I’ve upgraded from 15.3 to 16.0 and ditched all the WMs / DMs to install LXQT only, which brings in SDDM.
SDDM isn’t starting, it’s complaining:
Jun 28 15:57:05 XXXXX sddm[7938]: Adding new display...
Jun 28 15:57:05 XXXXX sddm[7938]: Loaded empty theme configuration
Jun 28 15:57:05 XXXXX sddm[7938]: Xauthority path: "/run/sddm/xauth_HMAfPG"
Jun 28 15:57:05 XXXXX sddm[7938]: Using VT 2
Jun 28 15:57:05 XXXXX sddm[7938]: Display server starting...
Jun 28 15:57:05 XXXXX sddm[7938]: Writing cookie to "/run/sddm/xauth_HMAfPG"
Jun 28 15:57:05 XXXXX sddm[7938]: Running: /usr/bin/X -nolisten tcp -background none -seat seat0 vt2 -auth /run/sddm/xauth_HMAfPG -noreset -displayfd 16
Jun 28 15:57:05 XXXXX sddm[7938]: Failed to read display number from pipe
Jun 28 15:57:05 XXXXX sddm[7938]: Display server stopping...
Jun 28 15:57:05 XXXXX sddm[7938]: Attempt 1 starting the Display server on vt 2 failed
how can I reset all relevant configs to default so I can localise where the issue lies?
…I’ve been fiddling to try & get it to work.
I’ve added to the service the 10 sec startup delay mentioned elsewhere, and have reverted the NVIDIA driver to G06 (it had upgraded) and the NVIDIA proprietary drivers as the Open Sauce driver wasn’re recognising the card.
My aim is to have local graphical login when needed but mostly use it through RDP / VNC, i.e. multi-user as default target not graphical.
Apologies - misinformation as I was getting stressed (running out of time). Box was at 15.6 before upgrade to 16.0.
I think I have solved it (will repost if not):
The NVIDIA driver wasn’t loading and I hadn’t checked throughly. The proprietary driver for G06 (v 580.159.03) work for my card (GeForce GT 1030) but the open driver doesn’t. I’d not checked after the switch to the “recommended” drivers. I’ve now backtracked and have the proprietary NVIDIA driver and that’s working.
Now doing a reinstall of LXQT after having uninstalled SDDM & LXQT & X11 & xdg and anything else which looked like a part of the graphical desktop.
That’s fixed it. Now to get persistent VNC through SSH tunnel working, and I’ll be happy.
Need to log a ticket for the open NVIDIA driver having different capabilities to the proprietary driver with the same version number. Tomorrow morning - all done for today.
The proprietary driver with the same version number does support that card, and the documentation on openSUSE says “find what version you need and install that”, which gives G06 for my card.
The families of card - I’ve no idea about. If you want to add a comment in there about either the age of the card or card family identifiers (I wouldn’t know a Pascal from a Kilobar) within scope that would be helpful. There was nothing I saw there which indicated that the open driver was limited.
…aah, discovered what a Pasacl is - it’s an architecture! I didn;t know I had one of those: it’s never been significant before.
This is added now (so that you don’t need to click a link):
For newer GPUs from the Turing, Ampere, Ada Lovelace, or Hopper architectures, NVIDIA recommends switching to the open-source GPU kernel modules.
For older GPUs from the Maxwell, Pascal, or Volta architectures, the open-source GPU kernel modules are not compatible with your platform. Continue to use the NVIDIA proprietary driver.
Thanks Hui. I’m not sure it would have prevented the problem (I haven’t paid any attanetion to the architectures of any parts of the computers I use) but that may have helped me find the problem faster.
And, if it was you - thanks for the CUDA installer. I spend an hour or two every time there’s an upgrade needed trying to work out which versions fit together - and usually get it wrong the first time.
