Sun-Spots-Request@Rice.edu (William LeFebvre) (10/10/88)
SUN-SPOTS DIGEST Thursday, 6 October 1988 Volume 6 : Issue 250
Today's Topics:
Re: RPC failure in OS4.0 client boot
Re: Serial line H/W flow control under 3.5 and 4.0
Re: .cshrc vs .login
Re: gnuemacs keyboard bindings
Re: Interpreting ID PROM info
DEC LAT daemon for the SUN
Crashing problems on a 3/280 running 3.5
SunOS 4.0 /etc/printcap line for ALM-2 parallel port?
Formatting a CDC WrenIV/344?
Correspondence between raw disk devices and physical layout?
3.2 and 3.5 NFS incompatibilities?
Sun 3E (eurocard) compatible serial cards w/drivers for SunOs?
binding Fn keys?
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: Thu, 29 Sep 88 17:49:16 EDT
From: steinmetz!grymoire!barnett@uunet.uu.net (Bruce G. Barnett)
Subject: Re: RPC failure in OS4.0 client boot
>Diskless clients cannot boot
>whoami RPC call failed with status 10
I have been SCREAMING at Sun for two months about this problem. I still
don't have a solution.
As far as I can tell, when a 4.0 client boots up, it does a broadcast RPC,
using UDP, requesting hostname, domain and route to the server. The
broadcast is network wide (no subnet mask), so that it can go thru
subnets.
Some systems wrongly respond to this with a ICMP: "destination
unreachable". Possible machines that might(?) cause this problem are
Ultrix 2.3 machines, or perhaps VMS machines with either Wollongong or the
Ultrix Bridge product. (I hear DEC has a patch for Ultrix Bridge).
I may not quite understand the problem. But we have 800 machines on our
ethernet, and can't get patches to every single implementation.
We plan to install 90 new Suns soon. These will be SunOS 4.0 machines, but
we are unable to install them until we get a fix from Sun.
If you are contemplating upgrading to 4.0, you may want to think twice.
--
Bruce G. Barnett <barnett@ge-crd.ARPA> <barnett@steinmetz.UUCP>
uunet!steinmetz!barnett
------------------------------
Date: Mon, 03 Oct 88 12:30:03 +1000
From: Craig Bishop <munnari!lupus.cc.deakin.oz.au!craig@uunet.uu.net>
Subject: Re: Serial line H/W flow control under 3.5 and 4.0
This is mail I just sent to a colleage in New Zealand, I think it will go
part way to solving your problems. We tried the MDMBUF stuff but it never
seemed to work right in SunOS 4.0.
Craig
------- Forwarded Message From: Craig Bishop <craig>
Firstly we made the same mistake you have the standard sun ports are not
RS232 they are RS432 (I think, I'd have to check with our engineer to be
positive about the number).
Yes it can be done, and it took a bit of heart ache.
We didn't however use DTR we used RTS CTS. To do the following you will
need source to the printer software you are using. And also be using SunOS
4.0. Sorry if this is not help.
You have to change the all the IOCTL's to use termio instead of sgttys.
Once you have done that, add to the termios (note the "s" in termio, there
is a new termio structure termios) the flag CRTSCTS, this enables RTS CTS
handshaking on the line. Note also that the line is NOT configured as a
modem port.
Finally if you don't want the printer to be a bit bucket when it is turned
off, the following wiring should be used,
3/60
+-------+
| 2 | -------------------------- Data
+-------+
| 5 | ---------*---------------- Hardware flow control
+-------+ |
| 7 | ---------|---------------- Ground
+-------+ /
| 8 | ---+ \
+-------+ | / 4.7 K Resistor
| 20 | ---+ \
+-------+ |
| 25 |----------+
+-------+
Lots of luck.
