XServe G5 has no hardware deffect, so this *IS* a udev bug :/
tags 403136 + d-i
tags 403136 + needhelp
After speaking some with various folks, and someone else testing the same d-i
which failed here on other XServe's, altough maybe not from the same
generation (mine is from september 2005 i was told), i became suspicious, and
brang the machine to an apple technical center, for defect testing.
The machine came back today, and it was working perfectly, passed all apple
hardware diagnostic tests, and ran full tests of mac-os-x during various days
Furthermore, YDL 4.1 installs fine, and also has no problems whatsoever with
the (somewhat older) udev installation there, and since this is an issue which
surfaced around november rather suddenly (it was in a datacenter a couple of
weeks, then upgraded, and at the first reboot, everything broke), i suspect it
is indeed some strange udev issue.
Let's again summarize the situation :
1) udev does not create the /dev/sd* device nodes, either in initramfs-tools
or in d-i. Probably other nodes are affected.
2) the udev d-i scsi-devfs.sh scripts, which is in charge of creating that
node, dies when writing to stdout, which is piped to udevd.
3) ubuntu (of late november) exhibited the same problem.
4) yaird did not work, but for some other reason (not recognizing a given
node, didn't investigate more).
This makes the box fully unusable and unsuported both in d-i and in normal
debian, thus the RC status, furthermore something very strange is going on
Next step would be :
1) write a program writing to stdout and dropping the actual error message
2) contact udev author and ask for his help, since Marco said he didn't have
a further clue, and this may be an upstream problem.
3) fix yaird to work on it, and see if the machine is stable with yaird and
More to this once i have the box back, and am back from the Solution Linux
show in paris.