Ok.. Strangely enough, I tried running Solaris for a few more days. Then,
I "downgraded" to NetBSD 1.5 instead of 1.5.1, and I haven't had any SCSI
errors such as the ones that I was experiencing with 1.5.1. I guess I'll
poke around in cvs logs, though I know nothing of C...

On Wed, 1 Aug 2001, Jeff Wyman wrote:

> I got the block number to use on reassign from the output of an SCSI error
> message, which was "Check condition on CDB" and printed "CANNOT READ BLK
> X" on the Info Field. Reassigning never fixes this for NetBSD, even though
> the scsi reassign command is successful.
> 
> > If you were feeding the block numbers from fsck into "scsictl
> > reassign", that may be the reason it didn't work.  You need to get the
> > physical block number.  If you have the SCSIVERBOSE option in your
> > kernel, the error message will be displayed in a quasi-readable
> > format.  The physical sector number, if I remember correctly, will be
> > in the "Info Field" of the error sense code dump.
> > 
> 
>