Sun-Spots Digest, v6n24
William LeFebvre
Sun-Spots-Request at RICE.EDU
Wed Mar 2 09:19:40 AEST 1988
SUN-SPOTS DIGEST Tuesday, 1 March 1988 Volume 6 : Issue 24
Today's Topics:
Fig is now in the archives
Re: single-user boot: the final word
Re: make question
Re: Disk Space Limitations For 3/260, 3/280
More info on 68881 (chip masks)
Using the Interphase Cheetah with the Hitachi Gig Disk
Swap space problems
3.5 ifconfig on SUN 2 problem
More than 2 ethernet boards in a fileserver?
Apollo vs. Sun?
Non FTP sources of PD software?
RSCS software for the SUN OS?
clearing a color screen?
sun-4 as a terminal server?
Determining display type from a shell file?
Sun 4/280 power?
SunCore trashing Unix environment?
Send contributions to: sun-spots at rice.edu
Send subscription add/delete requests to: sun-spots-request at rice.edu
Bitnet readers can subscribe directly with the CMS command:
TELL LISTSERV AT RICE SUBSCRIBE SUNSPOTS My Full Name
Recent backissues are stored on "titan.rice.edu". For volume X, issue Y,
"get sun-spots/vXnY". They are also accessible through the archive
server: mail the word "help" to "archive-server at rice.edu".
----------------------------------------------------------------------
Date: Tue, 1 Mar 88 17:00:15 CST
From: William LeFebvre <phil at Rice.edu>
Subject: Fig is now in the archives
Okay. I finally did it. "Fig" is now in the sun-source archives,
packaged in eight shar files. They are named "sun-source/fig.shar.0X"
(with the X taking on appropriate values) and range in size from 45172 to
46869 bytes. Archive server users can get them with a request that looks
like "send sun-source fig.shar.0X", but remember that the server bundles
all your requests together. So if you ask for all eight parts
simultaneously it will put all of them into one return mail message! For
more information about the archive server, send a mail message containing
the word "help" to the address "archive-server at rice.edu". Those with FTP
access can retrieve them from the host "titan.rice.edu". This is not the
new version that was promised to me a few weeks back. When that becomes
available, I will update the archive's copy.
William LeFebvre
------------------------------
Date: Mon, 22 Feb 88 15:36:12 EST
From: Bruce Barnett <barnett at ge-crd.arpa>
Subject: Re: single-user boot: the final word
In the past I have suggested that people add the line
login root
to /.profile to prevent undesirables from booting up the system un
single-user mode.
DO NOT DO THIS!
What happens is that the system could crash, and fsck cannot fix up the
disk automatically. It then stays in single user mode and executes
/.profile.
This will execute login, which will then prompt for a password. But if
the user doesn't type in the password, login terminates and the system
procedes to multi-user mode WITHOUT REPAIRING THE DISK!
The following program example can be used to solve both problems.
Compile it and modify root's /.profile to be something like:
__________
stty dec
PATH=/etc:/usr/etc:/usr/ucb:/bin:/usr/bin:
export PATH TERM
/lockup
---------------lockup.c------------------------------------
/* example written by Bruce G. Barnett <barnett at ge-crd.arpa> */
#include <stdio.h>
#include <signal.h>
#include <pwd.h>
struct passwd *pwd;
struct passwd *getpwuid();
char *strcpy();
char *crypt();
char *getpass();
char *pw;
char pwbuf[10];
main()
{
signal(SIGINT, SIG_IGN);
signal(SIGQUIT, SIG_IGN);
signal(SIGTSTP, SIG_IGN);
/* get the password entry for root */
/* use 0 if you want to hard-wire the passwd for root */
/* else use getuid() */
pwd=getpwuid(getuid());
if (pwd == NULL )
(void) fprintf(stderr,"Cannot get password entry for root");
while ( 1) { /* forever */
(void) strcpy(pwbuf,getpass("Password:"));
pw = crypt(pwbuf, pwd->pw_passwd);
if (strcmp(pw,pwd->pw_passwd) == 0 ) {
return(0);
}
}
}
__________
This program is larger that necessary, but can be used in multi-user mode
in addition to single user mode. The password is determined by the entry
in the password file.
Bruce G. Barnett <barnett at ge-crd.ARPA> <barnett at steinmetz.UUCP>
uunet!steinmetz!barnett
------------------------------
Date: Tue, 23 Feb 88 10:16:41 -0500 (EST)
From: Bradley White <bww at cs.cmu.edu>
Subject: Re: make question
> Make (on a Sun 3-50/SunOS 3.2) produces the following diagnostic:
>
> $! nulled, predecessor circle
>...
The following is from "An Augmented Version of Make" by E. G. Bradford,
which describes the Columbus version.
`There is an interesting "hidden" variable in this version of make: "$!".
It represents the current predecessor tree. In the following makefile:
all: cat
@echo cat up-to-date
cat: cat.c
echo $!
when the "echo $!" is executed, "$!" evaluates to
cat.c cat all
which is not all that useful! Further, it occasionally prints a message
$! nulled, predecessor circle
This message means that the predecessors of a file are circular. The
actual evaluation of the "$!" macro was aborted, and its value set to
null. Otherwise there is no effect.'
The BSD version, being "pre-Columbus" based, does not include this
"feature".
Bradley White <bww at cs.cmu.edu> +1-412-268-3060
CMU Computer Science Department 40 26'33"N 79 56'48"W
------------------------------
Date: Tue, 23 Feb 88 20:28:13 EST
From: citi!dwon!lokkur!scs at rutgers.edu (Steve Simmons)
Subject: Re: Disk Space Limitations For 3/260, 3/280
In SunSpots Digest Vol. 6n17 mutchler at sun.com (Dan Mutchler) writes
>The difference between the 3/260 and 3/280 disk space is very simple. The
>3/260 is intended to be configured with pedestal SMD drives....The 3/280 is
>rack mountable and can use the Super Eagles (575MB each) with the same
>controller limit of four drives. This gives the limit of 2GB.
"Intended to be figured"? Translates to "that's how we sell it" in this
case.
Sorry to dispute you Dan, but we run 3/260s with dual or quad double
eagles with no problems. The only difference between the 260 and 280 is
the backplane size and the physical packaging. Electrically they are
identical. So why does Sun say different size limitations? Marketing, I
think. If you have a 3/260 and a disk rack, buy yourself two controllers
and four double eagles and have a good time.
In fact, of our 16 file servers 15 are either 3/160s or 3/260s supporting
the same drives as a 3/180 or 3/280.
Steve Simmons
Schlumberger CAD/CAM Div.
Ann Arbor, MI.
------------------------------
Date: 23 Feb 88 15:26:21 GMT
From: roy%phri at uunet.uu.net (Roy Smith)
Subject: More info on 68881 (chip masks)
It seems I spoke too soon last time, and slightly messed up the
information about the 68881 chip masks. Here is the (I hope) real scoop.
Sun's Floating Point Programmer's Guide (Appendix F) has the details of
how a A93N and A79J 68881 chip behave. There is apparantly one remaining
bug in the A93N mask, and there is a new mask, called B81G, which fixes
that. There is also a 68882 which is a higher performance 68881, and
which, if I understood the Motorola rep properly, is fuctionally the same
as a B81G 68881 (but faster). Does this mean you can drop a 882 into a
3/50? I think so, but I'm not sure. The point is moot, at least for a
while, as I don't believe 882's are shipping yet. My source hints that
they will be sampling soon, but I don't know how long it will be until
consumers can buy them Q1. On the other hand, people like Radius (maker
of 68020 add-on boards for Macintoshes) are already offering both 881 and
882 options.
Also, various people have accused me of being paranoid about taking static
precautions, and have said that they have not had any problems installing
68881's without any special anti-static precautions. They may be right,
but when working on a board that will cost me $1300 and a month of down
time to get repaired(*), I don't mind being paranoid. Also, New York in
the winter is dry, dry, dry. Low humidity means lots of static. If you
live where it is more humid, you will have less to worry about.
