Re: Huge unreliability - does Linux have something to do with it?

[Jim Nelson <james4765@cwazy.co.uk>]

jerome lacoste wrote:
> [Sorry for the sensational title]
> 
> I have had this laptop for three years. It ran Linux (Debian unstable)
> from the start and its hardware has been very unreliable: I changed
> hard disks twice and the motherboard thrice. My DVD drive started
> failing some days ago (this one is 'original', 3 years old). But I
> don't mind as I am not under warranty anymore... This morning the
> machine booted with fsck errors on my hard disk. I am not sure if I
> did the right thing, but I said clear the inodes, and I ended up
> loosing some programs(*) (du, dircolors, etc..). The day starts well
> isn't it? Sounds like I will have to switch disks again...
> 
> I halted the machine correctly yesterday night. I never dropped the
> box in 3 years. Am I just being unlucky? Or could the fact that I am
> using Linux on the box affect the reliability in some ways on that
> particular hardware (Dell Inspiron 8100)? I run Linux on 3 other
> computers and never had single problems with them.
> 
> How can the file system (ext3) be messed up the way it was this
> morning after I stopped the machine correctly yesterday?
> Could a hardware failure look like bad sectors to fsck?
> 

It can.  I had a drive crash on my server a couple of months ago, and I had ext3 
errors show up before the syslog filled up with the ide errors.  The hard disk was 
only 1 1/2 years old.

If the bad sectors happen where directory inodes are written, your directory 
structure will be turned into swiss cheese.  That will *definitely* cause ext3 
errors, and dump you (in Red Hat systems, at least) to a shell on reboot.

> Attached the output of smartctl -a /dev/hda, whatever that helps.
> 
> Jerome
> 
> (*) I accept tips on discovering and maybe recovering which files have
> been taken out of my system...
> 

You might not have any luck.  After fsck -f, I thought I had saved the drive, 
copied everything that was left onto another machine, and found that most of the 
larger files had holes in them - mp3's had skips, jpegs were completely corrupted, 
etc.

That's what made me get a backup FireWire drive... :)