My boot process locks up at the time it hits the nvidia module:
It was booting just fine earlier, only there was an upgrade today which I ran from X-Windows and later when I rebooted that's what I've got. I though I rebooted once in between though. I tested with various resolutions and two monitors, but before the reboot I'm on the same monitor I started with and same resolution. I really don't see why it would block unless the upgrade did break something.
What would you do from that point? Attempt a repair?
Note that the DRM says "VGA only" although the output is HDMI.
Updates:
- Does the previous kernel still boot (at grub, select advanced... and choose a previous kernel)?
Testing now... but somehow I could not boot from my USB Disk with the basic Ubuntu 18.04 installed nor the Recovery version.
Old kernel is 4.15.0-20-generic. Current is 4.15.0-64-generic.
In this case it gets stuck somewhere else, possibly because nVidia is not part of that kernel? Last it says:
i40e fw 3.1.55727 api 1.5 nvm 3.31 0x80000c92 1.1747.0
Before that it handled the USBs and ahci.
- How did you install the nvidia driver (where did you get the driver too)?
Automatically with a apt-get command.
sudo ubuntu-drivers autoinstall
sudo reboot
Source: How to install Nvidia driver in ubuntu 18.04?
- What Nvidia hardware and which driver were you using?
I have a standard RTX 2060 (not Super).
The driver was the latest, something around 340 430.26.
Old Kernel Recovery Mode Started!
So I tried to also start in Recovery Mode using the old kernel and that worked. So I have a recovery prompt now. I'll verify that the newer kernel is up to date (all installation ran as expected) and it looks like one of my fstab entries may cause problems too (an LVM creating an RAID device which is not yet working right).
Boot Fixed?
I got it to boot again. Here are the three things that changed before my failed reboot:
- Installed the upgrades, that included some nVidia stuff, but when I went to repair mode, I could not see anything broken there.
- I installed
lm-sensors, however, runningsensorsin my command line worked just fine so I don't think that has had any negative effects - I tried to have a new hard drive using LVM auto-mounted and even when going into recovery, it took forever for that recovery to fail and let me in, but at least I could see the info on my screen
Ah! About point 3, I paid attention on my last reboot and the screenshot I have up there is where it stops. That's more or less the point where nVidia takes over and switches to graphics mode. However, as we can see the video card output remains frozen to whatever was there at the time of the unfinished switch. So there was probably more output, it was just not visible. Were I able to see that, maybe I'd see the attempt to mount the LVM which takes minutes before continuing.
I know that with older versions, the nVidia driver tends to time out quickly. If your system takes too long to boot, then you are in a weird mode where the graphical target is only half way initialized and there is nothing you can do (in my case, somehow, the keyboard would die; but otherwise you may be able to use Ctrl-Alt-F12 to get out of that mode and back to the text console. From there you can then look for what's taking so long that X doesn't get started in time for the nVidia driver to be happy).
So I think I'll change my default target to console and start X-Windows manually. Over time, it will probably help me not waste hours trying to debug something which isn't a problem in itself (i.e. since it probably would have booted as expected had I waited long enough; although I thought I did wait at least as long as it took to boot in recovery mode...)
