Upgrade-2008 Headbutts

Here are the problems I encountered with this upgrade:

• Initially, the system would not boot due to /etc/fstab having some explicit mounts listed which were no longer possible (it should have just booted up without those mounts, in my opinion, but instead it died when it couldn't e2fsck the failed mounts). I booted the F8 DVD in rescue mode, made a minimal fstab (involving VolGroup 00 only), and then it was ok. I don't think it mattered what SATA port I plugged the boot-drive into (due to the drive being ``labeled''), but I plugged it into SATA1. (It was SATA2 on the previous motherboard.)

• Next, X could not launch (understandably). I let the error-handler reconfigure for generic 800x600 VGA mode (the VESA driver), then downloaded the proprietary ATI Radeon driver (ati-driver-installer-8-6-x86.x86_64.run), ran it, chose the precompiled Red Hat driver fglrx v8.501 (``for X.Org 7.1 or later''), logged out and back in, and everything seemed fine until the next reboot (see below).

• Running aticonfig messed up my xorg.conf in various ways. (Perhaps the lossage occurred earlier when X crashed and reconfigured me for the VESA driver.) The first thing I noticed was needing to refresh the setting ``System / Preferences / Hardware / Keyboard / Layout Options / Ctrl key position / Make CapsLock an additional Ctrl'' (very important for emacs users!). However, I was later reminded (by an old xorg.conf file) that the line

  	Option	"XkbOptions"	"ctrl:nocaps"
  

  in /etc/X11/xorg.conf would take care of this, which is why the problem came up since that line was now gone. Also, my keyboard type had been changed to pc105, so I changed it back to microsoft. My very specific monitor entry had been whacked as well. Ultimately, I ended up completely rewriting xorg.conf by hand, referring to earlier saved versions. I'm pretty sure aticonfig used to be more helpful, generating comments in the file suggesting various options one might want.

• My HP laserjet 1220 did not work at first because the new motherboard has no parallel port -- Doh! Fortunately, when you take it apart (wincing as bits of plastic sound like they are breaking off and flying in all directions -- perhaps I should oil the case a bit), you find that it also has a USB port, so I just had to dig up a cable. Phew!

• On my first reboot, it hanged after starting udev (presumably reading the xorg.conf file, in hindsight). Rebooting in single-user mode, I determined it was the ATI Radeon driver (fglrx) that was hanging. I reverted to the backed up xorg.conf in /etc/X11, which called for a standard VESA driver, and then I could reboot. Originally I had installed the recommended ATI driver for generic X11 systems. Since I was losing, I tried building one specifically for Fedora/F8. (The -listpkg option lists all supported cases.) Alas, the build failed with the following message:

  error: %changelog entries must start with *

  Talk about a bozo bug! I got the same failure trying to build for Fedora/F7 and Fedora/FC6. Finally, simply reinstalling the generic fglrx driver as I did originally worked. (Exiting single-user login proceeded to run-level 5 which worked normally.)

• After the next reboot, same problem! Apparently, I have to boot up in single-user mode, then exit to go to run-level 5! Otherwise the machine just reboots over and over again. I verified that the radeon driver did not work at all (``no devices''); it might actually function if the source were hacked to list the 4850 as a known variant, but I'm going to limp along with the fglrx driver for now.

• Something changed my network settings from a fixed IP address to DHCP! Windows XP / Samba took it in stride, but other machines could no longer find my Linux box. I re-entered the fixed IP address et al. in ``System / Administration / Network'', typed service network restart in a shell, and things were back to normal.

I am pleasantly surprised by how cool and quiet the new system is. Since it draws much more power, I was worried that my old Antec Sonata case would not pull out enough air with its one quiet fan in the rear. So far, however, the core temperatures reported by lm_sensors⁹are always less than 40$\hbox{${}^{\circ}$}$C under ``normal'' maximum load for me (compiling a large book in LaTeX). The ``red line'' is said to be at 100$\hbox{${}^{\circ}$}$C. In contrast to this, my two-year-old HP Pavilion notebook (§) idles at core temperatures in the 60s (AMD Turion64 dual-core), and a long project compile drives them up into the 80s and 90s. (The laptop has also shut itself down several times due to overheating.)