Skip to main content

During the installation of the NKP cluster you need to set an IP-address for the CONTROL_PLANE_ENDPOINT. 

Unfortunately the rocky linux node (nkp-rocky-9.5-release-1.31.4-20250409153435.qcow2) also tries to obtain an IP-adress via DHCP on the same nic. 

As a result the nic gets 2 IP-addresses in the same subnet (1 from the installer and 1 from the DHCP)  and reports back to nkp with the DHCP address. 

This results in a failing install of nkp cluster. 

The Control Plane Endpoint is like a VIP and is essential for you to communicate with the k8s API otherwise it won't work so you need to put in a unique IP that isn't used for the control plane endpoint   The node gets its own IP so it's different, so I believe what your seeing is more expected.  


I don't think you know what the issue is. 

The control endpoint becomes multihomed on the same IP-range and reports back to the installer with a different IP address than the installer expects. 

This causes the installation to fail while it keeps waiting for the full cluster to report back. 


Problem persisted at the keyboard ;-)

The Vlan we assigned the control plane to was not able to communicate with IP's outside the VLAN and thus not with the Nutanix platform in an other vlan. 


Reply