Can't update nvidia driver in CentOS 6.8

We run Resolve Studio on Linux, CentOS 6.8, the "Legacy DVD Install" with our advanced dongle. Trying to upgrade to 14.3 but I'm having an issue with the nvidia drivers. I sent this to BMD support a minute ago but I'm hoping someone here might have an answer:


We're trying to upgrade our Linux Resolve Studio system to 14.3 (CentOS 6.8), but we're having problems with the nvidia drivers, following the instructions in the Resolve readme.

Specifically, I'm doing the following:

Tried switching the runlevel to 3 from a terminal. When I do this, our screen goes black rather than dropping to text mode, (likely because of our Raritan KVM system). I'm able to ssh into the machine as root from my mac to complete the process however.

I follow the instructions in the Resolve 14.3 Readme:

Code: Select all

---------------------------------------------
[root@gui Downloads]# cd /home/resolve/Downloads/
[root@gui Downloads]# sh NVIDIA-Linux-x86_64-384.69.run --silent --no-network
Verifying archive integrity... OK
Uncompressing NVIDIA Accelerated Graphics Driver for Linux-x86_64 384.69..........................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................

WARNING: The '--no-network' option is deprecated:  nvidia-installer will ignore
        this option.


ERROR: You appear to be running an X server; please exit X before installing.
      For further details, please see the section INSTALLING THE NVIDIA DRIVER
      in the README available on the Linux driver download page at
      www.nvidia.com.


ERROR: Installation has failed.  Please see the file
      '/var/log/nvidia-installer.log' for details.  You may find suggestions on
      fixing installation problems in the README available on the Linux driver
      download page at www.nvidia.com.
---------------------------------------------


So to confirm I'm not running X at all, I modified /etc/inittab so that it reads:

Code: Select all

id:3:initdefault:


...and rebooted the computer, and logged back in via ssh (because again, the Raritan won't display the default resolution in text mode). Tried to run the driver install again, but again got the same error, telling me that X is running. The nvidia log file referenced in the error says this:

Code: Select all

---------------------------------------------
nvidia-installer log file '/var/log/nvidia-installer.log'
creation time: Mon Feb 19 11:41:13 2018
installer version: 384.69

PATH: /usr/lib64/qt-3.3/bin:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin:/opt/ibutils/bin:/root/bin

nvidia-installer command line:
   ./nvidia-installer
   --silent
   --no-network

Using built-in stream user interface
-> Detected 56 CPUs online; setting concurrency level to 32.
-> The file '/tmp/.X2-lock' exists and appears to contain the process ID '7688' of a runnning X server.
ERROR: You appear to be running an X server; please exit X before installing.  For further details, please see the section INSTALLING THE NVIDIA DRIVER in the README available on the Linux driver download page at www.nvidia.com.
ERROR: Installation has failed.  Please see the file '/var/log/nvidia-installer.log' for details.  You may find suggestions on fixing installation problems in the README available on the Linux driver download page at www.nvidia.com.
---------------------------------------------


I guess my question is - do I have to run this version of the driver even? wouldn't it be easier to update the driver via yum? Or is there something about the latest version that is incompatible with Resolve? The one recommended in the readme is from last year.

Or is there another way to exit X? I was under the impression that switching to runlevel 3 would do it.

Thanks,

-perry

Checking on the process that has the lock, I see that it's this:

Code: Select all

resolve   7688     1  0 11:28 ?        00:00:00 /usr/bin/Xvnc :2 -desktop gui:2 (resolve) -auth /home/resolve/.Xauthority -geometry 1680x1050 -rfbwait 30000 -rfbauth /home/resolve/.vnc/passwd -rfbport 5902 -fp catalogue:/etc/X11/fontpath.d -pn


Does that help any?

I've tried killall X, but it said there were no processes to end. So I'm stumped here.

[Dwaine Maggart]

You have to install the NVIDIA driver from run level 3 in text mode.

I don't know that an ssh login will work. I'm not SURE it won't, but you really need to do this from the normal KVM.

When you do the init 3 and it drops to text mode, try ALT-F2, ALT-F3, ALT-F4, ALT-F5, etc. You should find a text login prompt on most of those. Maybe you'll have to bypass your Raritan KVM if you don't.

Hi Dwaine. I'll give it a try but it means bringing an keyboard and monitor into the server room and connecting directly. Our IP based kvm is fussy about monitor mode changes so this kind of thing can't be done through that. Switching to run level 3 gets us a blank screen.

I'm kind of surprised ssh to that machine doesn't work for this though. I wouldn't think X would be required of that.


Sent from my iPhone using Tapatalk Pro

[Dwaine Maggart]

Again, I'm not sure ssh won't work, but I've run into a few other things that won't work over ssh, like this command to enable VNC to work:

gsettings set org.gnome.Vino require-encryption false

I tried for way too long to get VNC working with this, until some Googling pointed out that you have to enter this locally, and it won't work over ssh.

Based on your described results, it seems something similar may be happening to you.