[olug] Be wary of preupgrade from Fedora13 to Fedora15 - systemd troubles

[Rob Townley]

Since others are probably contemplating Fedora15...

First, i did not do preupgrade like i normally do.
  Edit /boot/grub/grub.conf first to increase the timeout to 30 seconds.
  yum clean all
  yum update yum preupgrade rpm something else
  backup /home/

i have preupgraded about ten different machines since Fedora 9 (maybe
8) numerous times and not once has it failed out of about 40x till
now.
Had done preupgrade from Fedora 10 to 11 to 13 successfully on this
machine.  So i skipped 12 and decided to skip 14.  According to some
systemd comments, skipping F14 was probably the fatal mistake, so i
really wish preupgrade did not allow me to do so just like it has not
allowed some version jumps in the past.  Probably need to file a bug
report.

(Our home cox connection is slower than my 2.5G cell connection, but
unusable till i get new phone and did not want it to take forever.
Not very competitive of cox.net to raise rates to $25.99/month for
1Mbps down / 256kbs up, but that is an entirely other subject.)

The preupgrade from F13 to F15 seemed to be going along fine.  The new
2GB /boot/ partition gave it plenty of space to store the upgrade
image, so no warning about needing a ethernet cable.  It rebooted and
installed over 2,000 packages successfully until preupgrade said
MingW32-dev was corrupt and the preupgrade had to be cancelled.  It
was a <90 day used very inexpensive seagate 1TB drive, so i figured
bad harddrive.

The 3 hour BIOS level test of my new harddrive tested out ok, but it
does not provide any SMART data or other real information.  But at
least according to this one test, there is nothing wrong physically
with the drive itself.

After holding down the up-arrow key during boot i was able to get the
grub menu to come up, but that took a long time to get the keypresses
just right.   After running 'mount -a'  /boot/ was mounted rw so i
could change /boot/grub.conf.

So i attempted to boot into the older F13, but it comes up with the same:
  "Welcome to emergency mode.  Use "systemctl default" or ^d
  "systemctl default" results in
  "Failed to issue method call:  Transaction is destructive."

There are some websites about creating a default softlink for the new
systemd upstart replacement, but that appears ok already, but i need
to double check.  (Not sure why they felt the need to have yet another
sysV init replacement as F14 boots up pretty fast as it is.)

Once i properly backup /var/log/yum.log,  /home/ and /etc/ and
/usr/local/etc/, everyone says to do a default install from the CD.
Thereafter, /home/ should by default have its own VolumeGroup and it
is safe to do repeat for the next version.  But it appears to me, that
leaves out my old /etc/ and /usr/local/  configured files.  Anything
else i am not remembering to backup?