nvidia - service lightdm modifing /etc/alternatives/x86_64-linux-gnu_gl_conf - Ask Ubuntu

i've installed nvidia-367 on system (ubuntu 16.04). reason don't know, restarting lightdm service leads modification of /etc/alternatives/x86_64-linux-gnu_gl_conf.

i run manually:

root@tron:~# update-alternatives --config x86_64-linux-gnu_gl_conf there 3 choices alternative x86_64-linux-gnu_gl_conf (providing /etc/ld.so.conf.d/x86_64-linux-gnu_gl.conf).    selection    path                                       priority   status ------------------------------------------------------------   0            /usr/lib/nvidia-367/ld.so.conf              8604      auto mode   1            /usr/lib/nvidia-367-prime/ld.so.conf        8603      manual mode   2            /usr/lib/nvidia-367/ld.so.conf              8604      manual mode * 3            /usr/lib/x86_64-linux-gnu/mesa/ld.so.conf   500       manual mode  press  keep current choice[*], or type selection number: 0 update-alternatives: using /usr/lib/nvidia-367/ld.so.conf provide /etc/ld.so.conf.d/x86_64-linux-gnu_gl.conf (x86_64-linux-gnu_gl_conf) in auto mode

and service lightdm restart ending with:

there 3 choices alternative x86_64-linux-gnu_gl_conf (providing /etc/ld.so.conf.d/x86_64-linux-gnu_gl.conf).    selection    path                                       priority   status ------------------------------------------------------------ * 0            /usr/lib/nvidia-367/ld.so.conf              8604      auto mode   1            /usr/lib/nvidia-367-prime/ld.so.conf        8603      manual mode   2            /usr/lib/nvidia-367/ld.so.conf              8604      manual mode   3            /usr/lib/x86_64-linux-gnu/mesa/ld.so.conf   500       manual mode  press  keep current choice[*], or type selection number:  

again. problem is, wrong ld.so.conf nvidia driver can not loaded:

/var/log/xorg.0.log:

[  1649.307] (ii) loadmodule: "nvidia" [  1649.308] (ww) warning, couldn't open module nvidia [  1649.308] (ii) unloadmodule: "nvidia" [  1649.308] (ii) unloading nvidia [  1649.308] (ee) failed load module "nvidia" (module not exist, 0)

any ideas what, , why restart touches these alternative-settings?

root@tron:~# dpkg -s nvidia-367 | grep 'version' version: 367.57-0ubuntu0.16.04.1 root@tron:~# dpkg -s lightdm | grep 'version' version: 1.18.3-0ubuntu1

p.s. tried purge nvidia* , reinstalling driver, without success.

i having same issue dell vostro 1500 (geforce 8400m gs, nvidia-340 driver) after updating recent ubuntu 16.04.2. took me day find out.

it turned out after

• setting alternatives auto (i.e. nvidia proprietary drivers)
• running ldconfig
• and manually inserting dri modules (sudo modprobe nvidia_340 nvidia_340_uvm nvidiafb)

startx run commandline started x correctly with proprietary driver. display manager when run manually (not systemd service) using proprietary driver.

after more investigation turned out culprit /usr/bin/gpu-manager (run via /lib/systemd/system/gpu-manager.service). log /var/log/gpu-manager.log saying:

looking nvidia modules in /lib/modules/4.4.0-72-generic/updates/dkms found nvidia module: nvidia_340_uvm.ko nvidia loaded? yes nvidia unloaded? no nvidia blacklisted? yes 

and updated alternatives mesa/ld.so.conf.

from point on easy find out nvidia-340 module indeed blacklisted in /etc/modprobe.d/bumblebee.conf

this apparently leftover previous driver nvidia-304, using bumblebee. nvidia-340 driver doesn't use bumblebee.

so deleting /etc/modprobe.d/bumblebee.conf solved problem. other alternative purge delete previous nvidia driver installation, other tutorials say. haven't sufficiently purged previous driver installation...