On Wed, Apr 16, 2008 at 10:56 PM, Howard Sanner
<[EMAIL PROTECTED]> wrote:
>         For a few weeks I've been getting unrecoverable read errors on
>  bootup. It has only affected various printer config files, and,
>  since I don't have a printer hooked up, that doesn't matter. I
>  have been madly backing things up and figured I was going to have
>  to install a new HD.

These errors did not trigger an fsck?  At what stage in the boot did
they occur?  BIOS/POST? GRUB/LILO? Kernel startup?  Or after the
kernel mounts the root partition (this usually happens ~3s after the
kernel messages start spewing up the screen).

>         The question is this: Do I have a major HD problem lurking that
>  fsck hid, or did fsck really make the problem go away?

It's certainly more likely that you have a hardware problem, although
that could be something as simple as a loose cable.

I would weigh the costs of downtime and a restore from backup (are you
using Amanda? ;) against the costs of a new drive and a controlled
transition.  I generally use LVM, so in this situation I have usually
added a new (and inevitably larger) drive, pvmove'd the volumes to it,
and then used the old drive as /tmp, /scratch, Amanda holding, a RAID
mirror, or swap until it died.  Now that I'm thinking more about green
computing, my preference is to simply retire the old drive, despite
its "mostly working" status.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com

Reply via email to