Prerequisites for Bare Metal Workload Clusters

Meet the following prerequisites before you register the hosts and create the cluster. They also apply to every host that you add to a cluster later.

Providers

Install these cluster plugins on the global cluster:

  • Alauda Container Platform Cluster Bootstrap Provider
  • Alauda Container Platform Infrastructure Provider for Bare Metal, which installs the bare-metal manager and elemental-operator

If the global cluster was installed on Bare Metal with ac, it already runs both. Otherwise, download them from ACP Common Aligned Extensions (see Download), upload them with violet push (see Upload Packages), and install them as described in Cluster Plugin. In a disaster recovery pair, install them on both global clusters.

Platform Certificate

Applies to: a platform that uses a user-provided certificate for the platform access address. Skip this section when the provider is already installed and elemental-operator is running.

Run the following commands from a machine where ac is logged in to the platform:

ac config use-cluster global
ac -n cpaas-system get secret dex.tls -o jsonpath='{.data}' | jq -r 'keys | join(",")'

Continue when the keys include ca.crt. Otherwise, set elemental.tls.caCertSecretKey: tls.crt in spec.valuesOverride of the ModuleInfo of the Infrastructure Provider for Bare Metal:

spec:
  valuesOverride:
    acp/chart-cluster-api-provider-baremetal:
      elemental:
        tls:
          caCertSecretKey: tls.crt
  • If the plugin is not installed yet, install it with YAML, as described in Installing via YAML, and include this value in the ModuleInfo. The Web Console form has no field for this setting.

  • If the plugin is already installed, add this value to its existing ModuleInfo and keep the other keys under valuesOverride. The ModuleInfo name is generated at installation, so look it up by the plugin label:

    ac get moduleinfo -l cpaas.io/module-name=cluster-api-provider-baremetal

    Then run ac edit moduleinfo <name>, replacing <name> with the NAME in the output.

Network Connectivity

Hosts register, install, and join the cluster over the network. Before the first host boots the ISO, open these flows:

The platform maintains the host firewall on an Alauda OS node; see Managing Firewall Ports.

Do not boot a workload host from an ISO that was built while the global cluster was being installed. Such a host registers with the bootstrap node.

Control plane endpoint. For a Self-built VIP, the VIP must be in the same Layer 2 domain as the control plane node IPs, and its VRID must be unique in that domain. VRRP and gratuitous ARP must be allowed. For an external load balancer, provision the listener and backends before you run create; see Plan the Control Plane Endpoint.

First-boot addressing. If a host gets no DHCP address in the live ISO, configure a static address from the host console; see Network in the Live ISO.

Image Registry

Decide which registry the cluster pulls images from before you run create; you cannot change it after the cluster is created.

  • The registry of the global cluster, the default. ac reads its address from the registry key of cpaas-system/public-registry-credential on the global cluster. The Secret must exist even when the registry needs no authentication.
  • A dedicated registry for the platform component images. Follow Choose the Image Registry for a Workload Cluster.

With either choice:

  • Each time a host is reinstalled, it pulls its OS image from the registry of the cluster, with the registry account of the cluster, whatever registry the image catalog entry names. If the registry requires authentication, the account must be allowed to pull the OS images. With a dedicated registry, push the OS images into it as well; see Push the OS Images.
  • The ISO is the exception: the global cluster builds it from the baseImage of the SeedImage, exactly as written there.

Hosts

Prepare the OS as described in Prepare Alauda OS for Hosts, and register the hosts as described in Register Hosts. A cluster needs one available host for each node: as many hosts as controlPlane.replicas plus the replicas of every node group in computes. For example, 3 control plane nodes and 3 worker nodes need 6 hosts.

Node Labels and Taints

To reserve the nodes of a pool for monitoring, logging, or ingress, plan their role labels and taints before you create the pools, as described in Infra Nodes.