Sun-Spots-Request@Rice.edu (William LeFebvre) (10/25/88)
SUN-SPOTS DIGEST Monday, 24 October 1988 Volume 6 : Issue 272
Today's Topics:
Re: nd question
Re: Hard vs Soft Mounts
Re: Encoding ASCII Files
Re: ispell and "emacs"
Re: Line drawing programs
Re: SunOS 4.0 Suntools speedup
Proxyarpd: release 2
SUN PCNFS & Micom NI5210
TAAC mailing list
Dist'n format for the 386i: tape vs. disk
SunOS 4.0 netstat brokenness
Problems with pc/nfs + IBM worm disk
88k for Sun 3/4 any news?
What is autocall?
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: Wed, 19 Oct 88 14:53:12 PDT
From: pixar!r2d2!dv@sun.com
Subject: Re: nd question
> From: Bennett Todd <bet@bent.mc.duke.edu>
>
> As far as I know, under SunOS 3.x there is no way to overlap use of the
> disk between nd and normal partitions. You will have to repartition the
> disk.
No, you don't have to repartition. While you can't have a normal (non-nd)
filesystem anywhere but the beginning of a partition (so it wouldn't be
possible to have nd stuff at the beginning, and the non-nd stuff
afterwards), there's nothing saying that the normal filesystem has to take
up the whole partition. If you use newfs to create the filesystem, then
it will, as newfs figures out how big the partition is, and uses that as
the size. However, if you use mkfs, and give it a size smaller than the
partition, you can use the remainder of the partition for nd.
This is in fact what I did here (a Sun2/170 running 3.4, with 2 clients
(okay, all you sun 3'ers can stop groaning, and ship me a sun 3 :-)).
Here's a sample comment from my /etc/nd.local:
# 0 64400 74520 94760 104880 125120 135240 165600
# | | | | | | | |
# |<--- 64400 --->|10120|<-20240->|10120|<-20240->|10120|<-- 30360 -->|
# | | | | | | | |
# | | | | | |wood-| |
# | Spare area |ducky| ducky |jedi | jedi |stock| (woodstock) |
# | Can either |root | swap |root | swap |root | swap |
# | be mkfs'ed | UNUSED | | | | |
# | locally or | | | | | | |
# | via 'nd' | | | | | | |
# | | | | | |
# | | | | | |
# | Currently used for accell | | | | |
# | | | | | |
#
The first part of it is a normal filesystem (/dev/xy0g), which was created
with: mkfs /dev/rxy0g 64400 The rest was created via /dev/ndl*
The most important part about doing any change like this is to *PLAN
AHEAD*. Create some diagram like this, so you know exactly how big each
partition is, and where it starts.
Name: David Vezie | Voice: "Hey, Dave"
Address: pixar!lfl!dv | Phone: (415) 662-1885
Company: Lucasfilm Ltd. | Fax: (415) 662-2186
------------------------------
Date: Wed, 19 Oct 88 17:00:38 PDT
From: jimc@math.ucla.edu
Subject: Re: Hard vs Soft Mounts
etnibsd!vsh@uunet.uu.net asks about the difference between hard and soft
mounting. With a hard mount, the client will retry NFS transactions
forever. Thus if the client is writing, the write absolutely will take
place (after the Sun FE shows up, replaces the board, etc. on the
server), provided the client isn't rebooted by an impatient user.
Needless to say, this is not always desireable.
With a soft mount, the client kernel is permitted to tell the user, "NFS
read/write failure" and punt the packet. This means you have to educate
your users, in case of editing a file, to stash it on /tmp and to rescue
it when their NFS site comes back. Fortunately our servers have been
fairly reliable so it hasn't been a big problem.
A bigger problem is that the application program may write a block, the
client may try to send it to the NFS server and encounter sluggish
response, and only upon closing the file will the kernel really flush the
file and discover that the write isn't going to happen. Now how many
programs do you know that properly check the return code from close()? The
file becomes truncated with no warning. I think that the crashing client
business may be related: somebody reads an executable file from a dying
server, doesn't get it all, doesn't check, and executes garbage --
possibly as root. I've never seen this happen, but it's remotely
conceivable.
We solve the problem this way: the discless client hard-mounts filesystems
of its own server, and all machines soft-mount other filesystems. We
encourage users to execute on their home server or its discless clients,
but they frequently use other sites anyway because of the convenience and
for load-levelling.
Here is a sample line from our fstab. We find that these parameters
(under SunOS v3.5) give reasonable freedom from complaints when the NFS
server is sluggish, yet let the users get on with their business when it
crashes:
redwood:/home/redwood /home/redwood nfs bg,rw,soft,timeo=20,retrans=5 0 0
James F. Carter (213) 825-2897
UCLA-Mathnet; 6608B MSA; 405 Hilgard Ave.; Los Angeles, CA 90024-1555
UUCP:...!{rutgers,ucbvax,sdcrdcf,{hao!cepu}}!ucla-cs!math.ucla.edu!jimc
ARPA: jimc@math.ucla.edu BITNET: jimc%math.ucla.edu@INTERBIT
------------------------------
Date: Thu, 20 Oct 88 03:15:53 EDT
From: Rayan Zachariassen <rayan@ai.toronto.edu>
Subject: Re: Encoding ASCII Files
As long as the encodings are compatible, it doesn't much matter which
implementation you use. However, one is widely available from the various
comp.sources.misc archives near you (or local packrats), something called
b{en,de}code. It is an updated version of the rscsencode mentioned
earlier, and should be used in preference. How to bootstrap? Send
cleartext code and get the recipient to undo any transit damage.
rayan
------------------------------
Date: Wed, 19 Oct 88 09:44:29 EDT
From: aad@stepstone.com (Anthony A. Datri)
Subject: Re: ispell and "emacs"
I get somewhat perturbed when I see people referring to "GNU Emacs" as
"Emacs" -- acting as if Unipress weren't there. I possess an ispell.ml
file for Unipress Emacs that I'll send to anyone who wants it.
