Sun-Spots-Request@RICE.EDU (William LeFebvre) (06/14/88)
SUN-SPOTS DIGEST Monday, 13 June 1988 Volume 6 : Issue 111
Today's Topics:
Re: Sun losing track of console
Re: text table full
Sorcerer's Apprentice -- Multiple invocation of Daemons
SunOS 4.0 doesn't work with A79J MC68881
Using previously created colormap segments
why is my 3/60 so slow?
subnet mask console message?
Troff viewer?
DES driver?
VT100 emulators?
Questions about Sun systems
Computer aided test procedure/case development
Send contributions to: sun-spots@rice.edu
Send subscription add/delete requests to: sun-spots-request@rice.edu
Bitnet readers can subscribe directly with the CMS command:
TELL LISTSERV AT RICE SUBSCRIBE SUNSPOTS My Full Name
Recent backissues are available via anonymous FTP from "titan.rice.edu".
For volume X, issue Y, "get sun-spots/vXnY". They are also accessible
through the archive server: mail the request "send sun-spots vXnY" to
"archive-server@rice.edu" or mail the word "help" to the same address
for more information.
----------------------------------------------------------------------
Date: 8 Jun 88 03:04:57 GMT
From: tekbspa!tss!joe@uunet.uu.net (Joe Angelo)
Subject: Re: Sun losing track of console
I've seen a number of messages regarding "Sun's losing track of console"
and wish a working solution were posted as well! Here's some more input to
anyone trying to solve the problem:
- In our Palo Alto office, a 3/260 uses a wyse75 as system console
on ttya -- all's well as MOTHER EARTH grounds the joint.
- At one point, in bean town, a PILOT was configured with a 3/260
and a wyse75/ttya system console. All worked fine. Later, during
system integration time, 6 3/260's where configured with wyse75/ttya
system consoles. All worked fine. Till...
... Till one day when an ISOLATED GROUND UPS was installed, plugged
in and powered up... system console started a' locking a' up. Not
all at the same time; sometimes 3 at once, 2 at once; but eventually
(by the end of the day) all consoles would lock up.
- All machine's are running SINOS3.4; use standard null modem
cables and are slightly configured differently with various hardware
options (MCP, SCP, ALM, MEM, IPC, XY, XT, SD, ST, etc.)
Once a console is locked up, all input is inhibited; including a REAL
break. If the rs232 signals are dropped (via powering down the terminal or
disconnecting the cable) the CPU will branch to monitor mode. Killing the
getty or currently running process (csh) certainly solves the problem and
Silly Sally can log in properly.
It appears, but never verified, that when the ethernet becomes full
("ie: IF_SND FULL") the lock up sequence begins.
Someone mentioned rm'ng files in /dev; I can't belive that removing
dev-nodes could relate to this problem. As I'm most sure no one tries to
run suntools from the console; I'm also very sure that /dev/kbd,
/dev/mouse, /dev/fb, etc. aren't accessed (period) in the Boston sites
(they are tightly controlled production machines...)
It's my feeling that a strange ground loop or lack of proper ground is
what's causing the problem; after all, the only place our [console]
terminals lock up is where the UPS is. I really wish I knew more about
EE'ing, perhaps Tonto, myself and a case of silver bullets ;-) could solve
this one...
Ps: If anyone else that has similar console lockups [w/ Sun h/w] [is
reading this] and would mail me thier electric config info (MOTHER EARTH
vs UPS MAN), I'll be glad to summarize.
Joe Angelo -- Senior Systems Engineer/Systems Manager
at Teknekron Software Systems, Palo Alto 415-325-1025
uunet!tekbspa!joe -OR- tekbspa!joe@uunet.uu.net
------------------------------
Date: Wed, 8 Jun 88 11:32:34 BST
From: James Davenport <jhd%maths.bath.ac.uk@nss.cs.ucl.ac.uk>
Subject: Re: text table full
My views on this are unchanged since v5n69, when I wrote
Dan <@flash.bellcore.com:dan@wind> mentioned that the default for ntext
(24 + MAXUSERS) was probably inappropriate. It certainly is - I have
found that a Sun with this (and MAXUSERS = 4 or less) to be unusable as a
single-user system. Try running tbl | eqn | ptroff. My first hack was to
go to 24+MAXUSERS*2, which improved things, but didn't fix all the
problems. After analysing an "idle" system with one person logged in at
the console running suntools, I decided that 30+MAXUSERS*2 was a better
formula. Since then, no-one has complained (do they dare??).
