This system is reliable in Alphs3 but in Beta1 it gets to probing hardware fades it hangs dead, no keyboard lights. This makes the entire beta impossible to use so I've marked it as high/high.
The trigger for these hangs turns out to be kudzu. If kudzu runs the machine crashes a little while later without kudzu its fine
Does this box have firewire?
No its pre firewire era hardware: AMD751/756 chipset (DVD-ROM drive on the onboard IDE) PS/2 keyboard PS/2 mouse (2 button - old microsoft) USB (unused) 3 quad head S3 video cards (unused) Nvidia Riva 128 AGP video (dual head, 1 used) ISA SCSI card (unused) 3com 3c590 (used) CMD680 PCI IDE controller (with disks on) Alpha3 is fine, 9.0 is fine.
Does it persist when running kudzu post-boot? If so, what does a strace (or similar) say when it's failing?
Running kudzu makes the box crash a bit later, it doesnt crash while kudzu is running just later on
If you're in text mode, do you get any kernel output when it crashes?
Nope it just stops
Alan, you were experiencing a total system lockup? In Bug 99528 I experienced only a frozen boot process while I am still able to switch VT's. After killing rhgb's X in VT7 with CTRL-ALT-Backspace then hitting ALT-F7, I see a corrupted looking kudzu that seems frozen. I suspect that Bug 99528 and Bug 91697 are duplicates of this bug, although only 99528 is public. Can we unhide this one? Bug 100894 is related, perhaps another duplicate.
Bug 100925 also related. Perhaps one bug should be an umbrella bug for "kudzu GUI" so we can clearly eliminate duplicates?
It was a total lockup yes
Bug 102647 also related.
Assigning to kernel, nothing kudzu does should kill the box.
Created attachment 95382 [details] screen grab of strace'd kudzu startup I see this too occasionally, but only on my Vaio. Can't reproduce it at all on any of my other systems. Attached is a screengrab of an strace during startup. It always hangs in the same place.
Really quite weird, but FC2test1 no longer triggers this problem so whatever caused it has gone away somewhere.