nvidia often fails to work after kernel update?

It is nvidia-bumblebee. Followed all tutorials and it used to work with kernel update.

But two recent kernel updates both break nvidia-bumblebee. I had to reinstall the whole nvidia-bumblebee thing a few weeks ago after a kernel update, wishing it might be some major kernel update.

But today after a kernel update, nvidia fails to work again. “modprobe nvidia” shows “modprobe: FATAL: Module nvidia not found”

It is fine if it happens only occasionally so I will spend another 30 mins to reinstall the whole nvidia thing. But if it happens all the time, I wonder what is the cause? Any permanent fix?

Sorry - but not an answer

I’m having exactly the same problems:
Every odd time after a kernel update the nvidia driver can’t load - just as you describe.

With the update published today the normal fix (delete all nvidia packages and then reinstall these)
hasn’t worked sofar.

This is becoming a little bit more than a nuisance.

According to the modprobe man-page the directory /lib/modules/uanme -ris used
to load modules.
In the /lib/modules directory I notice that there are directories for:
3.16.6-2
3.16.7-13
3.16.7-21
3.16.7-7
3.7.10-1
.
.

All are dated today

What pussles me is what the 3.7-stuff is doing - that is from a previous 12.3 installation (the current 13.2
is a direct upgrade from 12.3)
And why does the deinstallation/installation og the nvidia packages touch the old directories

I really hope somebody can shed some light on this

Regards
Martin

The previous kernel update (to 3.16.7-13 as I recall) broke compatibility for kernel modules such as nvidia. The update was withdrawn. The recent update to 3.16.7-21 fixes that and restore compatibility.

But – there’s a problem. If you did install 3.16.7-13, then Nvidia would have stopped working. So you had to re-install Nvidia drivers to fix that. But, as a result, your installed drivers are compatible only with 3.16.7-13, and are not compatible with 3.16.7-21. So your best bet is to reinstall again. Hopefully this mixup won’t happen again.

# zypper ref
# zypper up

# zypper rm nvidia-*
# zypper rm bumblebee primus dkms bbswitch
Delete all mentions of nvidia in /etc/sysconfig/kernel
----"zypper if xf86-video-intel"
# mkinitrd
 
reboot
# zypper in bumblebee
# zypper in dkms
# usermod -G video,bumblebee *username*
# systemctl enable bumblebeed
Check  /etc/modprobe.d/50-blacklist.conf and make sure the line "blacklist  nouveau" is present (even if you plan to use nouveau driver). If not,  add it and as root, run:
# mkinitrd
----------------------Installing nvidia driver
# zypper in nvidia-bumblebee
# systemctl enable dkms
# zypper in nvidia-bumblebee-32bit
 
reboot
"optirun --status" and "optirun glxspheres" for testing


Hope this helps ur reinstallation. It is a routine I do every time it fails after a kernel update. It costs about 30 mins to finish.

Thanks. BTW, when you change stuff like this, will it be posted to inform the users?

I’m just an end user. This seems to have been an unintended mistake. It’s hard to warn people of that, until the problems have been reported.

It seems you missed something essential when you joined for these forums. We are all users like you!