Adding compute nodes

Edit online

After installing RHOCP on IBM Z, you can add more compute nodes to the cluster.

The process that is described here assumes that you already have an existing RHOCP cluster running for more than 24-hours. In case your cluster is younger than 24-hours a compute node must be deployed in the same way as during the installation process but using an updated worker.ign file. This method consists on extracting the digital certificate from the running RHOCP cluster and used it to replace the one existing in the current worker.ign file. This role-based access control of RHOCP allows to precisely define, which permissions are granted to which set of users:

Preparing the infrastructure

There are a few things to update in the existing infrastructure like DNS, Load Balancer, and in the hypervisor to add the VM definition for the new compute node.

  1. The DNS Service configuration needs to be updated: The zone record needs to include the 'A' pointers to the new node, the compute and the etcd.
  2. The new compute node must be added to the Load Balancer: In this example, HAProxy is used as a Load Balancer.
  3. A new virtual machine needs to be created: In this example, the VM uses zFCP and not ECKD for the installation of Red Hat CoreOS.
  4. Transfer the initramfs, kernel, the parameter file, and the worker.exec to the new compute node virtual machine disk, it can be done by using FTP or the application VMUR (s390utils-base-2 rpm) if using a RHEL guest as a helper on the same LPAR as the VM guests. If installing a cluster across multiple LPARS, a RHEL guest must exist on each LPAR to upload the boot files to the VM guests on that LPAR.

Creating a new compute node based on an existing compute Ignition file from the previous RHOCP installation process

The following steps define how to start the process to install RHCOS to the new compute VM:

  1. Using a 3270 terminal emulator, the worker.exec file must be run.
  2. The node must be started to get installed with the new Ignition file.
  3. The new CSR must be approved.

For detailed commands on how to add a new compute see: