Nvidia driver installer problem [Solved]

[roundup976]

Here are my steps, I will post them anyway.

• 1. run the command: sudo ddm-mx -p nvidia
  2. open MX Package Manager, go to stable repos, sort on nvidia, uninstall all nvidia products
  • a. my main issue was ddm-mx did not remove the nvidia 510 version of nvidia-detect
  3. reboot, verify nouveau is being used in Quick system info
  4. open MX Package Manager
  • a. go to stable repos
    b. filter on nvidia
    c. check boxes to install on :
  • • 1. xserver-xorg-video-nvidia-tesla-470
      2. nvidia-tesla-470-driver
      3. nvidia-settings-tesla-470
  • d. at the bottom check: also install "Recommended" Packages
    e. click install
  5. You should get a pop-up confirming the install, click more info and verify that many more nvidia packages are getting installed
  • a. accept the and install the packages
    b. a pop-up should appear at some point warning about a conflict with nouveau and that a restart is required, accept this and continue
  6. reboot after packages installed
  7. run Quick System Info and verify the nvidia driver is loaded or open nvidia settings
  
  8. additionally, you could disable ahs as dolphin_oracle suggested. I did this in /etc/apt/sources.list.d and opened mx.list and commented out the ahs repo and then re-ran sudo apt update.
  Originally I did this first, but then the nvidia 470 drivers are not available.

I actually un-installed everything and put nouveau back on and then did this again just to verify it and it works.

[Huckleberry Finn]

... You can also:

Code: Select all

sudo apt purge nvidia*

For a "clean job" (for step 1 & 2 ) :)

[CharlesV]

@roundup976 Thank you for that detail!!

[FlyBy]

Here are my steps, I will post them anyway.

• 1. run the command: sudo ddm-mx -p nvidia
  2. open MX Package Manager, go to stable repos, sort on nvidia, uninstall all nvidia products
  • a. my main issue was ddm-mx did not remove the nvidia 510 version of nvidia-detect
  3. reboot, verify nouveau is being used in Quick system info
  4. open MX Package Manager
  • a. go to stable repos
    b. filter on nvidia
    c. check boxes to install on :
  • • 1. xserver-xorg-video-nvidia-tesla-470
      2. nvidia-tesla-470-driver
      3. nvidia-settings-tesla-470
  • d. at the bottom check: also install "Recommended" Packages
    e. click install
  5. You should get a pop-up confirming the install, click more info and verify that many more nvidia packages are getting installed
  • a. accept the and install the packages
    b. a pop-up should appear at some point warning about a conflict with nouveau and that a restart is required, accept this and continue
  6. reboot after packages installed
  7. run Quick System Info and verify the nvidia driver is loaded or open nvidia settings
  
  8. additionally, you could disable ahs as dolphin_oracle suggested. I did this in /etc/apt/sources.list.d and opened mx.list and commented out the ahs repo and then re-ran sudo apt update.
  Originally I did this first, but then the nvidia 470 drivers are not available.

I actually un-installed everything and put nouveau back on and then did this again just to verify it and it works.

This is fantastically helpful. I followed it to the letter and all is well. The install did terminate with an error (see attached log), however following the reboot everything's working. A shame a beer isn't possible here, thank you.

[roundup976]

@FlyBy I went back and checked and I am having an issue with nvidia-persistenced as well. I did some reading, and honestly I am not even sure it is truly needed. I am no graphics expert, but from the guide on nvidia I found this:

nvidia-persistenced is intended to be run as a daemon from system initialization, and is generally designed as a tool for compute-only platforms where the NVIDIA device is not used to display a graphical user interface. As such, depending on how your package is typically used, it may not be necessary to install the daemon to run on system initialization.

from this site: https://download.nvidia.com/XFree86/Lin ... enced.html

I did more reading and just like always, nvidia is a pain to install and remove. MX actually makes it the easiest I have ever had. In my day job I work with Oracle Linux running RTOS applications, and getting nvidia running on them is a pain.

anyway, it looked like this was not installed, but really I think it got partially installed. I also notice the latest version in the stable repo is 418.56 which is technically a mismatch with tesla-470.
try this, it looks to have resolved my issue:

Code: Select all

sudo apt remove nvidia-persistenced
...stuff removed

sudo apt install nvidia-persistenced
...stuff installed

note: if you disabled the ahs repo you will need to enable and run sudo apt update first.

If that actually removes something then it was probably partially installed.
reboot.

then run this and see if the daemon is running.

Code: Select all

ps aux | grep nvidia-persistenced

on my system the output shows:

Code: Select all

nvpd      2161  0.0  0.0   4700  1656 ?        Ss   12:33   0:00 /usr/bin/nvidia-persistenced --user nvpd

I have not seen an error since, but I will keep looking now that you pointed it out. I also was running without it for a while, and I did not notice any difference on my system.
I did not test this, but @Huckleberry Finn pointed out sudo apt purge nvidia* could replace the steps I listed in 1 and 2, which may have avoided this problem altogether.

[Huckleberry Finn]

It looks like "optional" :

Code: Select all

$ apt show nvidia-persistenced
...
Priority: optional
...
Pre-Depends: init-system-helpers (>= 1.54~)
Depends: libnvidia-legacy-390xx-cfg1 | libnvidia-legacy-340xx-cfg1, adduser, libc6 (>= 2.7)
...
Description: daemon to maintain persistent software state in the NVIDIA driver.
When persistence mode is enabled, the daemon prevents the driver from releasing 
device state when the device is not in use. This can improve the startup time of 
new clients in this scenario.

[FlyBy]

@roundup976 Have you had any issues where the nvidia gpu isn't being used by an application as had been expected?

[roundup976]

@roundup976 Have you had any issues where the nvidia gpu isn't being used by an application as had been expected?

No I have not. I would also think on my system it is not possible. I am using a desktop with a PCIe nvidia card. Since I am physically plugged directly into the card it has to be used. I have also not seen any issues of the card not loading the correct driver. To be honest though, the only time I can tell a difference between the nvidia driver and open source driver is playing steam games.

Remember back in the day the static on color TV's? Sometimes after I turn back on from a hibernate or suspend, windows that were left open will occasionally look like that. I had that issue before this whole driver incident though.

On a different system, my work laptop runs gnome using Oracle Linux 8 and has an Nvidia card. I loaded the nvidia driver on it and thought everything was good. Turned out it was fine in the xserver, but as soon as I would try switching to a login shell the colored static window was the only thing that showed up there. It made the login shells unusable, so I reverted back to nouveau on that system.

In the end I really get annoyed with the Nvidia drivers in general. If I don't need them I don't load them. So far gaming is the only place I have had a need. I do not edit videos either, maybe someone who does has a take on that side.

[FlyBy]

I have decided to continue using MX Linux without nvidia drivers installed. Although work-arounds suggested initially appeared to fix the issue, subsequently those work-arounds created problems of their own. I do appreciate those that took time to look at this, thank you.