Have you tried the Code Ready Containers release candidate? It's OpenShift 4 on your laptop!

[flozano]

Visit try.openshift.com, click 'Get started' and then 'laptop'. Click the link that relates to your operating system.

Or download the installer directly to https://mirror.openshift.com/pub/openshift-v4/clients/crc/latest/

Installation instrutions are at: https://cloud.redhat.com/openshift/install/crc/installer-provisioned

The previous beta releases were very stable on my machine.  :-)

[flozano]

Caveat: Before Code Ready Containers reach GA, you have to download each new release (each month or so), destroy and recreate your crc VM.

[joelbirchler]

Seems to work, although oc get cs shows an unhealthy controller-manager.

[DanielD]

Everything is fantastic - Excep...  For normal situation CRC is not exactly useful. 
Mind you, Normal means different things to different people - to me, itr means Libvirt is already installed on the laptop.

I assume everyone has an installation of libvirt on their laptops - especially if you happen to, also, run a Windows workstation.
You do, of course, run Windows in a Librt VM, don't you ?!? 
(I am aware that, sadly, it is not the case !)

All joke/irony/sarcasm aside, CRC is trying to install Libvirt and build a new VM.

How can we circumvent that ? 

I think we are somewhat forced to try "Installing OpenShift 4.1 Using Libvirt and KVM"
(https://servicesblog.redhat.com/2019/07/11/installing-openshift-4-1-using-libvirt-and-kvm/)

Please let me know if anyone figured how to run only the "single host cluster configuration" that happens on the VM

[flozano]

On my RHEL8 machine CRC used the distribution's packages for libvirt that I already had installed. But yes, CRC wants to create a new VM (running RH CoreOS) and manage it. You need to run CRC on your bare metal host OS. If you notebook runs Linux, you download and run CRC for Linux If you notebook runs Windows, you download and run CRC for Windows.

[DanielD]

Thank you,

Prior to my initial question, I did run the installer (crc) in a pre-existing VM. This is a nested virtualization example and - you can imagine - things don't rrun very fast in my CRC instance.

I need to run things in a bit of different way because the libvirt networks and stroage are, alraedy, predefined,

What I do not know is:

1. how to obtain the disk image that's being installed on the VM
2. how to convince "crc setup" that it does not need to create a VM, since I have the VM ready and waiting.

My mistake - I was expecting something like installing a 'controller/master' (may even be named "crc") in a container, on the "host" (bare metal laptop) that would talk to a RHCoreOS instance  installed in the VM

[flozano]

I don't know if CRC supports the kind of tweaks you want out-of-the-box, much less if it would easy or hard to achieve.  Maybe you can find something on CRC's project at GitHub:

https://github.com/code-ready/crc

On OpenShift 3 there was 'oc cluster up' that run the OpenShift master and node services as a single container, using the container engine of its host to run pods. OpenShift 4 does not provides that option anymore, and wants to control the VM it runs in.

Here I run CRC alongside other VMs that I create with virt-manager on my RHEL machine. If you realy require a very customized setup maybe you need to perform a bare metal installation, that will require at least three master VMS, and I don't know if you can tweak then to also take worker roles.