Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site ariel.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxn!mhuxj!houxm!vax135!ariel!jlw
From: jlw@ariel.UUCP (J.WOOD)
Newsgroups: net.periphs,net.wanted
Subject: Re: Question about disk drives
Message-ID: <804@ariel.UUCP>
Date: Fri, 21-Dec-84 09:23:21 EST
Article-I.D.: ariel.804
Posted: Fri Dec 21 09:23:21 1984
Date-Received: Sat, 22-Dec-84 02:10:23 EST
References: <213@psivax.UUCP> <427@amdcad.UUCP>
Distribution: net
Organization: AT&T-ISL, Holmdel, NJ
Lines: 39
Xref: watmath net.periphs:613 net.wanted:5410

We have had problems getting our RA-81s and 60s running well
on UNIX System V.  We sort of ported what was originally the
Rice? Univ. BSD driver.  When we bought the drives we also
ordered dual UNIBUS adaptors so we could run the RAs off
one UNIBUS and all the other junk off the other.  We have had
myriad problems.  We hacked SV to add dual adaptor support
but have never been able to get NPR devices to work on
UNIBUS 1 (ie the second UNIBUS).  Query, has anyone out there in
netnewsland done this? 
Also in our rage of rewirings we discovered that with an
expansion UNIBUS there are cartain configurations that don't work.
This may not be too clear.  We were in the throes of trying
anything that would work.  DEC recommends that the UDA50 be
the last guy in the UNIBUS so the buckpassing lets other devices
of the same priority class,  in our case DZ/KMC and DMRs, get
access first.  Since the UDA has  buffering it supposedly
can stand some delays of the millisecond range while the others cannot.
We had the UDA50 adaptor boards in the last slot of the internal
750 UNIBUS.  We wanted a single UNIBUS system with this at the very end
of the chain, so we went from the adaptor on the processor for
UNIBUS0 out to the expansion cabinet and then back into the
main cabinet to the internal slots.  This configuration won't
boot because of the power-up test sequence of the 750 powers the
internal UNIBUS first and then the expansion box, causing the
power up diagnostics to fail and lock the processor. Aw darn.

So now we are running single UNIBUS and aren't
too happy with the performance, and every once in a while
the system crashes due to, we think, an interaction of
the DMR11 (spell KMC11-DMC11) and the UDA50.  Sigh.



					Joseph L. Wood, III
					AT&T Information Systems
					Laboratories, Holmdel
					(201) 834-3759
					ariel!jlw