 in CRDs (#8687)
- CAPBK: replace reflect deepequal in bootstrap kubeadm (#8266)
- CAPD: Automatically set kubelet args for capd (#8881)
- CAPD: CAPD: only ignore necessary kubeadm preflight errors (#7911)
- CAPIM: Add startup timeout to the in memory provider (#8831)
- CAPIM: Add watch to in-memory server multiplexer (#8851)
- CAPIM: CAPIM: Enable update for coreDNS and kube-proxy (#8899)
- CAPIM: Fixups for watch in in-memory apiServer (#8898)
- CAPIM: Improve etcd management in CAPIM (#8906)
- CAPIM: add ClusterClass support for in-memory provider (#8807)
- CAPIM: fix readme for in-memory provider (#8905)
- CAPIM: test/e2e/in-memory: enable unit tests (#8886)
- CAPIM: test/e2e/in-memory: improve performance by fixing locking issues (#8895)
- CAPIM: test/e2e/in-memory: sync in-memory provider ClusterClass with test (#8892)
- CAPIM: test/infra/inmemory: rename controller files (#8830)
- CI: Add area/dependency label for dependabot when bumping gomod (#8410)
- CI: Add dependency label to github actions update (#8845)
- CI: Switch cloudbuild.yaml to e2 machine type (#8428)
- CI: verifier: change back to pull_request_target to make it work again (#8637)
- ClusterCacheTracker: ClusterCacheTracker: improve error when workload cluster is not reachable (#8801)
- ClusterCacheTracker: Deprecate DefaultIndex usage and remove where not needed (#8855)
- ClusterCacheTracker: Use ClusterCacheTracker consistently (instead of NewClusterClient) (#8744)
- ClusterCacheTracker: Use rest config from ClusterCacheTracker consistently (#8894)
- ClusterClass: cluster/topology: use cached MD list in get current state (#8922)
- ClusterClass: deprecate rolloutAfter in cluster topology (#8324)
- ClusterClass: upgrading control plane should only be blocked if MD are upgrading (not just rolling out) (#8658)
- Core: Cache unstructured in Cluster, MD and MS controller (#8916)
- Core: Remove unnecessary requeues (#8743)
- Dependency: Bump EndBug/add-and-commit from 9.1.1 to 9.1.2 (#8584)
- Dependency: Bump EndBug/add-and-commit from 9.1.2 to 9.1.3 (#8621)
- Dependency: Bump actions/checkout from 3.3.0 to 3.4.0 (#8321)
- Dependency: Bump actions/checkout from 3.4.0 to 3.5.0 (#8389)
- Dependency: Bump actions/checkout from 3.5.0 to 3.5.2 (#8540)
- Dependency: Bump actions/checkout from 3.5.2 to 3.5.3 (#8837)
- Dependency: Bump actions/setup-go from 3.5.0 to 4.0.1 (#8664)
- Dependency: Bump docker/distribution to v2.8.2 (#8645)
- Dependency: Bump gcb-docker-gcloud from v20230424-910a2a439d to v20230522-312425ae46 (#8770)
- Dependency: Bump gcb-docker-gcloud image (#8570)
- Dependency: Bump github.com/go-logr/logr from 1.2.3 to 1.2.4 (#8461)
- Dependency: Bump github.com/onsi/ginkgo/v2 from 2.10.0 to 2.11.0 (#8891)
- Dependency: Bump github.com/onsi...
v1.4.4
👌 Kubernetes version support
- Management Cluster: v1.23.x -> v1.27.x
- Workload Cluster: v1.21.x -> v1.27.x
More information about version support can be found here
Changes since v1.4.3
📈 Overview
- 12 new commits merged
- 6 bugs fixed 🐛
🐛 Bug Fixes
- CAPD: Add kind mapper (#8903)
- CAPD: test/capd: fix kind mapper entry for v1.25.11 (#8915)
- CI: Always build Kind images for upgrade tests (#8871)
- CI: Ensure ALWAYS_BUILD_KIND_IMAGES is defaulted correctly in ci script (#8876)
- e2e: Use known kindest/node image versions by sha for e2e upgrade tests (#8868)
- CAPBK: ignition: start kubeadm after network.target (#8803)
🌱 Others
- Dependency: Update KIND to v0.20.0 (#8911)
- Dependency: Update cert-manager to v1.12.2 (#8888)
- KCP: fix noisy error log triggered by missing patch helper (#8866)
- CI: Add dependency label to github actions update (#8847)
📖 Additionally, there have been 2 contributions to our documentation and book. (#8813, #8806)
Thanks to all our contributors! 😊
v1.3.9
👌 Kubernetes version support
- Management Cluster: v1.20.x -> v1.26.x [1]
- Workload Cluster: v1.18.x -> v1.26.x
[1] When using Cluster API with the experimental CLUSTER_TOPOLOGY feature, the Kubernetes version of the management cluster must be >= 1.22.0.
