Just a log of one user's experience trying to follow the instructions

I found it helpful to ssh to diydice and grep through various existing profiles

That's what made me realise I needed to change all the occurrences of 'cc' in the recommended starting point to 'ht' (my own username).

I'm guessing that the inv.sno is what's given on the label of my machine as "Barcode"

No idea how to find the allowed formatting for the inv.location field

If you have a high-res display, add the following line:

!xfree.displaymodes_main mSET("1600x1200" "1280x1024" "1024x768" "800x600")

Once I got my profile to compile (it's at /diyhome/ht/lcfg/source/ignatius if you're interested), the next hitch was that the boot options menu was not as described in step 7 of the instructions, rather it gave me five options, fc3 and fc5, and defaulted to fc5 while I was still reading!

That was OK, since I wanted FC5

Next pblm: couldn't find my profile. Some detective work established that the correct answer is

http://diydice.inf.ed.ac.uk/profiles/diy.inf.ed.ac.uk/ignatius/XML/profile.xml

obviously your machine name goes where 'ignatius' is above.

Next pblm:

LCFG fstab: /root/var/cache/afs already exists and populated
LCFG install: install method failed

and I got dropped into a shell. After looking around a bit, I simply exitted, which rebooted, and this time I got in quickly enough to type 'fc5' at the boot menu prompt.

Note always have to F12 on reboot during all this, to get to option 2 -- default/option 1 leads to Grub failure, whatever that is

That worked better -- looked like the installation rebuilt the disks from scratch, and started the mega-RPM fetch

2821/2821 rpms updated/installed, but "LCFG updaterpms: rpmtsRun failed" and out to shell

Commented out the #define FIRST_INSTALL in hopes that would keep the disks from being scrubbed

That didn't work :-(, disks rebuilt, rpms installing. . .

OK, this time I saw some of the errors, e.g.

%pre(dice-submit-2.65.5-1.i386) scriptlet failed, exit status 127
(code 7210499)
failed to open logfile : /var/lcfg/log/upd[*ERR*]

Now hung on rpm number 2211, after 1 hour 10 minutes, with pidof[19901]: stat failed for /proc/19901/root! displayed higher up the screen. . . This actually happened once before, eventually gave up that time and rebooted, which of course then took another 90 minutes . . .

OK, but this time, after about a 20 minute hang, it picked up again and carried on -- go figure.

10 minutes later, now up to 2478, a very small 'beep', maybe, and it's hung again -- no, wait, after 5 minutes it's off again, slowly. . .

installrpms error as above

Tried the hack below to keep rpms from being trashed, that worked, got to login OK!

Outstanding questions

• When is it right to comment out the #define FIRST_INSTALL?
  • After the first complete successful install. . .
• How do I prevent the disks from getting scrubbed, if things fail part-way through the rpms, or even at the end?
  • Not recommended, but try

!fstab.preserve_hda1            mSET(yes)
!install.imethod_mkdir_dev mSET("%oneshot% mkdir -p /root/dev") 
!install.imethod_mkdir_etc mSET("%oneshot% mkdir -p /root/etc")

-- Main.ht - 01 Feb 2007