Description of problem:
The minimal iso created during a cluster install fails to boot with no visible error printed to the console.
Assisted service was deployed using the community operator from operator hub, and a cluster installation was started using the kube native API.
Version-Release number of selected component (if applicable):
100% on this specific hardware. Not reproducible using a UEFI VM
Steps to Reproduce:
1. Deploy assisted service
2. Create clusterdeployment, infraenv
3. Download minimal iso
4. boot the iso
Boot from CD fails and host proceeds to boot from a separate bootable drive.
Boot into live iso.
"PowerEdge R340 , 64 GB memory , 223 x 2 disks . IDRAC version Firmware version 4.40.00.00 and I’m mounting the iso through the Virtual Media feature as remote file share as CD"
Problem also occurs with one ironic provided ISO, but things work properly when changing the ISO_IMAGE_TYPE to full-iso in the assisted-service.
Can you give some more details on the (In reply to Ulrich Schlueter from comment #2)
> Problem also occurs with one ironic provided ISO, but things work properly
> when changing the ISO_IMAGE_TYPE to full-iso in the assisted-service.
Can you give some more details about the "ironic provided ISO"?
How was it created? Is that something we can get access to?
Looks like this was server related after all. The server showed additional (corrupted) entries in the UEFI boot sequence that most likely confused the boot process (trying to boot from a non-existing virtual CD, not the one shown in the UI) . Removing these entries help with the ironic case.
Next step retesting with the minimal-iso
Re-testing with the minimal ISO worked.
So the corruption in the boot sequence seems to have caused the observed symptoms. Unexplained things: 1) why the boot process worked with the full iso. 2) what created the rogue entries.
Anyway, from my perspective this BZ can be closed.