Openshift IPI on VMware

[DerekYeo]

Hi,

I trying to understand how does the openshift 4.5 IPI architecture works. From the document excerpts below, it says required two static IP addresses. How should I configure my load balancer (HAproxy)?

  Required IP Addresses

An installer-provisioned vSphere installation requires two static IP addresses:

• The API address is used to access the cluster API.

• The Ingress address is used for cluster ingress traffic.

You must provide these IP addresses to the installation program when you install the OpenShift Container Platform cluster.

Thanks.

[franrebo84]

I have the same doubt. Did you solve this?

[benko]

Hey, all,
Just a heads up - IPI does not need a reverse proxy as it provisions those VIPs as keepalived pods in OCP.
You can have a look at that shortly as DO322 is about to be released soon as a video classroom course.
Cheers,
Gregs

[bchardim]

Hi,

As @benko said, on an OCP4 IPI installation on VMWare (vSphere) you do not need to configure the OCP4 LB, the installer does it for you. The IPI installer configures for you the keepalived static pods and VIPs on OCP4 nodes that provide the API and APP Ingress LB entry points for your OCP4 cluster.  The OCP4 API LB keepalived static pods run on the control plane nodes and the OCP4 APP Ingress LB keepalived static pods run on the compute nodes.

If the OCP4 cluster is designed to provide public services, the administrator must configure API and APP Ingress LB VIPs to make them available on the public network. See https://docs.openshift.com/container-platform/4.6/installing/installing_vsphere/installing-vsphere-i...

Here you have a detailed diagram about the OC4 LB implementation for IPI vSphere. Although this diagram is for an OCP4 IPI installation on OpenStack, the API and APP Ingress LB implementations are the same as for OCP4 vSphere IPI installation.

On DO322 course, you can find more detailed information. Hope this helps!