NVIDIA-Redhat 7.5
Hi Guys,
Need urgent help!
I tried installing Nvidia 390.87 in my HP-elite(intel core i7,Geforce GTX 1080). After installing the drivers I am getting blank screen. I also contacted Nvidia support, they hinted me to use init3 before installation and init5 after installation of drivers.
So init5 is giving me this now:
PolicyKit daemon disconnected from the bus.
We are no longer a registered authentication agent.
Can you please help me with this. I am a newbie with Red-hat
Thanks
Prem Rao
Responses
Hi Prem Rao,
Please examine this discussion https://access.redhat.com/discussions/3221971, and also this solution at https://access.redhat.com/solutions/1155663 and also this https://access.redhat.com/discussions/3101441 and also this https://access.redhat.com/solutions/1155663.
Let us know how this goes,
Regards,
RJ
Hi Prem ! :)
Did you install the drivers by using the .run file from NVIDIA ? Did your notebook ship with NVIDIA Optimus support ?
In case both is true and assuming that you have UEFI BIOS, please make sure that you performed the following steps :
Edit boot parameters :
sudo vi /etc/default/grub
Add these parameters :
GRUB_CMDLINE_LINUX="rd.driver.blacklist=nouveau modprobe.blacklist=nouveau nvidia-drm.modeset=1"
Update GRUB config :
sudo grub2-mkconfig -o /boot/efi/EFI/redhat/grub.cfg
Blacklist nouveau :
sudo vi /etc/modprobe.d/nouveau.conf
Add these lines :
blacklist nouveau
options nouveau modeset=0
Edit xorg.conf :
sudo vi /etc/X11/xorg.conf
Add these lines :
Section "ServerLayout"
Identifier "layout"
Screen 0 "nvidia"
Inactive "intel"
EndSection
Section "Device"
Identifier "nvidia"
Driver "nvidia"
BusID "PCI:1:0:0"
EndSection
Section "Screen"
Identifier "nvidia"
Device "nvidia"
Option "AllowEmptyInitialConfiguration"
EndSection
Section "Device"
Identifier "intel"
Driver "modesetting"
BusID "PCI:0:2:0"
EndSection
Section "Screen"
Identifier "intel"
Device "intel"
EndSection
Check for your BusID's by executing the following command : sudo lspci -k | grep -EA3 '3D|Display|VGA'
Regards,
Christian
What I forgot to mention, Prem - two days ago NVIDIA released the latest stable 410.66 drivers from the long-lived branch.
I recommend to uninstall the 390.87 drivers and to install these 410.66 drivers. Please read and follow the instructions ... :)
Regards,
Christian
Hi Prem,
Please execute sudo systemctl set-default graphical.target and, reboot the system ... you should boot into GUI. If that also doesn't let you access the GUI, contact NVIDIA support again and provide them with the exact error message and all details about what you've done so far. :)
Regards,
Christian
Hi Prem,
One additional setup adjustment you can try - not sure if it solves the problem though - in some cases it does, in others not.
Create two new .desktop files which will configure the video output source by using the xrandr tool on system startup :
sudo vi /etc/xdg/autostart/nvidia-optimus.desktop
sudo vi /usr/share/gdm/greeter/autostart/nvidia-optimus.desktop
Add the following text into both empty files (valid for system installations using the default GNOME desktop environment) :
[Desktop Entry]
Type=Application
Name=NVIDIA Optimus
Exec=sh -c "xrandr --setprovideroutputsource modesetting NVIDIA-0; xrandr --auto"
NoDisplay=true
X-GNOME-Autostart-Phase=DisplayServer
Once again ... good luck ! :)
Regards,
Christian
Hi Prem,
I assume CUDA 9.1 doesn't play well with the 390 drivers series. Please check if it works with
CUDA 9.2 ... or install the 410 drivers together with CUDA 10 - which is what I recommend. :)
Regards,
Christian
Well Prem, as you said that it worked with the 390 drivers installed before you installed CUDA,
I suggest that you ask what to do to get it also working with CUDA in the NVIDIA Linux Forum.
Chances are much higher to get a solution for that specific CUDA 9 related problem there ... :)
Regards,
Christian
Hi Prem,
Although I said that I'm running out of ideas - here is another one which solved the problem for another user I've supported. :)
Please add nvidia-drm.modeset=1 to the system boot parameters (and update the GRUB configuration), so that it reads : GRUB_CMDLINE_LINUX="rd.driver.blacklist=nouveau modprobe.blacklist=nouveau nvidia-drm.modeset=1"
sudo vi /etc/default/grub
sudo grub2-mkconfig -o /boot/efi/EFI/redhat/grub.cfg
Regards,
Christian
Hi Prem,
Maybe the preconfigured drivers from negativo17 or RPM Fusion work ... before installing them please remove
everything from NVIDIA and all configurations you made before, revert back to GDM and restart the system. :)
Regards,
Christian
Hi Prem,
Just for your information : NVIDIA released the new long-lived branch drivers 410.73 yesterday. Wanna give them a try ?
Although I don't exactly expect improvements in your case - maybe there's a slight chance that something was fixed ... :)
Regards,
Christian
Hi Prem,
Thanks for the feedback - but : "And In my case NVIDIA drivers have never been a issue." Now I am a bit confused - in your initial
post you said : "After installing the drivers I am getting blank screen." That's why I've tried to help you to get the NVIDIA drivers
working properly in the first place. CUDA problems you mentioned at a later time. Hmmm ... and I already suggested to ask the
NVIDIA experts how to get CUDA working, because chances are much higher to get a correctly working solution from them. :)
Regards,
Christian
I'm not so sure, Prem ... a whole bunch of Red Hat customers are using NVIDIA drivers and CUDA. RHEL is one of the most used systems in enterprise environments and if there would be a general problem with using NVIDIA and/or CUDA on that system - this would lead to a serious marketing problem for NVIDIA in the end.
I think you just couldn't reach the "best technician", which of course is not your fault.
First level support employees in most cases are not trained to solve deeper problems.
Please ask in the NVIDIA CUDA Zone Linux Graphics Forum for specific assistance. :)
Regards,
Christian
Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.