Craig Bishop ARPA: craig%lupus.cc.deakin.oz.au@uunet.uu.net
UUCP: ...!uunet!munnari!lupus.cc.deakin.oz!craig
------- End of Forwarded Message
------------------------------
Date: Mon, 3 Oct 88 08:00:13 EDT
From: Bennett Todd <bent!bet@mcnc.org>
Subject: Re: .cshrc vs .login
Your .login file serves two purposes. First, it should set up all aspects
of your environment (including in particular true environment variables)
which will be properly inherited by child shells. This includes your PATH,
TERM, (and, if appropriate, TERMCAP) selections, as well as such
program-specific customizations as PAGER, EDITOR, VISUAL, and MANPATH. The
other purpose is to perform any actions which you want to take place
whenever you log in, such as running fortune, and maybe calendar.
The .cshrc file is for one purpose only: to set csh(1) internal variables
to customize the shell environment. These are variables which are *not*
automatically exported and imported to and from the environment, so there
is no way for them to carry into the child shells. Examples of these
variables include prompt, history, time, filec, nobeep, and fignore, as
well as any csh aliases you might use. Since setting the prompt variable
in particular will break any programs that run a child csh for globbing or
whatever, it is considered wise to start off .cshrc with the following
incantation:
if ($?prompt == 0) then
exit
endif
This will cause non-interactive shells to exit .cshrc immediately, which
will both prevent programs from breaking and also speed them up.
In general (there are *always* exceptions...) if you aren't sure where
things go you could try the following arrangement.
In .login:
# environment variable setting:
set path=...
set term=...
setenv TERMCAP ...
setenv EDITOR ...
setenv VISUAL ...
setenv PAGER ...
setenv MANPATH ...
# any other environment variables...
# and run startup programs:
fortune
calendar
and in .cshrc:
if ($?prompt == 0) then
exit
endif
# Whatever csh internals you happen to like, such as:
set prompt=`hostname`\[!\]\
set history=50
set time=(30 "Time -- R:%E S:%S U:%U Mem -- R:%KK F:%F S:%W")
set filec=1
set nobeep=1
set fignore=(.o .out)
# and you choice of aliases:
alias l ls -F
Which is to say, .login gets setenv commands and program invocations, and
.cshrc gets the "if ($?prompt == 0) ..." starter followed by set commands
and alias commands. Note that the "path" and "term" variables are special
ones to csh -- unlike the others, csh automatically exports these to the
environment whenever they are set, and imports them on startup
bet@orion.mc.duke.edu (Bennett Todd)
------------------------------
Date: 3 Oct 88 14:11:04 GMT
From: Wayne Mesard <mesard@bbn.com>
Subject: Re: gnuemacs keyboard bindings
Reference: v6n245
P.C. Sutton says:
> Andrew Gerber asks:
>> How would one set up a .emacs file to bind the sun keyboard arrow keys
>> ([C [A [B [D) to move the cursor around in Emacs?
>
> This is very simple. The routine below does this. You can put this in your
> .emacs file,
[...]
> (global-unset-key "\M-[")
> (define-key global-map "\M-[A" 'previous-line)
[...]
It's even easier than that. Many GNU Emacs inventions turn out to be
reinventions due to the current [non-]existent state of the ELisp manual
[vapor-docs?], and a not-liberal-enough use of M-x apropos. Try:
(setq sun-esc-bracket t)
This causes R7 through R15 to be bound to functions similar to the PC
number pad.
--
void *Wayne_Mesard(); MESARD@BBN.COM BBN, Cambridge, MA
GNU Emacs--it's in der.
------------------------------
Date: Wed, 5 Oct 88 00:43:01 EDT
From: attcan!utzoo!henry@uunet.uu.net
Subject: Re: Interpreting ID PROM info
>On my machine, hostid returns 12005496. According to your article, it's
>telling me that my serial number is 5496. The serial number on the case
>of my machine is 725E2616....What's the deal?
The system serial number and the CPU serial number are unrelated. Hostid
is giving you the CPU serial number, which is what things like licensed
software packages want. The label on the case is giving you the system
serial number, which is what maintenance contracts and such want. Aren't
Suns wonderful?
Henry Spencer at U of Toronto Zoology
uunet!attcan!utzoo!henry henry@zoo.toronto.edu
------------------------------
Date: Mon, 3 Oct 88 11:05:34 EDT
From: jas@proteon.com (John A. Shriver)
Subject: DEC LAT daemon for the SUN
LAT is a proprietary (read: trade secret, unpublished) protocol of DEC.
