Fixes since v0.1.7

• #1173: Get updated cluster before removing finalizers

The controller manager image for this release is: us.gcr.io/k8s-artifacts-prod/cluster-api/cluster-api-controller:v0.1.8

Fixes since v0.1.6

• #1154: Don't swallow errors when reconciling noderef

The controller manager image for this release is: us.gcr.io/k8s-artifacts-prod/cluster-api/cluster-api-controller:v0.1.7

Fixes since v0.1.4

• Fixed issue handling provider IDs of the form :https:// (specifically, 2 slashes)
• Fixed issue where deleting a Machine could hang if it was the last control plane member
• Fixed issue where it could take ~10 minutes per machine when rolling out an update to a MachineDeployment
• Backported support for go modules; both dep and go modules are supported for v0.1.x. Master only supports go modules.

Fixes since v0.1.3

• Added support for remote node references (where the cluster running Cluster API is different from the workload cluster)
• ACTION REQUIRED: providers must implement a change to add support for this. See https://cluster-api.sigs.k8s.io/common_code/noderef_controller.html for details.
• Made ClusterNetworkConfig optional
• Added events to MachineDeployment operations
• Added events to MachineSet operations
• Fixed some RBAC permissions issues by adding missing permissions

Fixes since 0.1.2

• Fixed an issue where the distroless base image could have been out of date when building a release

Fixes since 0.1.1

• Switched to go 1.12.5
• Switched to a non-root user for cluster-api-manager container
• Enabled foreground deletion by default for MachineSets and MachineDeployments
• Stopped trying to reconcile deleted MachineSets and MachineDeployments
• Stopped trying to pivot deleted Machines
• Added short names for custom resources (e.g. “ma” for “machines”)
• Added foreground deletion by default to Clusters

Fixes since v0.1.0

• Added support to clusterctl to deploy multiple control plane machines
• Made the apply phase of clusterctl more resilient to the new cluster coming online but not ready yet
• Added /status subresource support for MachineDeployments and MachineSets
• Added ability for manager to watch a single namespace instead of all namespaces
• Switched to go 1.12

0.1.0 (v1alpha1)

This is the official v1alpha1 release for Cluster API, no further breaking changes will be made to the v1alpha1 types.

Providers that rely on the v1alpha1 types and are vendoring this repository should update their dependencies to use either the 0.1.0 tag or the release-0.1 branch to avoid pulling in any breaking changes.

The manager container for this release is available at gcr.io/k8s-cluster-api/cluster-api-controller:0.1.0

The manager container for this release is available at gcr.io/k8s-cluster-api/cluster-api-controller:0.0.0-alpha.4

Note: This "release" is intended to be a snapshot of the repository just prior to the migration to CRDs.

clusterctl is not included in this release as the binary from this repository longer supports the --provider flag. Follow the Getting Started instructions to find the appropriate binary for your provider.

Changes since 0.0.0-alpha.2

• Breaking API Change: MachineSpec no longer includes MachineRoles (#405)
• Small improvement in the duration of clusterctl create cluster by polling for changes immediately (#331)
• Populate the associated node's UID in the machine.status.nodeRef field (#348)
• Fixed an issue where controller logs would all be proceeded by the following error output: ERROR: logging before flag.Parse: (#349)
• MachineController can now delete nodes associated with master machines (#350)
• Added flag to clusterctl delete cluster to optionally override the provider components YAML associated with the cluster (#362)
• Increase clusterclient timeout and retry intervals (#356)
• clusterctl now requires passing in addons when creating a cluster (#376)
• The cluster name in generate-yaml.sh is now a random string (#370)
• Fix an issue where clusteractuator.Delete(...) would never succeed if the firewall rules it is attempting to delete do not exist (#378)
• The Cluster Actuator is now passed a deep copy of the informer-provided cluster object instead of the informer-provided cluster object directly (#426)
• Add a delete cluster command to clusterctl (#406)
• clusterctl no longer supports --provider google. To create a cluster on Google Cloud Platform, please use clusterctl from https://github.com/kubernetes-sigs/cluster-api-provider-gcp (#438)
• Add "kubectl validate cluster" that validates cluster and machine objects (#298)
• clusterctl no longer supports --provider vsphere. To create a cluster on vsphere, please use clusterctl from https://github.com/kubernetes-sigs/cluster-api-provider-vsphere (#440)
• The command cluster create can now accept a kubeconfig file and launch on an existing k8s cluster. This allows us not to use minikube if we do not want to (#442)
• Missing rbac resources added to apiserverclusterapiservertemplate file (#462)
• [clusterctl] the --cleanup-external-cluster flag has been renamed to --cleanup-bootstrap-cluster (#484)
• Added node-conditions into api for MachineStatus (#483)
• Remove priority field from the APIService manifest to make it compatible with 1.11+ bootstrap clusters (#486)
• Support cluster and machine in multiple namespaces (#509)