file system corruption

[jdow]

From: "chris" <fedora at parallelsw.com>
> Hi all,
> 
> I am trying to figure out whats going on with a system of mine, I have a 
> fairly cheap IDE drive in it (160GB Western Digital),  but over the past 
> 2 weeks 2 of the partitions ("/" and "/home") have been corrupted.  They 
> are both formatted ext3, and one of them is the root for FC4. 
> 
> I don't see any errors in /var/log/messages, and am not sure where to 
> look.  I just want to verify the drive is going out before replacing it. 
> 
> Can anyone point me in the direction of some sw tools are out there to 
> help in diagnosing the problem?

If this is a new problem that is getting bad fast then the drive is
going bad. Back it up and apply for a warranty replacement if it is
still in warranty. (Get their PowerMAX utility to test the disk. You
have to anyway to get an RMA. It will tell you what is wrong with
the drive.) If it is not in warranty then get a replacement. Meanwhile
tuck critical work you've done off on any other drives you can find.

Now, that said there is one other thing I found that can do this. I
had, back in the 2.0.24 days, a system with an Adaptec PCI SCSI
controller and both IDE and SCSI drives. For some reason that seemed
mysterious until I accidentaled upon the right place on the web the
IDE drive contents were getting mysteriously corrupted one byte at a
time. It turns out the Adaptec BIOS had a defect leading to this
effect. I updated the BIOS and never had a problem again.

Basically, errors don't happen at this level unless there is some form
of hardware problem. And drive do die early in their lives for some
people. That is the nature of "Mean Time Between Failures." That is
only an average time. It means for some failures will happen early
on. (Infantile failure rates can be rather high, too. Then failure
rates go down to a midlife plateau before getting bad again as the
drive approaches its MTBF.)

{^_^}