Architectural unit	Definition
OpenShift OKD cluster	A container for nodes.
OpenShift Node	Can be an instance of physical machines or VMs. The OpenShift nodes run pods.
OpenShift pod	A group of one or more containers that operates together and are deployed on the same host.
OpenShift cluster	A container for nodes that must contain at least one master node.
OpenShift master node	A server instance that is responsible for managing the entire cluster. It contains the controller manager, API server, scheduler, and etcd.

Check	Task
Nodes	•Linux running in a native LPAR or under IBM z/VM or K/VM hypervisor •Minimum of RAM: 8 GB •Minimum of hard disk space:  – 15 GB for /var/  – 1 GB for /user/local/bin  – 1 GB for /tmp/ •15 GB of unallocated storage for Docker storage. (See the OpenShift ODK official documentation: Docker storage configuration) •For more information, see OpenShift’s Configuring Docker Storage topic: https://docs.openshift.com/container-platform/3.11/install/host_preparation.html#configuring-docker-storage
Linux guest	Base OS: •RHEL or ClefOS 7.4 (or later) with the latest packages from the Extra channels. •Check OS release with this command: cat /etc/redhat-release
Devices	IBM ECKD, FBA, or zFCP

Environment details	Command-line display
Red Hat release: cat /etc/redhat-release
Kernel version: uname –a
Disk configuration: lsdasd

Workaround: The preceding command might fail with the error that is shown in Figure 6. In this case, you must downgrade the version of Python that is already installed in your host by using the following command. Then, rerun the failed installation after you downgrade the Python version. yum downgrade python-javapackages

Note: In the environment used for this installation, the package shown in “Installing centos-release-ansible26” on page 10 was already installed. The output of the command would be then different if you proceeded with a first-time installation of this package.

Note: To ensure that etc/resolv.con file is not overwritten on each reboot, verify that in the /etc/sysconfig/network-scripts/ ifcfg-enccw0.0.0d20 file, the variable PEERDNS is set to no (PEERDNS=no). Reboot the VM, and check that /etc/resolv.conf has not been overwritten.

Note: In this installation, we disable the installed firewall, and we use iptables rules. Otherwise, the OpenShift ansible playbook would disable it and mask it, and the defined rules would be lost.

Note: As mentioned before, we create the Ansible Playbook to install the all-in-one OpenShift cluster. It is recommended for a production environment to use dedicated infrastructure nodes to host the different components of OpenShift cluster.

Note: For this document, the explanation sections in Example 10 are in bold font. If you use this example inventory file as the basis for your own inventory file, first copy and paste the text into a raw-text editor like Notepad on Windows. This action ensures that the text is clear of extraneous formatting code.

Note: The prerequisites.yml file is already present in the following location: /usr/share/ansible/OpenShiftansible/playbooks/

Note: If this file is not located in the /etc/ansible/hosts directory, use the –i argument followed by the path of your inventory file in the previous command.

Note: To check whether some components such as the registry and router were correctly deployed, run the following command: oc get po -n default

Note: The host name (in our installation, OpenShiftmaster.mop.fr.ibm.com) must resolve to a name on the DNS server or in the server host file. If you try to access the OpenShift web console and the access is denied, you must add the following entry to your hosts file (/etc/hosts on Linux and C:WindowsSystem32driversetchosts on Windows): <Master_node_IP address> <hostname> For example, we entered the following in our C:WindowsSystem32driversetchosts directory: 10.3.58.89 OpenShiftmaster.mop.fr.ibm.com