#Huawei Cloud Stack Provider Release Notes

#Overview

This page tracks release notes for the Alauda Container Platform Huawei Cloud Stack (HCS) Infrastructure Provider. Each section corresponds to a release tag.

Pre-release tags (v1.0.0, v1.0) are internal builds and are not covered here. The first publicly released version is v1.0.1.

For the current full capability list and installation steps, see Huawei Cloud Stack Provider Installation.

#v1.0.4 (2026-09-01)

#Enhancements

• ACP 4.4 system component repositories — The provider selects the Kube-OVN chart repository from the target Kube-OVN version, the CoreDNS image repository from the ACP version suffix in the CoreDNS tag, and the kube-proxy image repository from the Kubernetes version. Kube-OVN v4.4 and later uses acp/chart-kube-ovn; CoreDNS tags for ACP v4.4 and later use acp; and Kubernetes v1.35 and later uses acp/k8s for kube-proxy. Earlier targets keep their legacy repository paths.

• Upgrade-time system component reconciliation — The provider configures the CoreDNS repository before control-plane creation, sets missing KubeadmControlPlane skip annotations so provider-managed system components are not overwritten, and reconciles CoreDNS and kube-proxy after the relevant control-plane rollout. This keeps system component images and charts aligned with the selected ACP and Kubernetes versions.

• Pod log placement on the containerd data disk — When a node has a dedicated /var/lib/containerd data disk, generated bootstrap configuration links /var/log/pods to /var/lib/containerd/pods so pod logs use the containerd data disk.

#Fixed Issues

• Persistent disk duplicate creation — The provider now reserves a persistent-disk slot in HCSMachineConfigPool status before submitting the EVS create request. Repeated reconciliation therefore waits for the in-flight volume instead of submitting duplicate CreateVolume requests.

• CoreDNS image pull secret reconciliation — CoreDNS pull secrets are now evaluated from the CoreDNS Deployment, preserving existing credentials and adding the sentry registry credential only when it is missing.

#v1.0.3 (2026-06-27)

#New Features

• Web UI node pool workflow stabilization — This release is the documented minimum HCS Provider version for the web UI workflow and requires Fleet Essentials 1.0.2 or later. It improves the Node Pools cards, prevents duplicate or stale subnet IP selections, validates required subnet and IP values, and corrects the day-2 Manage IPs workflow. When you create a control plane or worker node pool, the Availability Zone list uses zones reported by HCS and disables zones reported as unavailable. See Creating Clusters on Huawei Cloud Stack and Managing Nodes on Huawei Cloud Stack.

#v1.0.2 (2026-06-26)

#New Features

• Initial HCS web UI integration — The Provider package now includes the initial HCS web UI for guided cluster creation and Node Pools operations. The UI loads infrastructure values, including flavors and availability zones, through the selected HCS credential. For the documented web UI workflow and subsequent node pool fixes, use HCS Provider v1.0.3 or later with Fleet Essentials 1.0.2 or later.

• Control plane high availability (server group anti-affinity) — HCSCluster accepts a new spec.controlPlaneHA block with enabled and policy (anti-affinity or soft-anti-affinity). When enabled, the provider creates and maintains an HCS server group and spreads the control-plane ECS instances across different physical hosts. The feature covers the control plane only, the provider owns the server group rather than referencing an existing one, and existing clusters converge through control plane rolling replacement. See Configure HCSCluster for the field reference and Control Plane HA Placement Plan for planning guidance.

#Fixed Issues

• Machine creation backs off when the HCS API is rate limited — When the provider cannot create a control-plane or worker ECS instance, it now retries with a rate-limit-aware backoff instead of a fixed fast requeue. This prevents a request storm against the HCS API when one or more machines repeatedly fail to create, and the failure reason is reported on the machine's MachineReady condition.

#v1.0.1 (2026-05-20)

#Overview

v1.0.1 is the first public release of the HCS Infrastructure Provider. The provider manages the full lifecycle of Kubernetes clusters on Huawei Cloud Stack through the Cluster API model, including cluster creation, node pool scaling, control plane replacement, and machine configuration through reserved IP and hostname pools.

#New Features

• Cluster lifecycle on Huawei Cloud Stack — Create, scale, and delete Kubernetes clusters that use HCS as the infrastructure backend. Cluster creation reuses an existing VPC, subnet, and security group; the provider creates and binds the control plane ELB automatically and writes back the resulting controlPlaneEndpoint.

• Availability zone selection — Set HCSMachineTemplate.spec.template.spec.availabilityZone to place the control plane or a worker node pool in one provider-recognized HCS availability zone. Different worker node pools can use different templates and availability zones, but replicas in one pool are not automatically spread across zones. If the field is omitted, the Provider selects an availability zone reported as available by HCS. See Availability Zone Placement.

• Reserved hostname and IP pool (HCSMachineConfigPool) — Reserve fixed hostnames, subnet assignments, and static IP slots for cluster nodes. Slots are released and reused when nodes are deleted. Required when persistent node-local disks must be preserved across node replacement.

• Control plane replacement with static IPs — maxSurge: 0 rolling replacement is the default upgrade path for static-IP control planes. No additional IP slot is needed for replacement. When maxSurge > 0 or the control plane is scaled up, IP pool capacity must be expanded first.

• Cloud-init-based node bootstrap — Node bootstrap uses cloud-init user data. By design, the HCS provider uses cloud-init exclusively as the bootstrap mechanism; Ignition is not part of the HCS provider's design and is intentionally not supported.

• FQDN and hostname rendering — Node hostnames are rendered with the short name and FQDN separated correctly in cloud-init, matching customer domain naming requirements. Webhook validation enforces lowercase hostnames and rejects names containing consecutive dots; no CRD field change is required for this validation.

• Bootstrap encryption provider secret — HCSCluster accepts a reference to a Kubernetes Secret that provides the bootstrap encryption configuration, enabling encryption-at-rest for etcd data from cluster creation onward.

• OCI source and revision labels on container images — Built provider images include standard OCI source and revision labels for build traceability.

#Upgrade Notes

v1.0.1 is the first public release. There is no upgrade path from earlier internal builds; production deployments must start at v1.0.1 or later.