[Dirvish] Disk drive follies

Eric Mountain em-dirvish-1 at nerim.net
Mon Feb 28 06:11:07 PST 2005

On Friday 25 Feb 2005 22:00, Jason Boxman spake thus:
> On Friday 25 February 2005 15:38, Keith Lofstrom wrote:
> > This came at a good time for me;  I rotate 3 250GB hard drives for
> > backups, and I just lost one of the 3 (a Maxtor I purchased in
> > September).  It is still under warranty, so I Maxtor will replace it.
> Entirely anecdotal given my small sample size, but my past two of three
> Maxtors have failed on me.  The second was a replacement for the first.  I
> kept them from running extraordinarily hot to no avail.  I have had four
> WDs for about as long and none have failed yet.

More anecdotes: I'm running on an IBM DeskStar bought early September 2001. 
That series actually got nicknamed "DeathStar" due to allegedly high return 
rates.  Well... I can't complain, but I *backup*. ;-)

Hmmm, I backup to a Western Digital Caviar (13Gb).  I have to disable its 
write cache because it is buggy.  It used to be my main drive on my previous 
machine, and I use to get strange errors (the odd "deleted inode has zero 
dtime", and once lost /etc :( ).  When I got the new machine, I started off 
by testing ext3* on the WD drive.  My partitions became corrupt beyond 
imagination without me even "using" them.  Seems like if you have a bad 
drive, you are better off not having a journalling filesystem.  Anyway, now 
hdparm disables the write cache on the drive on boot and I haven't had any 
problems since!!!

* Linux Extended Filesystem 3 (has journalling, ext2 doesn't).

Eric Mountain

More information about the Dirvish mailing list