I am not convinced that increasing MAXUSERS is correct, since that can
break licencing arrangements and will increase other tables that are
quite large enough. The problem is that the relationship between MAXUSERS
and ntext is wrong.
James Davenport jhd@uk.ac.bath.maths
------------------------------
Date: Tue, 7 Jun 88 17:13:57 EDT
From: bzs%bu-cs.bu.edu@bu-it.bu.edu (Barry Shein)
Subject: Sorcerer's Apprentice -- Multiple invocation of Daemons
You would think sendmail with -bd invoked twice would fail because it
shouldn't be able to bind the SMTP socket again, I've never tried it.
Same with other similar TCP daemons.
[[ That's what I think, too. But I've never tried, either. --wnl ]]
I guess if I were to write wrapper programs I would consider simply using
flock()* on the executable (being as that path has to be in the wrapper
anyhow, why not?)
-Barry Shein, Boston University
* You probably don't want lockf() as that would disallow any system using
this executable on its server from starting the daemon, probably not what
you want, so in this case the weaker semantics of flock() wins.
------------------------------
Date: Tue, 7 Jun 88 11:11:10 BST
From: Adrian Cockcroft <mcvax!camcon!anc@uunet.uu.net>
Subject: SunOS 4.0 doesn't work with A79J MC68881
Looking through the SunOS 4.0 Manuals (I dont have tapes yet), in
particular the addendum to the Floating Point Programmers Guide I
noticed that support for the old 12.5 MHz, buggy A79J revision of the
MC68881 has been removed. Use /usr/etc/mc68881version to find out
what you have, we have one old 3/160 with A79J here. If you have an
A93N revision then you are OK. As compensation SunOS 4.0 will
support the MC68882 which is faster but saves more state than the
MC68881 so won't work under SunOS 3.X. You will either have to get
SUN to do the upgrade for you or fix it yourself.
See Sun-Spots Digest v6n16 16 Feb 88 for details on DIY upgrades, its
pretty easy as long as you can get the old chip out! To change from
12.5 MHz to 16MHz move CPU jumper N-11 from 5-6 to 7-8.
Adrian Cockcroft anc@camcon.uucp ..!uunet!mcvax!ukc!camcon!anc
Cambridge Consultants Ltd, Science Park, Cambridge CB4 4DW,
England, UK (0223) 358855
------------------------------
Date: Wed, 8 Jun 88 13:29:19 EDT
From: Matt Landau <mlandau@diamond.bbn.com>
Subject: Using previously created colormap segments
SunView shared colormap segments seem designed to let a window use only
part of the hardware colormap, and to let cooperating windows share the
SAME part of the hardware colormap.
It therefore seems strange to me that setting a window's shared colormap
segment name (pw_setcmsname) causes the colormap segment attached to that
name to be reinitialized to NULL, requiring you to immediately call
pw_putcolormap to reset the colors.
I would like to be able to share a custom colormap segment among several
cooperating windows. Ideally, I'd want one of the windows to generate the
RGB triplets and install them in a shared colormap segment with a well
known name that's been agreed upon by the cooperating windows, then tell
each of the other windows to use this named colormap segment.
Unfortunately, each time you set a window's colormap segment name, the
colormap entries are cleared, resulting in flashing of the windows as each
one sets its name and colors.
One might suggest creating an internal atomic operation that consists of
setting all of the names first, then calling pw_putcolormap to set the
colors, and indeed that's my current approach. The problem here is that
this approach doesn't work very well if you want to CHANGE the colormap
segment (not just the entries, but the acutal size of the segment as well)
for all of the windows while the process is running, without having a lot
of flashing and redrawing.
