After the update to Kernel 6.8.12-11 something very odd happened. I've spent weeks trying to resolve this.
Now, at boot, it gets part way through then just resets suddenly and without warning.
At first I thought BIOS watchdogs, memory or thermal issues, though why these should suddenly appear after an update, I don't know.
I did the usual, BIOS config reset, check cooling systems etc. No difference.
Then I explored with GRUB 'emergency', then recovery (aka '1') boot modes.
I found I could boot into emergency mode and mount disks. I left it several days with no reset.
I then found if I boot into recovery mode it does the same reset.
But, if I mask systemd networking in emergency mode then boot into recovery mode, then all is fine, no reset (and no networking).
But then I can manually configure network interfaces and routing and all is ok.
Now, I unmask networking and try to start it, the box resets again immediately.
On resetting there is nothing in the logs, no errors, no warnings.
So, something in the networking systemd tree is causing a reset, but whatever is doing it is catastrophic.
I've tried turning off NMIs & watchdogs, no luck.
I think this is nothing to do with the actual kernel, as booting into older kernels does the same. Something else updated during that kernel update is causing this.
Is there a way to 'single step' the processes started with the command "systemctl start networking" or perform some sort of bifurcation on them.
systemd networking is bloated, so finding which one of the resulting drivers or services seems almost impossible, especially as there is no log to review and the reset is sudden.
Once networking is unmasked, the only way to get back control is via emergency mode, then recovery mode.
Now, at boot, it gets part way through then just resets suddenly and without warning.
At first I thought BIOS watchdogs, memory or thermal issues, though why these should suddenly appear after an update, I don't know.
I did the usual, BIOS config reset, check cooling systems etc. No difference.
Then I explored with GRUB 'emergency', then recovery (aka '1') boot modes.
I found I could boot into emergency mode and mount disks. I left it several days with no reset.
I then found if I boot into recovery mode it does the same reset.
But, if I mask systemd networking in emergency mode then boot into recovery mode, then all is fine, no reset (and no networking).
But then I can manually configure network interfaces and routing and all is ok.
Now, I unmask networking and try to start it, the box resets again immediately.
On resetting there is nothing in the logs, no errors, no warnings.
So, something in the networking systemd tree is causing a reset, but whatever is doing it is catastrophic.
I've tried turning off NMIs & watchdogs, no luck.
I think this is nothing to do with the actual kernel, as booting into older kernels does the same. Something else updated during that kernel update is causing this.
Is there a way to 'single step' the processes started with the command "systemctl start networking" or perform some sort of bifurcation on them.
systemd networking is bloated, so finding which one of the resulting drivers or services seems almost impossible, especially as there is no log to review and the reset is sudden.
Once networking is unmasked, the only way to get back control is via emergency mode, then recovery mode.