Until recently, they would not consider letting anyone else touch it under
any conditions. Now, they have started to license the source, but I would
expect this is on a controlled and non-competitive basis. (Eg TIU vendors
are getting it, but I suspect Sun never would.) I have no idea whether
DEC would sue you for reverse-engineering LAT, the way they have with
MSCP, VAXBI, et. al.
LAT is also a VERY limited protocol. No network layer (not for any good
reason, it was just designed by a group of total novices). The
"transport" timers are fixed. It assumes bandwidth is very cheap. The
only clevernesses are (1) it is designed to minimize the packet-processing
load at the timesharing hosts, and (2) clever naming and fault-tolerance.
All of these could have been acheived inside the DECnet framework, but the
LAT designers were not architects.
The solution is to follow standards: IP, TCP, and Telnet. Ignore LAT.
Start buying TCP/IP TIU's, and put TCP/IP on the VAXes.
At least Sun is keeping their network protocols open, unlike modern DEC,
with proprietary protocols like LAT and LAVC.
------------------------------
Date: Mon, 3 Oct 88 9:48:57 MDT
From: liz@cgdra.ucar.edu (Liz Coolbaugh)
Subject: Crashing problems on a 3/280 running 3.5
I put a call into Sun's software services last Wednesday about the
following problem and have so far not even received a call back from the
engineer. This problem has only crashed our system once, but I would
still like to have some idea of what caused it and what I might be able to
do to keep it from crashing :-).
The system in question is a Sun 3/280. I recently upgraded the machine to
3.5 and within a week, the machine crashed with a "panic: Bus error"
message. The following is an excerpt from the messages file:
nfsd:
trap address 0x8, pid 89, pc = f038972, sr = 2004, stkfmt b, context 0
Bus Error Reg 80<INVALID>
data fault address 0 faultc 0 faultb 0 dfault 1 rw 0 size 0 fcode 5
KERNEL MODE
page map 20000000
D0-D7 0 3f 26 0 0 0 0 22
A0-A7 0 f187ecc 0 f07d9c0 f094a30 f338000 ffff85da ffff85b6
Begin traceback...fp = ffff85da, sp = ffff85b6
Called from f03a698, fp=ffff85fa, args=304 f338000 11f8 effffd4
Called from f00f668, fp=ffff861a, args=f1874d8 ffff8616 f3c3dec f187944
Called from f00e574, fp=ffff8686, args=f3c3de4 22 f187944 0
Called from f00f3c2, fp=ffff86ca, args=f3c3de4 f187e64 ffff86ea 2
Called from f015ace, fp=ffff874a, args=ffff86ea f187944 f06f572 ffff87b6
Called from f015bb0, fp=ffff875a, args=f187944 2000 ffff8782 f00df38
Called from f00df38, fp=ffff8782, args=f187944 0 3 f187944
Called from f048e78, fp=ffff87aa, args=ffff886a effffe0 0 0
Called from f0044cc, fp=effffbc, args=9b 0 1 2
End traceback...
Has anyone seen this problem? Is it, by chance, a known problem with 3.5?
Any information you can provide me will be appreciated.
Liz Coolbaugh
National Center for Atmospheric Research
(303) 497 1327
liz@cgdra.ucar.edu
..ncar!cgdra!liz
------------------------------
Date: Sun, 2 Oct 88 13:36:34 EDT
From: elsie!ado@ncifcrf.gov (Arthur David Olson)
Subject: SunOS 4.0 /etc/printcap line for ALM-2 parallel port?
A Sun 3/280S (running SunOS 4.0) crashes when I try to print
something after having put a
lp:lp=/dev/mcpp0:
line in its /etc/printcap file in an attempt to use its ALM-2's parallel port.
Clearly the above line is the wrong thing to use; what's the right thing?
--
ado@ncifcrf.gov ADO is a trademark of Ampex.
