Accessing SCSI-Devices >2TB

[Bernd Walter]

On Wed, Jun 08, 2005 at 04:55:27PM +0200, Raphael H. Becker wrote:
> On Wed, Jun 08, 2005 at 04:24:10PM +0200, Bernd Walter wrote:
> > > And 1kb would extend the 2TB limit to 4TB, what would work for me here.
> > 
> > Obviously, but that's not required - you can have SCSI disks with more
> > than 2^32 512 Byte blocks since more than 2 years.
> 
> Then we might have some other problems accessing that RAID. When
> configured as a large 2620354MB logical drive (no logical drive partitions):
> 
> [...]
> (da1:ahc1:0:0:0): No or incomplete CDB sent to device.
> (da1:ahc1:0:0:0): Protocol violation in Message-in phase.  Attempting to abort.
> (da1:ahc1:0:0:0): Abort Message Sent
> (da1:ahc1:0:0:0): SCB 9 - Abort Completed.
> (da1:ahc1:0:0:0): No or incomplete CDB sent to device.
> (da1:ahc1:0:0:0): Protocol violation in Message-in phase.  Attempting to abort.
> (da1:ahc1:0:0:0): Abort Message Sent
> (da1:ahc1:0:0:0): SCB 14 - Abort Completed.

Obviously something wents wrong.
You might want to add CAM and/or ahc debugging.
Asking on freebsd-scsi list would also a good idea.

> Using 1 logical drive with two partitions mapped to two LUNs (resulting
> in da1 and da2 in FreeBSD) works perfectly. 
> 
> BTW: I'm actually running a nearly-GENERIC kernel 5.4-RELEASE here, just
> with 
>         options SMP
>         device ccd 
> 
> added. Maybe I will need some other options? What about GEOM? 
> 
> But for me it smells more about the SCSI-protocol between RAID and
> Controller (Adaptec 39160), something about LBA-64?
> Maybe running -current will give me this feature?
> 
> Running ccd upon a partitioned RAID is "A Bad Idea[tm]".
> 
> Any pointers to documentation? Maybe I just need some deeper
> knowledge about whats going on here.

-- 
B.Walter                   BWCT                http://www.bwct.de
bernd at bwct.de                                  info at bwct.de