Thanks for sharing the script. Hopefully some others might find it useful. However, I don't think I need a solution along those lines. I have a fixed desktop system, I don't have any problems with the nvidia proprietary driver, and never have any need to switch it out. It's not a big hassle to reinstall the nvidia driver - it's an easy using Yast - but thought I comment here so that anyone who encounters the same update problem has an easy fix in the manner described by Philip, or by myself. It's also curious that not everyone with the proprietary drive is encountering the issue, which might be an issue in itself. Cheers, Michael On Thursday 25 February 2021, Cameron Seader wrote:
Maybe try my attachment script. Put your nvidia installer run package in a location the script can pick it up. I use this script when I'm docked to my eGPU and not on the road. I have an alternate script I run which converts my system back when i switch to going mobile. Let me know if your interested in my other script to go mobile. I think this script might fix up your problems.
Cameron Seader Technology Strategist
SUSE +1 208 420 2167 cs@suse.com
www.suse.com
________________________________________ From: Michael Hamilton <michael@actrix.gen.nz> Sent: Wednesday, February 24, 2021 4:12 PM To: factory@lists.opensuse.org Subject: Re: Beware upgrading to tw 20210222 if on nvidia GPU
On Thursday 25 February 2021, Cameron Seader wrote:
I've updated with this snapshot and its worked fine with my Nvidia 3070 I use proprietary drivers and use their scripted install. Works fine.
Cameron Seader Technology Strategist
SUSE +1 208 420 2167 cs@suse.com
www.suse.com
________________________________________ From: Michael Hamilton <michael@actrix.gen.nz> Sent: Wednesday, February 24, 2021 12:26 PM To: factory@lists.opensuse.org Subject: Re: Beware upgrading to tw 20210222 if on nvidia GPU
On Thursday 25 February 2021, Philip Raets wrote:
I had the same problem.
I rebooted to the previous kernel, reinstalled the nvidia-drivers (G04) with yast, ran mkinitrd, and rebooted.
Now it is working again.
Gr,
Philip
Similar experience here. I have a 1650 Super and use the proprietary drivers installed via the "easy" route. After 20210222, sddm failed to start. I logged into a console and used the cursors version of Yast2 to reinstall (forced "Update") any already installed packages from the nvidia repo, rebooted and was back in business. I didn't have to do anything after running yast2, I did not manually mkinitrd.
I poked around the logs a bit, but couldn't find any grumblings about modules, which I had seen under similar circumstances in the past (perhaps I was looking in the wrong places).
Michael
I don't know why it would work for some and not others. If "their scripted install" does not mean Yast, perhaps that's a factor.
Clearly the reinstall of the nvidia rpm's did the right thing, but it's weird the update did not, especially since the previous update including a kernel bump worked just fine. Maybe it had something to do with the state of my machine or combo of displays and cards.
I've look at the journalctl output for the boot that failed to start X11. Sddm did start, but did not identify any displays to add. Searching for nvidia failed to find any of normal kernel module load messages such as module license 'NVIDIA' taints kernel. The only nvidia related journal entries I could find was for audio/sound. For some reason the proprietary nvidia module wasn't included for boot - presumably it was missing from initrd.
Michael