Attempts to install nvidia-cuda-toolkit are not successful.
Below, the message that returns:
Sorry, the following packages cannot be selected:
Trying the alternative dkms-long-lived and nvidia related,
1 installation failed
there has been a problem during installation:
file /etc/ld.so.conf.d/GL.conf from installation of x11-driver-video-nvidia-long-lived-340.58-1.x86_64 conflicts with the file of package x11-server-common-1.18.4-2.x86_64
Maybe it is relevant to say that I have two graphic cards intel and nvidia (that demands bumblebee).
Damn, with nVidia Optimus?
I’ll try to find a computer with this chip unless someone else can test…
Thanks for all your patience and reports btw.
Yes, nVidia Optimus, but it works pretty well with OMV 2013, 2014.0, 2014.1, and 2014.2! I guess it will work with lx 3.0 as well as soon as I can install nvidia drivers.
Isn’t it just a x11 packaging problem?
Just filed a bug on x11-drivers dependence problem.
For my part idid not succeed yet installing this release, so I can’t check yet dep problems 
Raphael, sad to here you can’t install 3.0. Is this a proprietary driver problem? Right now as a group our support for proprietary drivers seems weak (being polite). I don’t understand when we have missing dependencies reported for days or weeks that can’t be fixed.
Forgot to come back here
I finally could install and confirm I have the same dependencies problem with my nVidia card.
As a complement, a fedora 24 user told me that its rpms for nvidia were not installed (there is no libGLdispatch as well) and, for proprietaty driver use he had to use the nvidia installer.
From Crispin Boylan on OM Cooker ML:
GLdispatch is in:
/usr/lib64/nvidia-current/libGLdispatch.so.0
so the file is not missing it was just an autodep failure of rpm.
fixed now in 367.44-1
So far, I could not install the new nvidia packages via MCC. I select dkms, nvidia cuda toolkit and nvidia cuda tookit devel (I need them) and all required dependencies. Then, when I apply, the window “Distribution Upgrade” pop up fully black, no download or installation information, and after a while this window vanish.
Don’t know yet if the cause is some internet connection problem here.
Tried command line (urpmi …) and the download seems too slow or not working. The following line is seen but nothing happens:
http://downloads.openmandriva.org/mirrors/openmandriva.3.0.x86_64.list: media/non-free/release/nvidia-cuda-toolkit-devel-7.5.18-1-omv2015.0.x86_64.rpm
It was an internet problem after all! Sorry for false alarm.
There has been this warning during installation:
warning: CICLO:
warning: to remove nvidia-current-doc-html-367.44-1.x86_64 “Requires: x11-driver-video-nvidia-current = 367.44-1” from tsort relations.
warning: to remove x11-driver-video-nvidia-current-367.44-1.x86_64 “Requires(hint): nvidia-current-doc-html = 367.44” from tsort relations.
You can safely ignore that warning.
Thanks Ben79.
Unfortunatelly, after rebooting sddm does not start and the system hangs out.
I had no graphical support.
It is really weird in this case because this laptop has its primary graphical support on intel graphical card.
I need nvidia support just to run some gpu demanding executables with the help of bumblebee.
It used to work properly at OMV 2014 but does not seems to work with OMV LX 3.0.
After uninstall of nvidia, dkms-nvidia, and its dependecies, the system started with sddm.
Does it look like the same bug as this one: 1870 – nvidia driver: blank screen instead of loading X11 ?
It can be similar. Don’t know if it has the same root. However, it seems to differ because here the basic graphical support is provided by the intel graphic card. NVidia is just another, more powerful, graphic card with GPU capabilities.
Another weird OMV LX 3.0 behaviour (related to uninstallation of nvidia drivers?):
Now mcc (drakconf) cannot be launch,
$ drakconf
No protocol specified
Cannot be run in console mode.
Tried,
$ urpmi --replacepkgs drakconf
but didn’t help.