Hi all,
I recently installed Debian 2.2 on my computer after having used Debian 2.0 for a year and a half. My computer had been complaining often about corrupted disk sectors under Debian 2.0, and I figured perhaps my HD was faulty or Debian 2.2 had a patch to fix the problem. After installing Debian 2.2 the problem persisted, even sometimes fsck would at bootup find irrecoverable HD errors and ask me to give root password for maintenence. So I bought a new 3.2 GB IDE disk, and again the same. I had been using a 1.6 IDE disk.
For both HDs I was using the same partitioning scheme: partitions for / /home /usr and swap, each less than 1GB. What I have noticed is that partitions at the beginning of the disk rarely have problems, whereas partitions above half the capacity of the disk suffer many errors and are slow to read, regardless of their size. So say, if your HD is 3.2 GB, make your partitions are in the first 1.6 GB. If you make even small partitions in the second 1.6 GB you'll find that they are slow to read and unreliable. Also, the HD heads make a lot of noise reading it. If you have programs stored in the higher regions of the disk they will suffer from slow or faulty startups.
So I recommend that you use half the surface in your disk to avoid problems. Also, I recommend that you mount swap right at the begining of the disk, you'll see it makes a big difference.
This doesn't happen only with Linux. I made a small partition at the end of my disk and installed Windows 95 on it, and it was a nightmare: continuous scandisks and surface checks. Even the formatting of the partition was trouble. I've tried to define my disk as LBA mode, Large mode, and ordinary CHS mode, and still the same: don't use the second half of the disk.
While this interim solution is working fine for me, I have several doubts: does this happen generally for IDE disks (I do know it happens to some other people), does it happen with SCSI disks? Does anyone know the cause? To me it seems that the surface is not uniform in quality, but it could be that there is a mechanical problem with the heads reaching far cylinders, or even that the standard HD drives are faulty.
I'll continue to experiment with the disk drive.
Has anyone had the same problem or know a solution?
Cheers,
Jaime Silvela from Spain