Fedora 22 NVIDIA proprietary drivers and NoMachine black screen

[tyrluk]

Hi there – hope someone can help point me in the right direction for this.

Running into an issue that appears to be as described in article ID AR09L00814 @ https://www.nomachine.com/AR09L00814

I am trying to connect to the primary physical display though – not specifically using virtual desktops.

My NoMachine server is running on Fedora 22, non-Wayland GNOME 3 session, with the NVIDIA proprietary drivers installed from rpmfusion.
When I login to this server from a Mac OS X, Linux, of Windows client I get a black screen.
I have tried making the change to /usr/NX/etc/node.cfg as directed in the article but this does not seem to work. The article also states at the bottom that this should not need to be done for libraries in the default path for libraries.

Any help would be appreciated. Thanks in advance.

Location of mesa-libGL files:
$ rpm -ql mesa-libGL | grep libGL.so.1
/usr/lib64/libGL.so.1
/usr/lib64/libGL.so.1.2.0
/usr/lib/libGL.so.1
/usr/lib/libGL.so.1.2.0

Have also searched the file system for *libGL* and found the following files:
$ sudo find / -name *libGL* | sort
/usr/lib64/libGLESv2.so.2
/usr/lib64/libGLESv2.so.2.0.0
/usr/lib64/libGLEW.so.1.10
/usr/lib64/libGLEW.so.1.10.0
/usr/lib64/libGL.so.1
/usr/lib64/libGL.so.1.2.0
/usr/lib64/libGLU.so.1
/usr/lib64/libGLU.so.1.3.1
/usr/lib64/nvidia/libGLESv1_CM.so.1
/usr/lib64/nvidia/libGLESv1_CM.so.346.72
/usr/lib64/nvidia/libGLESv2.so.1
/usr/lib64/nvidia/libGLESv2.so.2
/usr/lib64/nvidia/libGLESv2.so.346.72
/usr/lib64/nvidia/libGL.so.1
/usr/lib64/nvidia/libGL.so.346.72
/usr/lib/insync/libGL.so.1
/usr/lib/libGL.so.1
/usr/lib/libGL.so.1.2.0
/usr/lib/libGLU.so.1
/usr/lib/libGLU.so.1.3.1
/usr/NX/scripts/vgl/libGL.so
/usr/share/doc/libGLEW
/usr/share/doc/mesa-libGL
/usr/share/doc/mesa-libGLES

Edit that I made in /usr/NX/etc/node.cfg:
#
# Specify path to libraries to be added to the command for starting
# a new virtual desktop.
#
DesktopCommandLibraryPath “/usr/lib64/”

I have also tried in /usr/NX/etc/node.cfg:
DesktopCommandLibraryPath “/usr/lib64/nvidia/”

NoMachine version installed on server:
$ rpm -qa nomachine
nomachine-4.6.4-13.x86_64

NoMachine version installed on clients: v4.6.4

Server is using the NVIDIA akmod drivers from rpmfusion:
$ rpm -qa | grep nvidia | sort
akmod-nvidia-346.72-2.fc22.1.x86_64
kmod-nvidia-4.0.5-300.fc22.x86_64-346.72-2.fc22.1.x86_64
kmod-nvidia-4.0.6-300.fc22.x86_64-346.72-2.fc22.1.x86_64
kmod-nvidia-4.0.7-300.fc22.x86_64-346.72-2.fc22.1.x86_64
xorg-x11-drv-nvidia-346.72-1.fc22.x86_64
xorg-x11-drv-nvidia-kmodsrc-346.72-1.fc22.x86_64
xorg-x11-drv-nvidia-libs-346.72-1.fc22.x86_64

Content of server’s /usr/NX/var/log/nxerror.log:
Info: Handler started with pid 11563 on Tue Jul 14 08:26:38 2015.
Info: Handling connection from 10.1.1.12 port 63422 on Tue Jul 14 08:26:38 2015.
Info: Connection from 10.1.1.12 port 63422 closed on Tue Jul 14 08:26:43 2015.
Info: Handler with pid 11563 terminated on Tue Jul 14 08:26:43 2015.

Content of server’s /usr/NX/var/log/nxserver.log:
2015-07-14 08:26:22 704.733 10678 NXSERVER Starting NoMachine server 4.6.4 and services.
2015-07-14 08:26:38 783.827 11563 NXSERVER User ‘tyrluk’ logged in from ‘10.1.1.12’.
2015-07-14 08:28:11 863.221 11563 NXSERVER User ‘tyrluk’ from ‘10.1.1.12’ logged out.

[graywolf]

tyrluk, I saw similar issue to yours but it doesn’t seem related to the existence of NVIDIA driver in the system.

I think we found an incompatibility with GDM provided with Fedora 22. I’m trying to find a suitable workaround and possibly a solution. I’m going to try different display managers to find if we get the same issue with them.

[graywolf]

Switching display manager to LightDM is an effective workaround.

Install lightdm:
[root@test01 ~]# dnf install lightdm

Disable GDM:
[root@test01 ~]# systemctl disable gdm

Enable LightDM:
[root@test01 ~]# systemctl enable lightdm

Reboot to make changes effective:
[root@test01 ~]# reboot

[Author]

Posts