22 Jul
2002
22 Jul
'02
7:14 p.m.
--- Joe <rigdonj(a)cfl.rr.com> wrote:
It all depends on how modern your target box is (from a software
standpoint). In the case of UNIX, the low level format is typically
optional, but a newfs (and possibly a disklabel operation) isn't.
Same goes for AmigaDOS - if you can write to every sector, you can
bypass the low level format and go right to partitioning (either with
a mountlist for older versions of the OS, or with a more modern disk
tool, presuming your driver supports RDB (Rigid Disk Blocks)).
It's only a problem for certain embedded SCSI drives. The solution
under AmigaDOS, presuming you *don't* have an Adaptec bridge controller
or a multi-LUN SCSI CD-ROM disc changer is to disable LUN support in
the driver - the "modern" model is one target ID, one LUN (typically
'0'). Even the Adaptec can be set up with one LUN - only put one
ST506/ST412 drive on it. The difference is that since the Adaptec
won't respond to IDENT, it can't mis-report how many LUNs are populated.
I don't know about that - mostly, stuff should work fine. I just
wanted to point out a few places where things are substantially
different between using a bridge controller and having an embedded
SCSI drive. I presume much the same is true for the Emuled MD-21
ESDI bridge controller that was common back in the days of Sun3
(MC680x0) hardware... I've seen them but never had to format drives
on them. Once they are formatted and partitioned, they should behave
much the same as an embedded drive - the trick is getting past the
setup software.
-ethan
__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com
At 06:26 PM 7/22/02 -0700, Ethan wrote:
I figured, but I didn't want to come right out and make a blanket
statement and have someone who knows lots more than me quote chapter
and verse of a very late model board that I've never heard of.
--- Don Maslin <donm(a)cts.com> wrote:
That's correct according to what I've been able to find out. ...I see no reason why a regular SCSI
drive could not be installed in lieu of the ACB-4070 and ST-411/506
interface drive.
My experiences with Adaptec bridge controllers is that they don't
support an IDENT packet, so the host has to "know" the geometry,
That entirely depends on the software/firmware on the host.
meaning that _you_ will have to know the geometry
of the embedded
SCSI drive and tell the host since it will not think to ask the
drive itself.
Hmmm. But how do you tell it the geometry? The User's manual for the system indicated that
there are only two
drives that work with this system. I wonder if that's because the
geometry is hard coded into the HD setup program?
Very likely. In the case of the Amiga and SunOS (4.x), with which I have
lots of experience, you edit a text file (a mountlist for AmigaDOS and
format.dat for SunOS) and throughly describe the geometry to the SCSI
driver. The OS then uses that information to place various filesystem
goodies on cylinder boundaries for efficiency (switching heads is a lot
less time consuming than moving them).
Additionally,
whatever (low level) formatting software comes with the
host might or not might not work with an embedded drive... if there
is a canned procedure for installing the OS (old variants of UNIX come
to mind), it might or might not succeed with an embedded drive if it
was written to expect an ACB-4xxx.
That's what I'm afraid of. Finally, most
Adaptec bridge controllers I know of did support multiple
LUNs. Some SCSI drives won't like being probed for LUNs and may return
a hit for every LUN probed (I've seen that happen with some drives on
an Amiga 3000 which optionally supports multiple LUNs per target ID)
Wonderfull! More problems. If none of the
above differences trip you up, then you should have no
problem using an embedded SCSI drive in place of your ACB-4070.
I think that should have been <B>IF</B>!