Prerequisites for Hosted Control Plane clusters

The prerequisites before you create virtualized or Bare Metal Hosted Control Plane clusters.

1. From the Red Hat® Operator Catalog, install MetalLB 4.16 or higher.
   Add the MetalLB Operator to your cluster so that when a service of type LoadBalancer is added to the cluster, MetalLB can add a fault-tolerant external IP address for the service. For the procedure to install and validate MetalLB, see https://docs.openshift.com/. Go to your specific version of OpenShift® Container Platform and check MetalLB details.

   Note: When you set up the load balancer, other applications can also get the advertised addresses. You must have enough addresses for any workloads on this cluster and created OpenShift Container Platform clusters.
2. Create MetalLB CR based on the following example:

   apiVersion: metallb.io/v1beta1
   kind: MetalLB
   metadata:
     name: metallb
     namespace: metallb-system

3. Create an IPaddress pool.
   Note: Reserve a set of unused IPs on the same CIDR as the Bare Metal network for MetalLB. The MetalLB serves these IPs to any load-balancer service that is installed on the cluster and not just the Hosted Control Plane.

   Example:

   apiVersion: metallb.io/v1beta1
   kind: IPAddressPool
   metadata:
     name: metallb
     namespace: metallb-system
   spec:
     addresses:
     - 9.9.0.51-9.9.0.70

   Note: When you set up the load balancer, other applications can also get the advertised addresses. You must have enough addresses for any workloads on this cluster and created OpenShift Container Platform clusters.
4. Create L2advertisement based on the following example:

   apiVersion: metallb.io/v1beta1
   kind: L2Advertisement
   metadata:
     name: l2advertisement
     namespace: metallb-system
   spec:
     ipAddressPools:
      - metallb

5. Install Multi Cluster Engine for Kubernetes operator.
6. Create a Multi Cluster Engine CR instance based on the following example:

   apiVersion: multicluster.openshift.io/v1
   kind: MultiClusterEngine
   metadata:
     name: multiclusterengine
   spec:
     targetNamespace: multicluster-engine

   Change the parameter values according to your environment.
7. Wait until the Multi Cluster Hub instance is created, available, and in Running state.
   The Multi Cluster Engine instance must be available.

   Run the following command to check whether the Multi Cluster Engine instance is available.

   $ oc get multiclusterengine

   Example output:

   $ oc get multiclusterengine
   NAME STATUS AGE
   multiclusterengine Available 170m

8. Download hosted control plane CLI from OpenShift Container Platform console:
   1. Go to the Command Line Tools page.
   2. From the Hosted Control Plane - Hosted Control Plane Command Line Interface (CLI) section, download the CLI tar and extract from archive based on your platform.
      Note: In MCE, you can use the user interface instead of CLI. However, the CLI gives more options than the user interface.
9. To optimize etcd performance for hosted clusters on the hub cluster, configure LVM to use local drives on worker nodes as follows:
   1. Provision three dedicated drives on three separate worker nodes.
   2. Configure LVM on the worker nodes to manage the dedicated drives for etcd pods.
10. Follow the steps to configure the LVM:

    You must configure LVM so that the hosted clusters have a good etcd performance, which requires three dedicated drives on three worker nodes.

    1. Create a configmap similar to the following example. This configmap is used by IBM Fusion to create a LVM cluster.

       Example configmap:

       kind: ConfigMap
       apiVersion: v1
       metadata:
         name: lvm-config
         namespace: ibm-spectrum-fusion-ns  
       data:
         drives: |
           - '/dev/disk/by-path/pci-0000:61:00.0-nvme-1'
         computeNodes: |
           - compute-1-ru5.rackm01.rtp.mydomain.com
           - compute-1-ru6.rackm01.rtp.mydomain.com
           - compute-1-ru7.rackm01.rtp.mydomain.com
         nodeType: compute

       The LVM cluster uses the NVMe drive /dev/disk/by-path/pci-0000:61:00.0-nvme-1 on the three compute nodes to provide PVCs for the etcd pods of the hosted clusters.

    2. To configure LVM clusters, specify the drives and nodes to be used. The drives field specifies the drives to be used for the LVM cluster. Use one or two drives per node. To specify two drives in a node, use the following format:

       drives: |
         - '/dev/disk/by-path/pci-0000:61:00.0-nvme-1'
         - '/dev/disk/by-path/pci-0000:63:00.0-nvme-1'

       The computeNodes field lists the worker nodes that provide disks for the LVM cluster. Ensure that the selected nodes are not GPU nodes.

       The nodeType field specifies the type of nodes to use for LVM clusters. Supported types are:

       1. Compute: uses disks from worker nodes.
       2. Control: uses disks from all master nodes to configure the LVM cluster.
11. Increase the maximum number of pods per node from the default value of 250 to 500. For the procedure to update the value, see Managing the maximum number of pods per node .