It seems as though one should be able force a pixwin to use a previously
created colormap segment without having to respecify the colors. Am I
just missing something, or is this really not possible in SunView?
Matt Landau
mlandau@bbn.com
------------------------------
Date: 7 Jun 88 18:47:37 GMT
From: Sjoerd Mullender <mcvax!cs.vu.nl!sjoerd@uunet.uu.net>
Subject: why is my 3/60 so slow?
I recently switched over from a Sun 3/50 to a 3/60, both running SunOS
3.5. I have a shoebox with a 70 Meg disk which came with the 3/50 on
which I swap. The problem is that, although a 3/60 is faster than a 3/50,
swapping is much slower. On the 3/50 I had to wait a bit from time to
time but on the 3/60 I very often have to wait a minute or so before a
process reacts if it was swapped out. Does anyone know what the problem
might be and does anyone have a solution. Could the problem be a timing
problem in the SCSI driver? We have source, so solutions involving source
change are welcome.
Sjoerd Mullender
sjoerd@cs.vu.nl
When this doesn't work try sending via uunet.uu.net or mcvax.uucp.
------------------------------
Date: Tue, 7 Jun 88 12:37:33 EDT
From: psw@johann.arpa (P. S. Wherry)
Subject: subnet mask console message?
Every couple of days for a while now, a number of Suns on our Ethernet
have been displaying the message "Setting subnet mask to 0xffff0000" in
the console window. Nobody I've spoken to has any idea what is causing
the message to appear (though I suspect it's someone changing something on
a system elsewhere on our network). Could anyone fill me in on the exact
cause of this message's appearance? Thanks.
Phil Wherry
The MITRE Corporation
psw%wolfgang@gateway.mitre.org
(note that this address differs from the address in the header!)
------------------------------
Date: Tue, 7 Jun 88 16:27:05 edt
From: cbmvax!carolyn@uunet.uu.net (Carolyn Scheppner CATS)
Subject: Troff viewer?
We are looking for a WYSIWYG troff viewer for a Sun2/120 running Unix 4.2
release 3.0. We are also interested in comments on the various laser
printers which can be used for output (speed, reliability, supplies,
etc.).
The printer will be used to create camera-ready copy for several fairly
thick manuals which need to be updated and currently exist as troff files
on the Sun. In addition, the printer will be used for checking output
format unless we can get a troff viewer.
All information appreciated. Please email to:
[ihnp4,rutgers,uunet]!cbmvax!rains (Nancy Rains)
Thanks.
------------------------------
Date: Tue, 7 Jun 88 12:09:17 EDT
From: smb@research.att.com
Subject: DES driver?
Has anyone figured out how to use (or buy from Sun!) the DES hardware
option? I'm told that there's an empty socket just waiting for an AMD9518
DES chip, and my Sun rep tells me all I have to do is buy one and plug it
in. Other folks have said that a custom PAL is needed to control the
chip; our local rep denies any knowledge of this chip.
--Steve Bellovin
ulysses!smb
smb@research.att.com
------------------------------
Date: Wed, 8 Jun 88 10:44:34 EDT
From: jonathan%bert.mitre.org@gateway.mitre.org (Jonathan Leblang)
Subject: VT100 emulators?
Does anyone know of any public domain VT100 emulators for the SUN? I am
particularly interested in the source code for such an emulator. Please
mail responses (or code?) to me directly as I do not get the sun-spots
digest regularly. Any help would be appreciated. Thanks,
Jonathan Leblang
The MITRE Corporation
(703) 883-5761
jonathan@bert.mitre.org
[[ I think I need to start a list of the "most often asked questions on
Sun-Spots. This one would certainly be near the top. I am including some
old messages in this digest that should answer this question. Edward
Lafferty, <ell%linus@mitre-bedford.arpa> wrote one. There is also one
written by Bruce Barnett <barnett@crd.ge.com> in the sun-spots archives
(under "sun-source" in the files "vttool.shar.1" and "vttool.shar.2".
