On Wed, Jun 05, 2002 at 12:52:50PM -0700, Matthew Jacob wrote:
> 
> Hmm. Very interesting. Thanks.
> 
> The drives that are failing for me negotiate at an intermediate rate between 5
> && 10MB/s:
> 
> sd0 at scsibus0 target 0 lun 0: <SEAGATE, ST31200N SUN1.05, 9424> SCSI2 0/dire
> ct fixed
> sd0: 1006 MB, 2700 cyl, 9 head, 84 sec, 512 bytes/sect x 2061108 sectors
> sd0: sync (124.0ns offset 15), 8-bit (8.064MB/s) transfers, tagged queueing
> sd1 at scsibus0 target 1 lun 0: <SEAGATE, ST1480   SUN0424, 7516> SCSI2 0/dire
> ct fixed
> sd1: 411 MB, 1476 cyl, 9 head, 63 sec, 512 bytes/sect x 843284 sectors
> sd1: sync (200.0ns offset 15), 8-bit (5.000MB/s) transfers
> sd2 at scsibus0 target 3 lun 0: <MICROP, 2217-15MQ1091501, 4930> SCSI2 0/direc
> t fixed
> sd2: 1683 MB, 2372 cyl, 15 head, 96 sec, 512 bytes/sect x 3447008 sectors
> sd2: sync (124.0ns offset 15), 8-bit (8.064MB/s) transfers, tagged queueing

  This looks familiar.  ISTR a kernel or two a while back (maybe last year?)
that reported transfer speeds as some strange number a little above 8 MB/s.
That was on a system that had previously and has since reported 10 MB/s.
Are you sure you don't have any sticky tags in your kernel source tree?

  If it matters to you, I can try booting a few of my old kernels to see if
I can spot one or more that exhibited this oddball rate.  (I'll be near the
console for a couple more hours or so.)

--Jim