More information about version support can be found here
Changes since v1.3.8
📈 Overview
- 5 new commits merged
- 3 bugs fixed 🐛
🐛 Bug Fixes
- CI: Always build Kind images for upgrade tests (#8870)
- CI: Ensure ALWAYS_BUILD_KIND_IMAGES is defaulted correctly in ci script (#8874)
- CAPBK: ignition: start kubeadm after network.target (#8804)
🌱 Others
- Dependency: Update cert-manager to v1.12.2 (#8887)
- CI: Add dependency label to github actions update (#8846)
Thanks to all our contributors! 😊
v1.4.3
👌 Kubernetes version support
- Management Cluster: v1.23.x -> v1.27.x
- Workload Cluster: v1.21.x -> v1.27.x
More information about version support can be found here
Changes since v1.4.2
📈 Overview
- 32 new commits merged
- 9 bugs fixed 🐛
🐛 Bug Fixes
- clusterctl: Return error on infra cluster and control plane discovery (#8609)
- e2e: Adjust machinepool helper e2e timeout (#8756)
- e2e: test/e2e check for machines being ready after provisioning on Runtime SDK test (#8646)
- e2e: test/framework fix docker pod log collector (#8643)
- KCP: Allow machine rollout if cert reconcile fails (#8737)
- KCP: Prevent KCP to create many private keys for each reconcile (#8619)
- CAPD: change the haproxy entrypoint to prevent getting stopped immediately after start (#8742)
- CAPD: Delegate CAPD port selection to the container runtime (#8655)
- CAPD: fix fail-swap-on=false flag not being part of kind images anymore (#8777)
🌱 Others
- API: Deprecate v1alpha3 and v1alpha4 in CRDs (#8700)
- Dependency: Bump docker/distribution to v2.8.2 (#8648)
- Dependency: Update cert-manager to v1.11.2 (#8639)
- Dependency: Update cert-manager to v1.12.0 (#8705)
- Dependency: Update cert-manager to v1.12.1 (#8752)
- Dependency: Update kpromo to v3.6.0 (#8682)
- Dependency: Update kubebuilder envtest (1.26.0 -> 1.27.1) (#8601)
- Devtools: Pin delve to version supporting go 1.19 (#8727)
- e2e: Disable fail-fast by default for e2e tests (#8659)
- e2e: Pin cgroup driver used in v0.3, v0.4 and v1.0 templates (#8693)
- e2e: Improve gomega fail handling in clusterClass rollout (#8771)
- e2e: e2e: log leftover processes to eventually detect zombies (#8671)
- Machine: use providerID string as-is (#8594)
📖 Additionally, there have been 10 contributions to our documentation and book. (#8730, #8720, #8674, #8633, #8608, #8607, #8539, #8713, #8677, #8652)
Thanks to all our contributors! 😊
v1.3.8
👌 Kubernetes version support
- Management Cluster: v1.20.x -> v1.26.x [1]
- Workload Cluster: v1.18.x -> v1.26.x
[1] When using Cluster API with the experimental CLUSTER_TOPOLOGY feature, the Kubernetes version of the management cluster must be >= 1.22.0.
More information about version support can be found here
Changes since v1.3.7
📈 Overview
- 20 new commits merged
- 8 bugs fixed 🐛
🐛 Bug Fixes
- clusterctl: Return error on infra cluster and control plane discovery (#8610)
- e2e: Adjust machinepool helper e2e timeout (#8755)
- e2e: test/e2e check for machines being ready after provisioning on Runtime SDK test (#8647)
- e2e: test/framework fix docker pod log collector (#8644)
- KCP: Allow machine rollout if cert reconcile fails (#8738)
- KCP: Prevent KCP to create many private keys for each reconcile (#8626)
- CAPD: change the haproxy entrypoint to prevent getting stopped immediately after start (#8741)
- CAPD: Delegate CAPD port selection to the container runtime (#8654)
🌱 Others
- Dependency: Bump docker/distribution to v2.8.2 (#8650)
- Dependency: Update cert-manager to v1.11.2 (#8640)
- Dependency: Update cert-manager to v1.12.0 (#8704)
- Dependency: Update cert-manager to v1.12.1 (#8751)
- Dependency: Update kpromo to v3.6.0 (#8683)
- Devtools: Pin delve to version supporting go 1.19 (#8726)
- e2e: Disable fail-fast by default for e2e tests (#8660)
- e2e: Pin cgroup driver used in v0.3 and v0.4 templates (#8692)
- e2e: Revert "Pin cgroup driver used in v0.3, v0.4 and templates" (#8697)
- e2e: log leftover processes to eventually detect zombies (#8672)
📖 Additionally, there have been 2 contributions to our documentation and book. (#8653, #8714)
Thanks to all our contributors! 😊
v1.4.2
👌 Kubernetes version support
- Management Cluster: v1.23.x -> v1.27.x
- Workload Cluster: v1.21.x -> v1.27.x
More information about version support can be found here
Changes since v1.4.1
📈 Overview
- 25 new commits merged
- 0 breaking changes
⚠️ - 2 feature additions ✨
- 9 bugs fixed 🐛
✨ New Features
- CAPD: Use k8s v1.27.0 in quickstart docs and CAPD (#8538)
- ClusterClass: add support or concurrent MD upgrades in classy clusters (#8528)
🐛 Bug Fixes
- ClusterClass: avoid errors when MHC and upgrade occur together in classy clusters (#8568)
- Dependency: Update kindnetd and kindest/haproxy (#8470)
- e2e: Catch failure when fetching component URLs (#8489)
- IPAM: fix gateway being required for IPAddress (#8574)
- KCP: Ensure nil-pointer check in KCP syncMachines (#8488)
- MachinePool: Add node watcher to MachinePool controller (#8474)
- Testing: internal/machine-controller fix flakes in phases test (#8590)
- Testing: machine-controller: fix phase tests race condition in tests on lastUpdated field (#8491)
🌱 Others
- Dependency: Update controller-tools to v1.11.4 (#8555)
- Dependency: Update cert manager to v1.11.1 (#8532)
- Dependency: Update kind to v1.18.0 (#8434)
- Dependency: Update docker to v20.10.24 (#8476)
- e2e: assert E2E error responses when waiting for MD nodes (#8516)
- e2e: use topology flavor for workload clusters in clusterctl upgrade test (#8550)
- IPAM: Make IPAddressClaim.Status.AddressRef optional (#8530)
📖 Additionally, there have been 8 contributions to our documentation and book. (#8582, #8562, #8537, #8512, #8513, #8482, #8556, #8589)
Thanks to all our contributors! 😊