[comp.sys.amiga.hardware] Multiple SCSI controllers

lphillips@lpami.wimsey.bc.ca (Larry Phillips) (05/27/90)

In <90146.174753BRANDON@ncsuvm.ncsu.edu>, BRANDON@ncsuvm.ncsu.edu writes:
>  I was wondering if I could have multiple SCSI host controllers on the
>same cable run.  I have 2 amigas with SCSI controllers and one disk drive.
>Currently, I am using the parallel port network from the Software Distillery
>to have both computers access the same drive.

Yes, provided that both controllers (their drivers as well as hardware) support
arbitration.

>  It would seem to me that there should be nothing wrong with having one
>disk drive and two host controllers on one SCSI "network" from a hardware
>point of view.  From the software end, I realize that I would have to have
>separate partitions on the disk to write to, as there would be no way to
>coordinate write requests from both machines.

That's the big problem. You can never wrote to the 'other' partition. I keep
meaning to sit down and figure a scheme that would be safe, to allow some sort
of completely shared disk, at the partition level. It seems to me that you
could come up with a way to have each machine write only to its own partition,
and to set up a couple of partitions to act as 'message ports', with a daemon
to look at the files on the message partitions from each machine.

>  Has anyone attempted this?  As all I need to do this is two header
>connectors and a few feet of ribbon cable, I'm tempted to try it just to test
>the other SCSI controller.

Not yet, but it sounds like it would be worthwhile. I may just try it before I
get rid of my other machine.

-larry

--
The raytracer of justice recurses slowly, but it renders exceedingly fine.
+-----------------------------------------------------------------------+ 
|   //   Larry Phillips                                                 |
| \X/    lphillips@lpami.wimsey.bc.ca -or- uunet!van-bc!lpami!lphillips |
|        COMPUSERVE: 76703,4322  -or-  76703.4322@compuserve.com        |
+-----------------------------------------------------------------------+

BRANDON@ncsuvm.ncsu.edu (05/27/90)

  I was wondering if I could have multiple SCSI host controllers on the
same cable run.  I have 2 amigas with SCSI controllers and one disk drive.
Currently, I am using the parallel port network from the Software Distillery
to have both computers access the same drive.

  It would seem to me that there should be nothing wrong with having one
disk drive and two host controllers on one SCSI "network" from a hardware
point of view.  From the software end, I realize that I would have to have
separate partitions on the disk to write to, as there would be no way to
coordinate write requests from both machines.

  Has anyone attempted this?  As all I need to do this is two header
connectors and a few feet of ribbon cable, I'm tempted to try it just to test
the other SCSI controller.
                                                  BlH

P.S. - To all you out there trying to get a 68000 running 'right' at 14MHz
      Why not just go for a 68020?  Once you correct the signals to get
      a 68000 running at 14MHz, you have nearly done everything necessary
      to hook up a 68020 at 14MHz also.  (Except figure out how to route
      the PGA signals to the right places on the 68000 socket...)

bscott@nyx.UUCP (Cos McCowboy) (05/27/90)

In article <90146.174753BRANDON@ncsuvm.ncsu.edu> BRANDON@ncsuvm.ncsu.edu writes:
>
>  I was wondering if I could have multiple SCSI host controllers on the
>same cable run.  I have 2 amigas with SCSI controllers and one disk drive.
>Currently, I am using the parallel port network from the Software Distillery
>to have both computers access the same drive.

I think it's been done.  A developer friend of mine who's fairly knowledgable
about hard drives uses what he calls a SCSInet implementation from CLtd to
use 2 A-1000s and however many HDs he has at the present time on one ribbon.
He says it's nice because to add another HD he just cuts the ribbon, and splices
connectors on either end, and voila.

I don't know much else about it; I know CLtd is gone.  I don't know where you
could get this software, or if you even need custom software for this, or how
well it works.  I do know he has to do a Diskchange if he writes to a partition
that is mounted on both machines. (on the other machine)

>P.S. - To all you out there trying to get a 68000 running 'right' at 14MHz
>      Why not just go for a 68020?  Once you correct the signals to get
>      a 68000 running at 14MHz, you have nearly done everything necessary
>      to hook up a 68020 at 14MHz also.  (Except figure out how to route
>      the PGA signals to the right places on the 68000 socket...)

If you would care to inform me of the procedure, I'd be grateful... I can't get
the 14 Mhz hack to give me more than a Green screen, and I have a 68020
chip sitting around from the time I tried a CSA Midget Racer and found it don't
work with the A590... I'm a menace with a soldering iron but I've been filthy
lucky so far messing around with my 500.

.                           <<<<Infinite K>>>>

|Ben Scott, professional goof-off and consultant at The Raster Image, Denver|
|Internet bscott@nyx.cs.du.edu, FIDO 1:104/421.2 or Arvada 68K (303)424-9831|
|"My brothers and sisters all hated me,'cause I was an only child!"-Weird Al|
|"Do I detect the smell of burning martyr?" - Basil Fawlty  | *AMIGA POWER* |

billsey@agora.uucp (Bill Seymour) (05/28/90)

In article <90146.174753BRANDON@ncsuvm.ncsu.edu> BRANDON@ncsuvm.ncsu.edu writes:
:
:  I was wondering if I could have multiple SCSI host controllers on the
:same cable run.  I have 2 amigas with SCSI controllers and one disk drive.
:Currently, I am using the parallel port network from the Software Distillery
:to have both computers access the same drive.

	You can do this easily, but with some ...features. In most usage,
your SCSI interface will only be looking for data on the SCSI buss for
requests that your interface has initiated. That means you can hook the
two up and mount the drive on both systems. The drive sends data out when
requested and both computers recieve that data, but only the computer that
asked for the data will actually use it. The biggest problem you run into
with this type of setup is that one machine will not know when the other
machine has made a change to a partition. If you have a partition that's
completely empty hooked to computer A and to computer B and you save a 50K
file onto the partition from comuter A, computer B will think the partition
is still empty. If computer B then saves a 20K file onto the partition,
computer A will now find a currupt file system when it tries to read. The
solution (short term) is to always do a diskchange before accessing a
partition, and never cause a file or directory lock that would defeat that
diskchange. The solution (long term) is to change the SCSI ID of one of
the computer interfaces and rewrite the file system to notify one SCSI
host when the other SCSI host makes changes... Or use ParNet to arbitrate
your changes... :-)

:  It would seem to me that there should be nothing wrong with having one
:disk drive and two host controllers on one SCSI "network" from a hardware
:point of view.  From the software end, I realize that I would have to have
:separate partitions on the disk to write to, as there would be no way to
:coordinate write requests from both machines.

	I actually used this type of 'network' on my system here at home
for a couple of years. I had Supra interfaces hooked to both the 1000 and
the 2000. They shared a (read only most of the time) workbench and each
had a partition that was earmarked for their personal use... I have
switched to ParNet now and am much happier... I also have to rebuild
my hard drive setup much less often. :-) ParNet runs about twice the
speed of a floppy, so it's not too bad for me to use.
	Note that I have an expansion card for my 2000 with another
parallel port on it, so printing while both machines are using the
parallel port isn't too much of a hassle...

:  Has anyone attempted this?  As all I need to do this is two header
:connectors and a few feet of ribbon cable, I'm tempted to try it just to test
:the other SCSI controller.
:                                                  BlH
:



     -Bill Seymour             ...tektronix!reed!percival!agora!billsey
=============================================================================
Bejed, Inc.       NES, Inc.        Northwest Amiga Group    At Home Sometimes
(503) 281-8153    (503) 246-9311   (503) 656-7393 BBS       (503) 640-0842