(*) 1,300 freaking dollars to change a $#%%@! 256k RAM chip? Give me a
break! Why won't Sun make the address/snydrome to chip location map
public so we can fix RAM failures ourselves? Open systems indeed! If
Volkswagon will sell me the parts, tools, and manuals needed to strip and
rebuild my car, why won't Sun give (or sell) me the 1-page chart I need to
do the simpliest and most common repair on my computer?
Roy Smith, {allegra,cmcl2,philabs}!phri!roy
System Administrator, Public Health Research Institute
455 First Avenue, New York, NY 10016
[[ William's second rule of system management: you can never be too
careful. Not taking precautions to eliminate static buildup is taking a
chance. You might get away with it (especially in Houston), but then
again you might not. --wnl ]]
------------------------------
Date: Tue, 23 Feb 88 08:40 EST
From: <STELLABO at CSHLAB.BITNET>
Subject: Using the Interphase Cheetah with the Hitachi Gig Disk
After months of tinkering we have finally installed the new Hitachi DK-815
gig disk in a SUN 3/280s using the Interphase Cheetah. The effort was
well worth it. The formatted disk capacity is around 870mb. The
performance is slightly better than the Cheetah with the Fujitsu Super
Eagle that we were using during our testing. The staff at Interphase was
very supportive often spending hours on the phone with us trying different
switch settings and diag options. If anyone is attempting to install the
same setup we would be happy to give you all the jumper settings and diag
options.
Problems still remain... Interphase's diag program does not know how to
slip sectors. This creates 2 problems .. first you must allow enough altn
cylinders to map all the bad sectors - this can substantially reduce the
formatted disk space. 2nd your disk must have enough defect free sectors
at the end of the disk to map to. Interphase claims they have a new diag
program to correct this problem.
This installation has been working error free for about 2 weeks now. I
have been doing fsck's on a daily basis with no errors. The Hitachi and
the cheetah are installed in our only file server machine; serving 13
diskless clients with an average user load of 15 - 20 users.
Again I would be happy to help anyone with any questions about this
installation.
Fred J. Stellabotte - Computer Systems Manager
Cold Spring Harbor Laboratory
Cold Spring Harbor, NY 11724
stellabo at cshlab
(516)367-8420
------------------------------
Date: Tue, 23 Feb 88 17:03:37 PST
From: blia!blipyramid!mike at cgl.ucsf.edu (Mike Ubell)
Subject: Swap space problems
We run a program that uses about 12meg of swap space. Often it runs out
of memory when pstat -s shows that there is in execess of 20 meg
available. pstat also says that the max process allocable is much less
than 20 meg. Does anyone know why this happens and how to avoid and or
fix it? (You can save your typing on the obvious "add more swap space"
answer. I'd like to avoid doing that.)
------------------------------
Date: 23 Feb 88 17:29:27 GMT
From: cyrus at hi.unm.edu (Tait Cyrus)
Subject: 3.5 ifconfig on SUN 2 problem
We have a SUN 2 that SunOS 3.5 has just been installed on. We are having
trouble ifconfiging the ethernet with the correct netmask.
Background:
Class B address: 129.24
subnet mask: 255.255.248.0
SUN 2 is on subnet: 129.24.13
Symtoms:
When we try:
ifconfig ec0 chama netmask 255.255.248.0 broadcast 129.24.15.255 -trailers up
followed immediately by:
ifconfig ec0
we see
ec0: 129.24.13.12 netmask 255.255.255.255 .....etc
if I set the netmask to 255.255.247.0 followed by an ifconfig ec0 I see
the netmask correctly set to 255.255.247.0
if I set the netmask to 255.255.249.0 followed by an ifconfig ec0 I see
the netmask correctly set to 255.255.249.0
But when I try to use 255.255.248.0 it gets screwed up. As a result of
this, we currently have the netmask set (incorrectly) to 255.255.0.0
which results is some strange arp flurries (not quite a storm yet :-)
Does anyone know of a solution? We had the same problem while running
SunOS 3.4 and had hoped that 3.5 would fix things.
NOTE:
-Our SUN 3's, running 3.4, don't have this problem (haven't upgraded
them to 3.5 yet).
