Hiig's Linux Troubles

[hiigaran]

(Note: I am frequently editing this post to reflect the current situation)

Been hijacking the other thread, so I figured I'd create a dedicated thread for this issue that's had me pulling my hair out, crying in the corner, and considering medication for hypertension! Let's get right to it!

As a note, considering how much work I've put into this in the last four days and how many different things I've tried (some of which I've lost track of), I'm going to start this post initially with what is happening on my current attempt.

For those of you just tuning in, the goal is to set up a headless F@H system. Before this can be done, I need to get fan controls to work, and that's where my problem has lied thus far. At this point, I should also mention that this system is getting an upgrade, replacing an old HDD with a M.2. The software installed on the HDD works as intended. I have tried installing the exact same operating system and drivers to replicate the drive's contents, but clearly there is something I'm missing.

Specs:

AsRock X99 WS-E
Xeon E5-2609
4x GTX 1080
1600w PSU
nVidia 390.25
Ubuntu Server 17.10
Kernel 4.13.0-32-generic

After a clean install, I have performed the following actions:

sudo apt update
sudo apt full-upgrade
sudo apt install build-essential
sudo nano /etc/modprobe.d/blacklist.conf
    blacklist vga16fb
    blacklist nouveau
    blacklist rivafb
    blacklist nvidiafb
    blacklist rivatv
sudo update-initramfs -u
sudo nano /etc/default/grub
    text
sudo update-grub
sudo wget http://us.download.nvidia.com/XFree86/Linux-x86_64/390.25/NVIDIA-Linux-x86_64-390.25.run
sudo chmod +x NVIDIA-Linux-x86_64-390.25.run
sudo reboot
sudo dpkg --add-architecture i386
sudo apt update
sudo apt install libc6:i386
sudo apt install libstdc++6:i386
sudo ./NVIDIA-Linux-x86_64-390.25.run
    Install 32 bit compatibility? (YES)
    nvidia-xconfig to automatically update X configuration file? (YES)
sudo reboot
sudo apt install libgtk-3-0
sudo apt install xinit

I'm using a .run file directly from nvidia to avoid having a display manager automatically installed. Installing nvidia-current or a specific version number from the graphics-drivers PPA will result in a display manager being installed, and subsequent boots landing on a GUI login screen. When this happens, the system freezes a few seconds afterwards. In these instances, I could still SSH into the system, but could not switch to tty locally.

Where display managers were automatically installed, nvidia-smi and dmesg showed these errors respectively:

Unable to determine the device handle for GPU 000:04:00.0: GPU is lost. Reboot the system to recover this GPU.

NVRM: Xid (PCI:0000:04:00): 79, GPU has fallen off the bus.

Again, since this system has worked with several months of uptime on the old HDD under full load, there is no hardware issue

In all cases, nvidia-settings gives me the following error:

Unable to init server: Could not connect: Connection refused
ERROR The control display is undefined; please run 'nvidia-settings --help' for usage information

At this point in my progress, this is the apt list --installed output of all installed packages.

 

[biodoc]

I opened up the synaptic package manager and searched for libgtk.

On my mint 17.3 system I have the following primary libgtk3 and libgtk2 packages installed:

libgtk-3-0
libgtk-3-common
libgtk-3-bin
libgtk2.0-0
libgtk2.0-common
libgtk2.0-bin

 

[hiigaran]

Okay, installed. Now I'm back to my original problem with nvidia-settings, which is connection refused and undefined display messages as shown in the first post.

EDIT: First post edited. Will continue to edit first post to show what is currently happening

EDIT2: The following update will not be reflected in the first post yet. If the changes are a step in the right direction, I will re-edit this post and the first with relevant information.

sudo apt install libgtk-3-0
sudo apt install xinit
sudo apt install lightdm
sudo systemctl disable display-manager.service
sudo apt install xorg
sudo apt install xorg-dev
sudo ./NVIDIA-Linux-x86_64-390.25.run

libgtk was installed to eliminate missing lib errors in nvidia-settings. xinit was installed to be able to use startx. lightdm was installed because i dont know. display-manager.service was disabled to prevent the GUI from appearing, as the 'text' parameter via GRUB does not work.

Now for what might be progress. Before installing xorg, xorg-dev and reinstalling the drivers, I tried using startx locally. Screen flickered off, then the terminal looked different. Smaller font or something. Either way, it was frozen. Inspecting the Xorg log, there was a mention about it failing to initialise the GLX module.

Doing some research, someone suggested that I would need to install the Xorg SDK and then reinstall the drivers so that the module can be built. Not sure what the SDK was, so I installed xorg and xorg-dev. The errors seem to have disappeared from the latest log.

Pastebin of current Xorg log.

Now, when I perform startx locally, I get the initial welcome messages about X version, config file locations used, etc. A few seconds later, the screen flickers, returns to the same unaltered terminal, and another line says:

Xinit: Connection to server lost
Waiting for X server to shut down. (II) Server terminated successfully (0). Closing log file.

Nothing seems to stand out in the linked log, so I don't know why startx stops shortly after. Assuming no error stopping it from functioning, the next step seems to be to try and find a way to actually keep it running.

 

[biodoc]

Would you mind posting the contents of your /etc/X11/xorg.conf file?

 

[biodoc]

This mint Tech Note for "Xserver fails to start" may be of use.

 

[hiigaran]

Will take a look at it later. I'm out of the country now for a week on work.