------------------------------
Date: Sun, 2 Oct 88 18:23:29 EDT
From: lj@spdcc.com (Len Jacobs)
Subject: Formatting a CDC WrenIV/344?
Has anyone successfully formatted a CDC WrenIV/344? I am interested in
knowing how you set the cylinders & sectors for maximum utilization of the
drive. Please mail me your results & I will summarize if the results are
useful and meaningful.
Len Jacobs
------------------------------
Date: 3 Oct 88 09:56:24 GMT
From: "David.J.Ferbrache" <mcvax!cs.hw.ac.uk!davidf@uunet.uu.net>
Subject: Correspondence between raw disk devices and physical layout?
At Heriot-Watt we are working on a hardware search system which requires
rapid access to a partition held on disk. The current configuration we are
utilising is a SUN 3/160 with a Micropolis SCSI disk.
The data rates achievable over the standard unix file system are in the
region of 200-250 K bytes/sec. Initial timings over raw and buffered disk
partitions seem to indicate data rates in the region of 350Kbytes/sec for
raw disk (block size 16K) and 90K buffered disk.
The above timings over raw and buffered disks employ random layout
strategies but indications are that modification of layout strategy can
cause significant changes in data transfer rates.
I have two main questions which I hope can be answered:
1. What is the mapping between raw disk partition offset and the physical
sector and track offsets on the disk.
2. What is the optimal layout strategy for blocks on disk, taking into
account rotational positioning and cylinder layout.
Thanks in advance.
------------------------------
Date: Sun, 2 Oct 88 15:24:20 EDT
From: aad@stepstone.com (Anthony A. Datri)
Subject: 3.2 and 3.5 NFS incompatibilities?
The scenario: a 3/180 server running 3.2, lots of 2/50, 2/120, 3/50
clients, with an odd apollo and HP thrown in for godd measure. The
villian: a 3/50 running 3.5 (prepackaged from Artecon). When I first
installed the 3.5 machine, I found that it took *forever* to access (read,
lookup, write...) files on nfs mounts. Logging in with a path with a
remote /usr/local in it took about 10 minutes, generating two sets of "nfs
server <3/180> not responding" and "nfs sever <3/180> ok" messages. I
hacked the fstab for the nfs mounts to have a higher timeo and even
rsize=2048, as we have to do on some of the sun2's because of speed, and
the error messages went away and the user of the machine never complained
about it. Now, when I moved the machine, it's magically back again.
Packets seem to get in and out okay -- telnet and ftp work quickly enough
--- but remote NFS mounts are slowed incredibly, and this 3.5 3/50's /usr
mounted from other machines gets a somewhat lesser slowdown.
Am I alone in experiencing this? I'm ready to back the machine down to
3.2 and see what happens.
Anthony A. Datri,SysAdmin,StepstoneCorporation,stpstn!aad
------------------------------
Date: 1 Oct 88 19:29:03 GMT
From: well!ejf@lll-crg.llnl.gov (Erik James Freed)
Subject: Sun 3E (eurocard) compatible serial cards w/drivers for SunOs?
I am looking to add some RS232 serial ports to a Sun 3E 6U vme card set
based system. Does anybody have any pointers to or direct experience with
solutions to this problem? I have heard that that Systech and Xylogics
both have SunOs drivers for their serial cards. Do they work and how good
are they? I would be indebted to anyone who could help me with this.
Thanks in advance,
Erik Freed
UUCP:{hplabs,pacbel,apple,cogsci,qantel}!well!ejf
415-461-5400
------------------------------
Date: Mon, 3 Oct 1988 10:36-EDT
From: Hans.Tallis@ml.ri.cmu.edu
Subject: binding Fn keys?
I'm having trouble binding the function (F1, F2, ...) keys on my Sun 3/60;
the man pages aren't too helpful. I just want to bind one to "(", one to
"-", and so on. Could someone mail/post a bit of code/shell script that
does the trick?
Many thanks, --Hans
P.S. If it makes a difference, I'm running Mach/X10.
[[ If your Sun is not running SunOS, I doubt that very many readers here
will be able to help you. --wnl ]]
------------------------------
End of SUN-Spots Digest
***********************