-Our SUN 2's are not on maintenance because it costs too much
(SUN are you listening? :-)
Thanks in advance for any help
W. Tait Cyrus (505) 277-0806
University of New Mexico
Dept of Electircal & Computer Engineering
Parallel Processing Research Group (PPRG)
UNM/LANL Hypercube Project
Albuquerque, New Mexico 87131
e-mail:
cyrus at hc.dspo.gov
------------------------------
Date: Fri, 19 Feb 88 17:38:27 CST
From: eta!como!mondo!jsalmi at uxc.cso.uiuc.edu (John Salmi)
Subject: More than 2 ethernet boards in a fileserver?
I have a need to install additional ethernet boards in my fileservers, ie,
three or four boards. I have heard rumors of machines with up to 6
boards, but noone seems to have the definitive answer. ALso, Sun says that
they "don't support" more than 2 boards.
I am guessing that the multibus adapter switches need to be tweaked, as
well as changing the address/interrupt.
Does anyone have firsthand knowledge of this task? If so, please drop me
a line. This seems like it may be a common problem, so I am failry sure
that there is a solution out there somewhere.
Thanks, as always, in advance!
John Salmi Email : jsalmi at mondo.lake.eta.com
Software Engineer Voice : (612) 642-3012
ETA Systems, Inc. SnailMail : 1450 energy Park Drive
St. Paul, Mn. 55108
------------------------------
Date: 22 Feb 88 17:04:09 GMT
From: freeman-andrew at cs.yale.edu (Andrew C. Freeman)
Subject: Apollo vs. Sun?
We are looking into purchasing a number of workstations for real-time
financial applications and are interested in opinions regarding the choice
between Sun and Apollo (all models of hardware, but mostly benefits and
drawbacks to system software). Among the major topics we would like to
know about are the distributed computing abilities of their respective
networks, availability of third-party software, vendor commitment to user
development issues, pure computing power, and quality of development
tools. Please reply by net mail.
Thanks in advance,
Andrew
------------------------------
Date: Tue, 23 Feb 1988 14:51:43 LCL
From: Gareth J. Barker <GJBARKER at UFFSC.BITNET>
Subject: Non FTP sources of PD software?
>From time to time various interesting/useful pieces of PD software have
been mentioned in Sun_spots Digest, most of which are available by FTP
from various archives. As I don't have FTP access to anywhere I'm looking
for ways to get hold of some of these through mail and/or archive server
file server programs.
I'm currently looking for:
Dsun - a troff previewer for the Sun
C++ - the C++ compiler mentioned by Michael Strong in v6n16
A complete VT100 emulator (including function keys). I know about
vtem/vttool (v6n12) on the SUG tape. They would be great, but I can't
afford even the fairly nominal fee the user group wants.
FIG - 'Macpaint for the Sun'
If anyone can send me copies of these, or give me directions to archives
accessible through BITNET, I'd be greatful.
Thanks a lot,
Gareth J. Barker,
Dept. Radiology, Phone : (904) 392-3087
College of Medicine,
Box J-374,
JHMHC,
University of Florida,
Gainesville,
FL 32610. Bitnet : GJBARKER at UFFSC
"All views expressed are my own, etc. etc."
[[ See my announcement about "fig" at the beginning of this issue. --wnl ]]
------------------------------
Date: Tue, 23 Feb 88 15:42:11 -0500
From: randy at ncifcrf.gov
Subject: RSCS software for the SUN OS?
I was wondering if anyone knew of any software available for the SUN to
drive the RCSC communications protocol used for Bitnet? I did a quick
scan through my SUN product list and Catalyst and didn't find anything.
Final goal is, of course, connecting a Sun 3/280s up to Bitnet. Any and
all answers appreciated. If there is interest I'll post a summary.
-- Randy
Randy Smith @ NCI Supercomputer Facility
c/o PRI, Inc. Phone: (301) 698-5660
PO Box B, Bldng. 430 Uucp: ...!uunet!ncifcrf.gov!randy
Frederick, MD 21701 Arpa: randy at ncifcrf.gov
------------------------------
Date: Tue, 23 Feb 88 17:23:17 EST
From: Ned Danieley <ndd at sunbar.mc.duke.edu>
Subject: clearing a color screen?
