While creating a platform agnostic UPI 4.14 cluster in a disconnected environment on vSphere using PXE boot with multiple disks, the cluster nodes fail to boot as the disk names are not persistent across reboots. The customer is using existing systems which have a 4.12 install and has some data in the secondary disk. The primary disk is of 120 GB and secondary disk is of 50 GB. With the same setup, installing a 4.12 cluster is successful, but 4.14 install fails. The customer wants to stick with even version so installation was not tested in a 4.13 install. As the install fails for 4.14, there seems to be some issues after introduction of RHEL 9. The customer cannot use disk names by UUID in their use case.   

Install a platform agnostic UPI cluster on 4.12 with primary and secondary disks. Write some data on the secondary disks. Now, on the same systems, try to install a 4.14 cluster. 

1. Install a platform agnostic UPI cluster on 4.12 with primary and secondary disks     
2. Write some data on the secondary disk.
3. On the same systems using same setup, install a 4.14 cluster.
    

The installation fails as the nodes fail to boot as the disk names change.

The disk names should be persistent across reboot and installation should succeed.

- The customer has shared the sos-report from a node, install-config file used and PXE configurations used. The below drive link holds these details.

Drive - https://drive.google.com/drive/folders/1QBB3UzFI-hrndhsivEBdjTDJtpR4I2KC?usp=sharing

- I tried to replicate the issue in my 4.14.10 cluster installed using ABI method. I tried to add a worker node to the cluster with a primary and secondary disk. The disk names are not persistent across reboots. The below drive link as the details. 

Re-producer Drive - https://docs.google.com/document/d/16tUCoNIVRkmWevw17L8KMT_kgr0f5CxQHAhhc9RKYjQ/edit?usp=sharing