I had a hardware problem on my system (user error in the end). After resolving it, Rocky wouldn’t load the NVidia driver. I was due for a clean install, so I put a fresh copy of the Rocky 8.5 ADSK ISO on it. Knowing that the DKU has trouble with the NVidia driver on my system (default config, not a reference system), I did the manual install of the NVidia driver for the A5000 which worked fine. System was operating with NVidia driver, installed DKU with --nonvdriver and installed Flame 2023.3. Flame launched and ran good. So far so good. DKU is 17.3.
Needed to make another tweak requiring a reboot (reload BMD driver, first after Flame install) and now I’m back to the same NVidia problem. In /var/log/messages there are entries that read
Failed to initialize NVidia kernel module. No devices detected.
Obviously hardware is fine because it worked just prior to the DKU install. I did notice that during the DKU install it said something about disabling graphics. And it in fact had set the systemctl default target to non-graphics. So I reverted that. But still no luck.
Figured maybe the DKU did something funny with xorg.conf. There are numerous in the folder, labeled ‘beforeDKU_install’, ‘dku_standard’, ‘flame.v2023.3’, ‘nvidia-xonfig-original’. I tried a few of them, but no luck.
After reboot it stays in text mode with just the blinking cursor in the top left. With Ctrl-Alt-F2 I can get to a shell and login, but can’t convince it to load NVidia driver properly.
Any ideas?