[K12OSN] RAID1 failure: need help

[Les Mikesell]

On Thu, 2005-10-20 at 12:14, Robert Arkiletian wrote:

> Les the same drive (channel B) has now been kicked out of the array 3
> times.  But I can always add it back with mdadm after a reboot. I have
> to mention that I have two scsi drives mirrored but they are on
> different channels (with seperate cables) but use the same controller.

I think the md driver is very sensitive to errors.  It will kick
drives out where a single drive would continue retries.

> Therefore I think it must be either drive B or cable B. I am wondering
> if I can exclude the cable by swapping the drives. So put drive A on
> cable B and vice versa. Since both drives have grub on the MBR (thanks
> to you) I  hope it will boot. Will it?  If it does work, I think I
> should be able to deduce if it's the drive or the cable, the next time
> it gets kicked out.

It may just be a matter of which has the first error.  Whichever one
stays active will have additional retries and may keep working.  Either
drive should boot, though.

> BTW I have always had the controller set to 160 for both channels (not
> 320). Also, cable A is 3ft round type and it says LVD/SE on the
> terminator. While cable B is 4ft flat twisted pair type and  it has
> the LVD symbol on the terminator. Both drives are on the last
> connector (just before the terminator as they should be for high speed
> scsi)

That all sounds right as long as there are no SE devices on cable B.
I always like to do a low-level format on scsi drives at the first
opportunity, using the same controller type that will be used for
normal use. Sometimes it helps, sometimes it doesn't.

-- 
   Les Mikesell
     les at futuresource.com