Re: write errors on secondary scsi LUNs

[Arjen Wolfs <arjen@xxxxxxxx>]

At 11:50 23-10-2001, Matteo Centonza wrote:
> Hi Arjen,
>
> > I have two hardware raid partitions, on an identical SCSI ID; one with LUN
> > 0, one with LUN 1. The partition on LUN 1, which consists of brand new hard
> > disks, is generating random write errors of the kind:
> >
> > SCSI disk error : host 2 channel 0 id 0 lun 1 return code = 8
> >   I/O error: dev 08:21, sector 310378498
> > XFS: device 0x821- XFS write error in file system meta-data block
> > 0x12800002 in sd(8,33)
> >
> > SCSI disk error : host 2 channel 0 id 0 lun 1 return code = 8
> >   I/O error: dev 08:21, sector 310397560
> >
> > Oddly enough, xfs_check and xfs_repair can't find anything wrong with it
> > after this happens. Considering kernels before 2.4.10 would not even
> > recognize SCSI devices on LUNs other than 0, I am not entirely sure wether
> > this is an XFS or a SCSI issue. Anybody have any ideas?
>
> I'm not an HW-RAID expert but i've seen this message before:
>
> http://marc.theaimsgroup.com/?l=linux-xfs&m=100319330327289&w=2
>
> indicating a raid layer level corruption. XFS notify this way that
> something is going wrong  with you  array. Do sanity check of your
> hardware (disks, controller, scsi terminations). It's likely that's
>
> a hardware based problem.

It's definately not a cabling issue; the partition on LUN 0 goes over the 
exact same cable/controllers and has no errors whatsoever. The HW RAID is 
external (i.e. in the disk cabinet), and shows no errors.

> Which controller are you using?

a Seek RAID unit connected to an Adaptec 2940U2W controller.

/A