Skip to main content
Version: v5.0 Stable

What's new in Platform v5.0

Platform v5.0 is built for providers who run isolated clusters for many customers. It adds Tenants as the boundary for each customer, and brings bare metal servers into a machine inventory that the platform hands out to those customers. This page covers the main changes. For the full list, see the Platform v5.0 release notes.

Give each customer a tenant​

A Tenant is a customer-organization boundary above projects. A platform admin creates a Tenant for each customer and grants it the node types, OS images, templates, SSH keys, connectors, and control plane clusters it may use, and inventory it isn't granted stays hidden from it. Inside the Tenant, a tenant admin manages users, teams, and projects without seeing other customers. Tenant admins can impersonate users within their own tenant.

Each Tenant can carry its own branding in the platform UI, and access to clusters through the platform requires an explicit cluster access grant.

Allocate bare metal to tenants​

A Metal3 or NICo node type either draws on machines assigned to a Tenant, so the customer gets dedicated hardware, or places nodes on unassigned machines. Administrators select machines by label, lend network environments to Tenants, and upload OS images that machines provision from. A NetBox connector imports tagged NetBox devices into the inventory for the Metal3 provider, and Metal3 network environments can use Aviz networking. Bare metal machines can be started, stopped, and rebooted from the platform.

Node pools move a cluster's node choices out of vcluster.yaml and into the platform. Administrators manage them during cluster creation, in templates, and on the cluster's Node Pools tab.

Node claims report clearer failures, such as provisioning errors and machines that never boot their deploy image. Platform admins can browse a server's Redfish resources from the UI.

Deliver software stacks to tenants​

Stacks deploy several applications to a cluster as one dependency-aware unit, such as a model server with the services it depends on. Administrators publish Stack templates, and project users apply them. Platform v5.0 completes the Stack workflows for administrators and tenants. The certified bundle adds NVIDIA NVKE Management Plane.

Connect and observe​

Connectors are now managed through a new Connector API, and platform admins can grant connectors to Tenants. NetBox and OS image store connectors are new in v5.0. The observability gateway forwards telemetry to additional destinations, and edge collectors can collect pod logs.

Recover and operate the platform​

A platform restored from a management backup keeps its identity. Administrators can recover access when no admin can sign in. Clusters registered in Argo CD carry project and instance labels for ApplicationSet generators. These also shipped in v4.12 patch releases.

Limited availability​

Platform v5.0 is a limited availability release, recommended only for new deployments. Keep existing deployments on Platform v4.12 and vCluster v0.37, the current LTS releases. For upgrade guidance, contact vCluster Labs.

Compared with v4.12, Platform v5.0 removes shared secrets, cost control, the project-level Argo CD integration, and the bundled infrastructure apps. Apps and the example Helm, Bash, and Manifests apps remain available. It deprecates project namespaces (spaces) and the Istio integration. See the removed features and breaking changes in the Platform v5.0 release notes.