[VHRISO] installation failed on HPE Alletra 4140 with Intel VMD VROC

[mecki]

MOD: split from post532222.html#p532222

Hi,

tested today with HPE Alletra 4140 Storage Server (next gen Apollo)

Installation failed.
Boot device is a Intel VMD (VROC=CPU based) RAID1 with two nvme SSD 480GB, not the usual NS204 (HW RAID1), and two SR Controllers RAID60 each 300TB
Anaconda/Kickstart see that VMD/VROC RAID1 correctly in GUI, but it looked like that it tried to install on first nvme ssd of the RAID1.
Installing RHEL 9.3 worked fine with that configuration. Not tried Rocky manual installation. I know there is a requirement to use RAID1 controller for boot, and 2nd RAID controller for data.

No support ticket open yet as we are uncertain if that config should work. Which parameters is kickstarter looking for?

Thanking in advance,

[HannesK]

Hello,
can you please open a support ticket and post the case number for reference?

I have installed the ISO on the the HPE Alletra 4140 in our labs with the NS204i card, but the VMD RAID might be an issue. If RHEL installs, then it should be possible to solve that problem.

In the support case, please provide information what failed and also screenshots how it looks in the storage selection would be helpful.

Best regards
Hannes

[mecki]

support case 07531569 opened right now Included screen shots of anaconda.log and gui.

lsblk see the md device as RAID but kickstart script result in a storage config file with a nvme device as boot device but not the md RAID device.
Anaconda installer fail to save config with a nvme boot device.

Best regards,

[mecki]

it also failed without and VROC RAID device configured, when using only two single nvme devices (and the two 300TB sd devices) with the same error. Only difference is that in anaconda GUI only the two 300TB devices are displayed.

[HannesK]

Hello,
thanks for the case number and the screenshots.

Looking at this document, the VROC RAID feels like it is a software RAID. Thanks @Regnor. Software RAID is known to fail, but I remember different errors than "kickstart insufficient" (currently no plans to support that).

For the standalone NVMe drives I can only guess that it has something to do with leftovers of the RHEL installation (we work on that for a different scenario which I ran into when reconfiguring the RAID-sets and the root cause might be the same).

The two NVMe drives that are in use now are connected to one of the SR controllers, right?

Best regards
Hannes