We recently purchased a 3/110 and two 3/60s, and I'm having some trouble
understanding how they work. If I start up a program like stringart on one
of the color machines, and then use, say, wall(1) to write to the screen,
whatever message I send stays on the display even when stringart redraws.
I understand that the color planes are separate from the monochrome plane,
but I can't figure out how to clear the mono. I've mucked around with
stringart with no success. Can anyone tell me what's happening?
(The reason I ask is that we use stringart as a screen saver, and usually
in the morning the color tubes will have system messages on them, with
stringart running merrily behind them.)
Ned Danieley (ndd at sunbar.mc.duke.edu)
Basic Arrhythmia Laboratory
Duke University Medical Center
Durham, NC 27710
(919) 684-6807 or 684-6942
------------------------------
Date: Tue, 23 Feb 88 16:02:53 EST
From: Jack V. Briner <jvb at cs.duke.edu>
Subject: sun-4 as a terminal server?
We are currently considering selling our large vax and using the money to
purchase some sun-4's. The major concern is whether the sun-4's async
ports and pseudo ttys will be able to handle many users. Our vax 8600 has
anywhere from 40-60 people on it mostly doing light tasks, such as reading
news, mail and editing text files. Most of these people are idle. We
would like to replace the machine with 3 sun-4's with ~20 people per
machine. The sun-4 is certainly fast, many of our tools see 5X over
sun-3; however, how gracefully does it handle the many context switches
that are required for an interactive machine with 20 users. Does anyone
use the sun-4 as a multiuser machine? (not just as a server or a compute
engine) How many people do you have on your machine via rsh/telnet and
via ttys? What do they do on the machine? Do you also use the machine as
a server? How many clients?
Thank you for your replies to jvb at cs.duke.edu. I'll summarize the replies
and tell you what we decide to do in a follow up article.
------------------------------
Date: Tue, 23 Feb 88 13:40:15 CST
From: kdye at como.lake.eta.com (Ken Dye)
Subject: Determining display type from a shell file?
Just looking for an easy way for a shell-script to determine what type of
display is in use (a 3/50 or 3/60). Any hints?
--Ken ..ihnp4!laidbak!eta!como!kdye
------------------------------
Date: 23 Feb 88 02:26:41 GMT
From: wucs1!br at uunet.uu.net (Bill Ross)
Subject: Sun 4/280 power?
So, all you Sun 4 owners out there, how much juice is in a Sun 4/280? Sun
claims that the machine is rated at 10 MIPS although I have heard that
this is somewhat inflated. I'm looking to buy some of these beasties to
replace my old flock of VAX 750's. I can run about 15 users doing C
compiles, nroff, macsyma, spice, ctrlc &etc on a 750 before it gets
unbearably slow. While the 750 is only rated at about 1 MIP because of
I/O considerations there is no way I can just scale this up to arrive at a
figure of 150 users on a 4/280. What is a good figure? How many users do
you support on your machine (25? 30?) and what type of jobs are they
running? Any and all insights welcome.
BR
Bill Ross Washington University ECL, St. Louis 314-889-5894
UUCP: br at wucs1.UUCP or ..!{ihnp4,uunet}!wucs1!br
ARPANET: wucs1!br at uunet.ARPA
CSNET: wucs1!br at uunet.ARPA%csnet-relay
------------------------------
Date: 22 Feb 1988 22:51-EST
From: Hans.Tallis at ml.ri.cmu.edu
Subject: SunCore trashing Unix environment?
I have a program which in which I do several system() calls to do things
like lpr a file it has written. After I intialize_core, however, a call
like system("lpr foo.PS") fails because unix can't find lpr (or anything
else not specified as an absolute pathname), as though the SunCore
initialization trashes the process's environment.
[[ It wouldn't necessarily be the entire environment -- just the "PATH"
variable. But that alone is bad enough. --wnl ]]
Has anyone else seen this, or perhaps even a fix for it?
Any pointers welcome, --Hans
------------------------------
End of SUN-Spots Digest
***********************
More information about the Comp.sys.sun
mailing list