On my test box, which uses an Asus H97M-E motherboard (kinda old!) with Intel firmware RAID, anaconda (via blivet) does not see the firmware RAID set as a viable target for install when booting in native UEFI mode. It does see it when booting in BIOS mode. In the journal and storage.log, I see this error: Mar 19 00:55:31 node-1w7jr9qh0gahqlm1u4ziyrtql.ipv6.telus.net org.fedoraproject.Anaconda.Modules.Storage[2358]: ERROR:blivet:failed to determine name for the md array a7ff7f19-f142-4329-6b52-1dbafa835906 This is reproducible with Fedora 40 Beta-1.2 and Beta-1.7. I will try with an older release shortly, I don't recall offhand if this was the case with earlier releases.
Gonna propose as a Final blocker for discussion, though I might well ultimately be -1 to it. Also as Beta FE just in case. It would be good if folks can test firmware RAID on UEFI on other systems to see if this is a general issue or specific to my motherboard and/or RAID set.
Created attachment 2022398 [details] full journal
It looks like this affects Fedora 39 Final as well, so it's not a 40 regression.
We signed off Beta today, so there's no point in Beta FE status any more. Clearing that nomination.
Discussed at 2024-03-25 blocker review meeting: https://meetbot.fedoraproject.org/blocker-review_matrix_fedoraproject-org/2024-03-25/f40-blocker-review.2024-03-25-16.00.html . We agreed to delay the decision on this as it's hard to tell how hardware-specific it is. We will try to test on more hardware, but also Vojtech, if you have any idea how widespread the effect of this is likely to be it would help with the blocker decision.
Discussed at the 2024-04-01 blocker review meeting: https://meetbot-raw.fedoraproject.org/blocker-review_matrix_fedoraproject-org/2024-04-01/f40-blocker-review.2024-04-01-16.01.html . We agreed to punt on this issue as the scope is not clear due to insufficient information required to make an informed decision. It was agreed that testing on more hardware would be very valuable to identifying the source of this regression. If anyone can test the available Beta artifacts against a system which uses Firmware raid (i.e., RAID implemented and supported by the motherboard, not by Software (e.g. mdraid) nor Hardware (external cards), it would be very welcome.
I've tested F40 Beta on a firmware RAID in UEFI mode yesterday on Gigabyte GA-Z170M-D3H (Skylake-S). It worked fine. So this is probably hardware specific.
-4 in https://pagure.io/fedora-qa/blocker-review/issue/1533 , marking rejected.
Adam, I just tested the same hardware from comment 7 again in BIOS mode, and found out that I don't see any RAID devices! The error was the same "failed to determine name for the md array". In lsblk --fs, I saw /dev/sda and /dev/sdb as isw_raid_member, but there were no md* devices underneath them. However, I found a solution, for my case. I had to go to BIOS and switch Storage Boot Option Control from UEFI to Legacy. Then the RAID started to be available in BIOS mode (and continued to be available in UEFI as well). So quite possibly if you tinker with your BIOS options, you might be able to convince it to run under UEFI as well (or at least switch it from BIOS mode to UEFI mode).