2400 RAID5 Problems

Jima jima at beer.tclug.org
Wed Dec 4 08:25:01 CST 2002

On Wed, 4 Dec 2002, jason andrade wrote:
> replace the drive.  wait for a short while.  then reboot the box (note,
> not power cycle, just try a reboot.  if this fixes it, great.  if it
> doesn't, you might want to wait for the rebuild to finish (which could
> take from 2-X hours depending on capacity of disks and load), and then
> power cycle the machine.
> alternatively, a (possibly riskier) approach is to reboot/power cycle
> the box (as needed) and then replace the disk.  this allows you to
> use afacli to verify the rebuild operation.

 While a bit primitive, watching the drive activity lights (with the front 
panel off) can give you a good idea that the array is rebuilding, if there 
isn't much other disk activity.  By watching the frequency of blinks you 
can tell which drives are being read (to produce the parity data) and 
which are being written to (hopefully the new drive!).  (This works with 
RAID 5, anyway.  RAID 1 might be a little harder to tell with.)
 It's far from an ideal solution, but it can do the job in a pinch, when 
downtime isn't an option.
 I suspect Jason is correct about a reboot fixing the afa0 problem.  
Either way, though, my question to Shawn is, what does /dev/afa0 look 
like?  Mine, on a PE2400 with RedHat 6.2 (2.2.14-5.0smp):

crw-------    1 root     root     254,   0 Apr 11  2002 /dev/afa0

 Just thought I'd offer my bit of experience on the subject.


More information about the Linux-PowerEdge mailing list