--wnl ]]
------------------------------
Date: Tue, 7 Jun 88 12:14:06 mdt
From: ziel@spot.colorado.edu (ZIEL FRED ADAM)
Subject: Questions about Sun systems
I'm currently reviewing a system we're on putting together, and I would
like to get some opinions/information from netland:
- Does SunOS support two (or more) Ethernet gateways connecting the same to
networks. I would appreciate any pointers on this. Our Sun rep says it
is possible but has yet to give us the details.
- Is anyone using WORM drives out there? We would like the ability to use
the worm as a random access device as opposed to a tape-like archive device.
Currently, we are looking at the system from Genesis Imaging Technologies.
Comments on this system or alternateive system would be appreciated.
- Is anyone aware of a PC Implementation of an NFS Server. As I understand it,
PC-NFS is client-only.
Thanks, hopefully if my Sun ever arrives, I can contribute more than questions.
Fred Ziel / ziel@spot.Colorado.EDU / These might not even be MY opinions.
uucp: scgvaxd!nbires!boulder!ziel usps: 1621 9th #7 / Boulder, CO 80302
------------------------------
Date: 7 Jun 88 23:30:00 EDT
From: "STEVE_CARR" <navmasso42a@nardacva.arpa>
Subject: Computer aided test procedure/case development
[[ This message is a reply to one posted in the software engineering news
list. He is wondering if anyone in the Sun world has seen something that
might help him... --wnl ]]
1. I am extremely interested in any response that you receive on this
subject. Please forward any direct responses to me. Our near term needs
are not as sophisticated as what you specified.
2. We have an environment wherein we must test contract deliverables.
This software consists mainly of business and logistics function AISs for
the Fleet for which we are the CDA (Central Design Agency). We receive
successive deliverables, and classically, delivery release "n" will be
tested for contracted functionality. When delivery release "n+1" arrives,
we have no time to manually retest all functionality which was certified
in delivery release "n". We expend all of our scarce functional expertise
painstakingly banging away at the keyboard testing new functionality
contracted for in delivery release "n+1". Yet we know from "Lehman's Laws
of Software" that errors will be injected into delivery release "n+1"
which will trash functionality which was validated in delivery release
"n".
3. We are further burdened by the fact that we must export software for
execution on 16 bit minicomputers of circa 1979 architecture. To say that
the hardware is overtaxed is putting it kindly. Therefore, we are not
interested in a testing tool which executes on the AIS target
architecture. Because of the target minicomputer hardware, we had been
thinking more along the lines of a Motorola 68020 or Intel 80386 driver
that would catalog test scenarios, and feed them into the 16 bit
minicomputers automatically. It would be nice if several terminals at
once could be emulated by the 32 bit microprocessor. We would like for
the 32 bit microprocessors to automatically capture results of individual
test transactions for perusal by the functional analysts after automatic
testing went to EOJ.
4. It seems to me we should be using relatively inexpensive yet powerful
1.5 to 5.0 MIP 32 bit engineering workstation microcomputers in a
multitasked UNIX environment. Together with the services of a true RDBMS
(Relational Data Base Management System) such as ORACLE, INFORMIX, INGRES,
or FOCUS, etc., we could build complex test cases, catalog them, and
retrieve them for modification without excruciating pain on the part of
the functional analyst. Personally, I don't see how such a tool could
ever be truly robust and user friendly to a functional analyst unless it
was RDBMS driven.
5. If anybody in the audience knows of such an environment which
approximates our needs, we would indeed be grateful for feedback.
Very Respectfully,
Steve Carr
LCDR, SC, USN
Navy Management Systems Support Office (Code 42A)
Naval Air Station
Norfolk, Virginia 23511-6694
Commercial: (804) 445-2171
AUTOVON: 565-2171
MILNET: navmasso42a@nardacva.arpa
carrs@nosc.mil
------------------------------
End of SUN-Spots Digest
***********************