During the last week I experienced some rather weird I/O behaviour on Huberts one-and-only harddisk. It’s been a 3Gigs Seagate Medalist 3221; yeah it’s been. Because now this is what I call a dead hard-drive. After I witnessed another I/O crash while accessing a website hosted on hubert, I today decided to replace it.

My first approach was:

1. get new hard-drive
2. create identical partition structure on new drive
3. copy partition data
4. have fun

Did you see the important thing I missed?
Right: I missed to install the bootloader. Silly me.

But it came worse and from behind. The first three points went good. But at reboot time the BIOS of my good’ol PII/BX-Mainboard wouldn’t accept the Hitachi drive. Great. Then the old Seagate just decided to die totally and everything I got from her wer some cryptic LILO errors where everything should run smooth and shiny.
Thus I struggled to get the old drive back one more time:

• tried to get into the LILO error - useless
• tried to boot Knoppix and reinstall LILO in hda’s MBR - no good!
  Knoppix prove to be unsuitable for such a task. It has a weird chroot behaviour.
• tried an Ubuntu Install-CD for LILO recovery - LILO reinstallation succeeded, but at boot-time this only proved that LILO was not the source of the error.
• I also tried to reconfigure the old HD in my BIOS, since I had to change everything for my new Hitachi-drive. - no succes with this either.

Well, anyone who knows me well enough, knows also that this is not the point where the Prakti resigns, isn’t it?
But then I was lucky too: found an old 10GB IBM HTTA drive, copied the data from the hitachi onto it, ran LILO and *tadaa* Hubert is back.

As a Summary:
The most-important commands today were: tar for copying consistently (cp would result in damaged symlinks) , mke2fs, cfdisk for partitioning, chroot, and of course mount. Everything was done from a shell from within a live-system of an Ubuntu-Install-CD.

Good Fight, Good Night.