#Kubeadm Provider Release Notes

#Overview

This page tracks release notes for the Alauda Container Platform Kubeadm Provider, the Alauda-packaged distribution of the upstream Cluster API kubeadm bootstrap and control plane providers for the Alauda Container Platform.

The Kubeadm Provider is platform-wide and is consumed by every infrastructure provider on this site (HCS, DCS, vSphere, Bare Metal) for cluster bootstrap and control plane management.

For installation and capability details, see the platform installation documentation.

#v1.0.14 (2026-08-31)

#Enhancements

• Alive rollout gate for bare-metal clusters — Control-plane machine replacement is now gated on Alive coverage of the remaining nodes. Deleting the last old control-plane node before its replacement has keepalived running would drop the Self-built VIP and leave the cluster unable to update the control-plane node identity. The gate blocks the deletion until the remaining nodes are covered.

  The gate is scoped to bare-metal clusters by default. Its release criteria depend on how Alive is actually deployed — the interface carrying the VIP, the app=alive static pod labels, and whether the VIP lands on a node interface — and that has only been verified on bare metal. Other infrastructure providers are not gated until their own deployment shape is verified.

#Fixed Issues

• Control-plane rollout recovers when Alive static pods are missing — Within the same bare-metal scope as the gate above, a rollout could stall permanently when the recorded control-plane node identity was already current but a node it claimed had no running Alive static pod, because the trigger and the readiness gate never exchanged state. The controller now detects that condition and re-renders the Alive chart to reinstall the missing static pods, throttled to avoid repeated re-renders, and with concurrent reconciles resolved so that only one of them performs the repair. Previously the only recovery was to hand-edit the Alive AppRelease values. A bare-metal reprovision clears /etc/kubernetes and removes Alive from a host after a successful install, which is how the condition arises.

• Kubernetes API server to kubelet permissions during control-plane replacement — Cluster API updated to a build that backports the kubeadm permission fix, resolving a control-plane replacement upgrade failure.

• CVE remediation across all three images — The Go builder base image was unified and vulnerable Go modules were upgraded in the bootstrap, control-plane, and rollout-gate images. Cluster API was updated to v1.10.6-alauda.17, which clears the reachable advisories in google.golang.org/grpc, go.opentelemetry.io/otel/sdk, and github.com/moby/spdystream.

#v1.0.12 (2026-07-28)

#Fixed Issues

• Standby Global Cluster protection — Kubeadm bootstrap and control-plane reconciliation remain inactive on a standby Global Cluster. Replicated workload-cluster resources are therefore not acted on until the standby cluster is promoted during disaster recovery.

#v1.0.11 (2026-06-10)

#Fixed Issues

• More reliable control-plane rollout readiness — The rollout gate now validates the identity of replacement control-plane nodes and waits for a stable control-plane node set before refreshing platform registry state. This prevents stale node records from leaving a rollout waiting indefinitely.

#v1.0.10 (2026-05-25)

#Fixed Issues

• Platform registry refresh after control-plane VM replacement — A Global Cluster now refreshes the platform registry state when a replacement control-plane VM reuses the prior node name and IP address. This allows the rolling update to continue after node-local platform services are prepared on the replacement VM.

#v1.0.9 (2026-05-21)

#Notes

Internal design documentation update for KCP rollout gating. No user-facing behavior change.

#v1.0.8 (2026-04-09)

#Enhancements

• Go 1.25 toolchain — Provider built with the Go 1.25 toolchain for security and maintenance updates.

#v1.0.7 (2026-03-06)

#Fixed Issues

• Upstream Cluster API CVE patches — Upstream Cluster API dependency bumped to a build that resolves disclosed CVE issues.

#v1.0.6 (2025-12-25)

#New Features

• Machine taints synchronization — Taints declared on the platform Machine resource are now synchronized to the underlying Kubernetes node by the provider, ensuring node taint state matches the declarative spec.

#v1.0.5 (2025-08-28)

#Notes

v1.0.5 is a re-tag of v1.0.4 (same underlying commit). No additional changes beyond v1.0.4; see v1.0.4 for the release content.

#v1.0.4 (2025-08-28)

#Enhancements

• Stabilization and dependency refresh — Maintenance release rolling up minor stabilization changes.

#v1.0.1 (2025-07-09)

#Enhancements

• Internal naming cleanup — Adjusted provider component naming for consistency. No user-visible API change.

#v1.0.0 (2025-06-18)

#Overview

v1.0.0 is the initial release of the Alauda Container Platform Kubeadm Provider. The provider packages the upstream Cluster API kubeadm bootstrap and control plane providers as an Alauda Container Platform module, plus Alauda-specific control plane rollout enhancements.

#Initial Capability Set

• Upstream Cluster API kubeadm bootstrap and control plane — Bundles the upstream Cluster API kubeadm bootstrap provider and the KubeadmControlPlane controller, distributed as an Alauda Container Platform module.

• Platform-aligned installation flow — Installs through the Alauda Container Platform's standard provider installation flow.

• KCP rollout enhancements — Alauda-specific rollout logic on top of the upstream KubeadmControlPlane controller, controlling control plane node replacement behavior during version upgrades and machine replacements.

• Foundation for infrastructure providers — Provides the bootstrap and control plane layer consumed by the Huawei Cloud Stack, Huawei DCS, and VMware vSphere infrastructure providers on this site.