>Date: Tue, 31 Jul 2001 01:10:07 -0700 (PDT)
>From: Jeff Wyman <wysoft@c481444-a.bremtn1.wa.home.com>
>
>I am not sure wether or not this problem is directly related to NetBSD,
>but it seems to be, so I will get on with explaining...
>
>I have a Seagate ST51260N hard disk hooked up via external drive enclosure
>to my SS5/110 running NetBSD 1.5.1. Everything seems to work fine - NetBSD
>formats the disk and installs happily, and ran for about 2 days straight
>before first reboot after I installed it.
>
>Now, if anything happens that causes the machine to reboot abnormally, the
>system performs an fsck_ffs on reboot and the disk stops cold with "Check
>error on CDB" or something of that sort, says "Media error" and complains
>that it cannot read block so-and-so. I would normally say that the drive
>is at fault, but I had this drive previously installed in a PC running
>FreeBSD with no such problems when running fsck. Plus, on the FreeBSD PC I
>achieved nearly 92% usage (2.1GB disk) and never encountered any block
>errors. Even Solaris' 'format' and 'analyze' tools found the disk to be
>OK, and Solaris never had any problems running fsck.
>
>I'm not particularily SCSI-knowledgable, but my termination seems
>fine. Could this be related to the tagged queueing bug, something strange
>in the system, or am I just an idiot? :)

Try reassigning the bad blocks.  E.g. if it complains about block
12345 on device sd0:

scsisctl /dev/rsd0c reassign 12345

--
entropy -- it's not just a good idea, it's the second law.