Weird Filesystem Problem on PE2650/RH7.2

jason andrade jason at rtfmconsult.com
Tue Mar 4 19:34:01 CST 2003


On Tue, 4 Mar 2003, Joe Stevens wrote:

> >>Feb 24 06:03:49 brickdb kernel:  I/O error: dev
> >>08:11, sector 15876992
> >>Feb 24 06:03:49 brickdb kernel:  I/O error: dev
> >>08:11, sector 15876992
>
> >Here, there's a problem writing (I assume) to the
> >disk. The sectors mentioned are probably bad.
>
> That was my first thought, but the volume in question
> is a RAID-5 set.
>
> Can that*happen*?  Shouldn't the disk with the bad
> sector be marked as "failed" and the raid continue in
> degraded mode, transparent to the OS? (which was my
> understanding of RAID-5)

it _can_ happen though obviously it is a lot rarer than
in single disk systems.  remember, raid is about mitigation
of risk but it doesn't eliminate it..

it could be

o a kernel error/bug of some sort

o a physical error with the system, e.g with cables, or
  the raid controller or some other piece of equipment
  apart from the disk drive itself.

o a simm that momentarily glitched on the raid controller

o cosmic rays ? :-)

regards,

-jason




More information about the Linux-PowerEdge mailing list