Path: utzoo!utgpu!water!watmath!clyde!rutgers!mit-eddie!uw-beaver!tektronix!zeus!amadeus!rogers
From: rogers@amadeus.TEK.COM (Roger Southwick (rogers@dadla.TEK.COM))
Newsgroups: comp.unix.questions
Subject: Re: 4.3 on VAX 8xxx?
Message-ID: <3007@zeus.TEK.COM>
Date: 19 Jan 88 19:49:18 GMT
References: <4350002@wdl1.UUCP>
Sender: news@zeus.TEK.COM
Reply-To: rogers@amadeus.UUCP (Roger Southwick (rogers@dadla.TEK.COM))
Organization: Tektronix, Inc.
Lines: 69

In article <4350002@wdl1.UUCP> bobw@wdl1.UUCP (Robert Lee Wilson Jr.) writes:
>
> We run 4.3 on a badly overloaded Vax 11/780. In planning an upgrade we
> confront a statement that Vax 8xxx machines don't/won't have BSD family
> Unix available. I would appreciate mailed comments on the following two
> questions:
> (as always, if there seems to be wide interest or a pattern in
> the answers I will distribute that back to the net.)
> 
> (1) How accurate is the statement regarding 8xxx systems and BSD? (We are
> of course checking with Mt. Xinu, but it is always nice to have other
> sources.)

I know that 4.3 will not run on an 8XXX machines, and have heard that
4.4 (now in Beta Test) WILL run on 8XXX machines.

> (2) (Perhaps even more important, for reasons of internal politics... :-))
> If we were to switch from 11/780 with 4.3 and a lot of local customization
> to 8xxx with Ultrix, what sort of incompatibilities could we expect to
> face?

I have one 4.3 BSD 780 and a 8700 running Ultrix 2.0.  While most of
our binaries work just fine, there are quite a few that require
re-compiling (anything to do with sockets) and I do have one program
that will re-compile, has no lint -habxc errors, and yet refuses to
work properly.  I don't guarentee binary compatibility to my users.

Now as far as customization in the kernel goes, forget it.  I got the
source, and it was incomplete...no drivers for anything on the Bi-bus
except for the dmb-32.  I've gotten a few binary kernel patches from DEC,
and no source patches....so I don't dare run a kernel compiled with my
source...

DEC also does not send you patches automatically...you have to wait until
you have a problem, and then they'll send you a patch...

One of our patches was for the tu81+.  It had a bug in the driver where
a bad tape would cause the driver to lock up.  The only solution was to
re-boot (ick).  When I called DEC in, they checked around, and  after a
few hours of looking found a patch that 'they had known about for 3
weeks'.  The patch they sent did not fix it, but provided a way to at
least re-set the error (mt clhrdsf).  Now this is wonderful, but a pain
if you get an error in the 3rd tape of a 3 volume dump.... You abort
the dump, and do the mt command, then restart the dump (and go thru
tapes 1 & 2 again).

> How well will Ultrix deal with our Ethernets and Cisco gateways
> going off to Arpa?

Don't know that answer to that one.

> How consistent are goodies like system calls, i.e. can
> we expect just to recompile C sources for local code and go?

Some things will work just fine.  Some don't...

> I would really appreciate hearing from anyone who has been through this
> switch: If your experiences are too complex to type in, send me a telephone
> number and I will be glad to foot the bill for a lengthier discussion.
> 
> Thanks for any information you can give,
> Bob Wilson at Ford Aerospace and Communications Corp.
> (bobw@ford-wdl1.arpa, (415) 852-6572)

If you want, Bob, drop me a note, and I'll mail you my phone number.

	-Roger (rogers@dadla.LA.TEK.COM)
		UUCP:	{ihnp4 | decvax | ucbvax}!tektronix!dadla!rogers
		ARPA:	rogers%dadla.LA.TEK.COM@RELAY.CS.NET