OCP Install on Nutanix In disconnect setup

[kiran34254]

Hello everyone,

i have one query can i Install OCP On nutanix in disconnect setup using UPI method

[Sunnykumar1]

@kiran34254 

Yes, you can install OCP on Nutanix in a disconnected environment using UPI. This is the recommended and supported approach for Nutanix + air-gapped setups.

IPI is not supported, however UPI is fully supported.

[kiran34254]

Thanks for replying,
Can I get documents for reference same

[Sunnykumar1]

@kiran34254 

Please refer to this document: https://docs.redhat.com/en/documentation/openshift_container_platform/4.20/html-single/installing_on...

[kiran34254]

I have already referred this document but for UPI method I need create ignition file with static ip and other host details and paste this file in user data in nutanix vm create time please correct me if I am wrong
And if you have any sample ignition file can you share with me

[Sunnykumar1]

@kiran34254 

Please note you don't have to manually edit the generated ignition files to add IPs and hostnames.

Static IP and host details are injected via kernel arguments or NetworkManager keyfiles, not by editing the ignition JSON directly.

Editing ignition manually is unsupported and risky.

Solution:

You can pass static IP details at VM boot time.

Example kernel args:

ip=192.168.10.21::192.168.10.1:255.255.255.0:master-0.ocp.example.com:ens3:none \
nameserver=192.168.10.10

This is the most common and Red Hat–recommended approach.

[kiran34254]

Thanks for your support I will do same as mentioned and update you

[Chetan_Tiwary_]

@kiran34254 

When you create your Nutanix VMs manually using the UPI method, the setup works a lot like a bare-metal install. Since you aren't using DHCP, you have to give each VM its "marching orders" at the very first boot.

You do this by adding specific kernel command arguments to the RHCOS boot screen. These arguments tell the VM its static IP address, nameserver, and gateway. You also provide the URL for your local HTTP server where you stored the Ignition files.

When the VM starts up, it uses those kernel args to get online. Once it has a network connection, it reaches out to your local server to grab its specific Ignition file (bootstrap.ign, master.ign, or worker.ign). The Ignition file then takes over to finish the configuration.

Because every node needs a unique IP, you will essentially have a specific boot string ( iPXE scripts ) for each master and worker. It's a bit more manual work, but it's the most reliable way to get OCP running in a restricted Nutanix environment.

https://docs.redhat.com/en/documentation/openshift_container_platform/4.20/html-single/installing_on... 

[Chetan_Tiwary_]

@kiran34254 Let me inform you that Red Hat's recommended fully supported OCP installation on air gapped network with scope of automation is IPI for obvious reasons of simplicity and agent based installer assistance. 

You can find the same in official RH and Nutanix docs :

https://docs.redhat.com/en/documentation/openshift_container_platform/4.17/html-single/installing_on... 

https://www.nutanix.dev/2023/12/28/openshift-on-nutanix-disconnected-installation/ 

https://portal.nutanix.com/page/documents/solutions/details?targetId=TN-2030-Red-Hat-OpenShift-on-Nu... 

Yes it is possible for a UPI based installation which is also supported in offline context but you must handle all provisioning and mirroring yourself. 

In a UPI setup, you take the driver's seat. Instead of the installer doing the work, you manually build the VMs, set up the networking, and handle the load balancers and DNS entries yourself. This path is perfect if you need total control over your environment or if automated methods aren't an option.

To get it running, you will need to:

Pre-build your RHCOS VMs for the bootstrap, control plane, and worker nodes.

Manage all the networking, including static IPs, DNS, and load balancing.

Host your Ignition files on a local web server so your nodes can grab their config.

Mirror every required container image to a local registry inside your secure network.

Every VM you create needs to follow a few simple rules to work correctly. First, make sure you use the right size and flavor as listed in the Red Hat support guides. Since you are in a disconnected setup, each VM must be able to reach your local mirror registry to pull images. You also need to ensure they can fetch their Ignition files over HTTP or HTTPS during the boot process. If you have ever done a bare-metal install using the UPI method, this will feel very familiar to you.