Pinnacle Magneto Optical Disks

[Bernard J. Duffy]

In article <216*finn.drablos at sintef.no> finn.drablos at sintef.no (Finn Drablos) writes:
>In our everlasting hunt for more disk space we have started to think
>about magneto optical disks. We have been contacted by a company
>selling Pinnacle systems, and we have been looking at the REO-650
>disk. It is a SCSI disk, and we want to connect it to a 4D25TG. But
...

Finn,  If you get an Optical drive working on a 4D25TG, let me know....

    We've tried Genesis' Sony drive on our 4D/25 (that good old Sync
SCSI) and it didn't work.  It had a nasty problem when formating/testing
platters where it took over the I/O SCSI system while it was doing 
the many read/writes to the drive.  The system basically "hung" still and
resumed after the operation was completed.  I don't know if we were
expecting too much, but the hard drives (CDC / WREN V[II]) units would
not do that when we would do the  fx -x  format.  We do one of their
units on a 4D/220 and also had it on a 4D/20 (PI-Async SCSI) and it
didn't have this side effect.
    We also tried Q-systems' unit and got similar problems.   It also
had some boot up problems where it failed the diags. and the IRIX
would prompt for "Hit return to continue".

    Since we were having such a hard time with the optical drives, we
decided to get 2 hard disks (760Mb) and an 8mm tape drive.  The drives,
which I believe operate in either SYNC or ASYNC SCSCI modes, have performed
like all the other hard drives I have "sucessfully" placed on other
machines.  I quoted "sucessfully" because we had 4D/2xx systems that
ran diags the drive could not handle.  The Genesis OS patch in accidently
fixed this problem.  The 4D/20 systems never had any such problems (yeah
for the ESD folks).  It's been a real treat to hear that "SCSI termination"
cry for the SGI folks.  That battle cry only came from the ASD folks where
they expected folks to be using SGI drives or ones that responded like SGI
drives to SCSI commands few people or OSs use (connect/disconnect).
    The 8mm (Exabyte) has been difficult and I'm still trying to get
it to work with the 2 extra drives.  I have got it to work with just the
one " /, /usr, swap " disk.  It also "breaks" the boot up diags which
results in my favorite "hit return to continue".  I'll be trying the "C"
(capital C boot flag) later today.  If anyone has an 8mm working on a
4D/25 let me know.

For you folks who need to deal with "SCSI termination" take this advice
from my experience with 8 disk drives and 3 optical drives from vendors
other than SGI (aka 3rd party) :

   Stick with the silver termination connector on the end of your
   SCSI bus (chain).  With that setup, removal all SIPs, transistor
   packs (sp?)  that provide "internal termination".  Also set the
   power for the termination (TERMINATIOR POWER) to external.  On 
   most SCSI boards it's the first two jumpers of about 8 ... generally
   it's configured in the horizontal position :

     .   .   .   .   .   .   .    the others are used to set the SCSI id

   ---------                      You can also leave it off, but we don't
   | .   . | .   .   .   .   .    want to be collecting these jumpers.
   --------- 

   Remember one other thing :  the SCSI id's only range from 0 to 7 , BUT
                               one of them is used by the controller.  I
                               accidently set a device's id to 0 on a SGI
                               and found out the odd way (no damage...
                               the system's diags tried to communicate
                               with a tape drive for SCSI checks      )

There's one other thing... In contradiction to the SCSI "standards", some
SCSI devices "require" to be placed in a particular order on a SCSI bus
in order to work.  They operate as if there was a termination problem
when they were placed in the "wrong" spot.  Typically this means the
unit wants to be closer to the SCSI controller.  I haven't had this
problem with any of the hard drives I've installed.

Sorry for the long message... it's just I've suffered enought with the
cold war on this termination stuff / 3rd party devices / SCSI devices you
had prior to moving up to a SGI.

    Bernie Duffy ...  SCSI victim

-- 
Bernie Duffy   Systems Programmer II | Bitnet    :  BERNIE at UMBC2
Academic Computing Services - L005e  | Internet  :  BERNIE at UMBC2.UMBC.EDU
Univ. of Maryland Baltimore County   | UUCP      :  ...!uunet!umbc3!bernie
Baltimore, MD  21228   (U.S.A.)      | W: (301) 455-3231  H: (301) 744-2954