Anthony A. Datri @SysAdmin(Stepstone Corporation) stpstn!aad
------------------------------
Date: Thu, 20 Oct 88 08:06:59 EDT
From: Chuck Musciano <chuck@trantor.harris-atd.com>
Subject: Re: Line drawing programs
We use Frame Maker to produce excellent quality viewgraphs containing
text, bitmaps, and line drawings. We also shoot 35mm slides directly from
LaserWriter output for slide presentations. Frame has a Mac-like
interface, but runs on a real computer.
I am in the process of getting a demo tape of Leonardo, a drawing program
from Qubix. I seem to have lost the address here, but it was advertised
in several of the trade magazines recently. Leonardo also handles splines
and 3D views of things, but I can't comment, since I've only read the
marketing stuff.
Chuck Musciano
Advanced Technology Department
Harris Corporation
(407) 727-6131
ARPA: chuck@trantor.harris-atd.com
------------------------------
Date: Thu, 20 Oct 88 08:49:39 EDT
From: steinmetz!grymoire!barnett@uunet.uu.net (Bruce Barnett)
Subject: Re: SunOS 4.0 Suntools speedup
One thing you can do to speed up SunOS 4.0 suntools is to trim EVERYTHING
you can out of the kernel. Every Kilobyte you can remove from the kernel
gains you 10 Kilobytes for user processes.
Bruce Barnett (barnett@steinmetz.ge.com)
------------------------------
Date: Sun, 25 Sep 88 14:06:09 +0100
From: mcvax!idt.unit.no!he@uunet.uu.net
Subject: Proxyarpd: release 2
Some time ago I received updates to my proxy ARP program from David
Robinson @ JPL that extended the program in a number of ways: SunOS 4.0
support was added, and the program was corrected so that it would run on a
Sun-4 as well (basically, by using "struct sockaddr_in" instead of
"u_long" a number of places). For Sun-3 systems running SunOS 3.X, there
is no change in functionality. The updated, merged version follows below.
Please note that I have been unable to really test this program under
SunOS 4.0 and on a Sun-4, so if anyone is able to verify that it works in
any of these configurations, please send a short message telling me so.
Haavard Eidnes, Div. of Computer Systems and Telematics, NIT, Norway
he@idt.unit.no
[[ This shar file replaces the old version in the archives. It is stored
under "sun-source" as "proxyarpd.shar" (30844 bytes). It can be retrieved
via anonymous FTP from the host "titan.rice.edu" or via the archive
server. For more information about the archive server, send a mail
message containing the word "help" to the address
"archive-server@rice.edu". --wnl ]]
------------------------------
Date: 20 Oct 88 12:21:25 GMT
From: unido!rmi!zentrale@uunet.uu.net (RMI Net)
Subject: SUN PCNFS & Micom NI5210
Has anybody used PCNFS with MICOM NI5210 card? What is the installation
procedure for this card?
Thanks,
Rupert Mohr
------------------------------
Date: Thu, 20 Oct 88 14:07:21 BST
From: everson%COMPSCI.BRISTOL.AC.UK@cunyvm.cuny.edu
Subject: TAAC mailing list
This message is to announce a new mailing list for users and potential
users of the TAAC - Sun's Graphics Accelerator. Other than using some of
their products (the TAAC included) I have no connection with Sun
Microsystems.
The list will be called:
taac-users@uk.ac.bris.cs for the list
taac-users-request@uk.ac.bris.cs administration/subscriptions
and will be available from 6am GMT 21st October, 1988. Prior to that email
me direct at the address at the end of this message.
Suitable gateways onto JANET are: from ARPANET via UCL (i.e. mail to
taac-users$-request%cs.bris.ac.uk@nss.cs.ucl.ac.uk) or via CUNY (i.e.
mail to taac-users$-request%uk.ac.bris.cs@edu.cuny.cunyvm).
I would be very interested in hearing from someone in the USA interested
in expanding the mailing list over there to reduce the amount of
trans-Atlantic email traffic.
Please try to avoid mailing to the list via uucp style mail (In Europe we
have to pay a lot for all mail received/sent by this method).
The intention is to provide a mechanism for potential/actual users of the
TAAC and as such the mailing list will be run on a simple non-moderated
echo basis to begin with.
Phill Everson
University of Bristol, UK
(everson@uk.ac.bris.cs)
------------------------------
Date: Wed, 19 Oct 88 21:09:19 EDT
From: gfr%wolfgang@gateway.mitre.org (Glenn Roberts)
Subject: Dist'n format for the 386i: tape vs. disk
If you're accustomed, as most Sun customers are, to being able to load
software off a remote tape drive, be careful on the 386i where this
capability is apparently not available for the standard "Application" and
"Developer's" distribution kits.
The cost difference between the 1/4" tape and the 3-1/2" diskette
distribution media is substantial ($234.00 in the current GSA schedule!!)
so we, with the concurrence of our Sun sales rep., opted to buy the tape
distribution thinking we could load it off one of our networked Suns. But
the distribution contains software 'clusters' (a nice way to package
pieces of Unix for optional loading and unloading) and uses a modified tar
format known as 'bar' format. These clusters are loaded with the 'loadc'
command, or individual programs may be loaded with 'load'. There is
currently (release 4.0.0) no support for using loadc on a remotely located
tape drive.
Sun loaned me a set of diskettes (20 high density 3-1/2" disks for the
Application set; 7 for the Developer's set) and I believe I have
successfully duplicated them as follows:
- use fdformat to format each disk
- for each disk do a cp /dev/fd0c temp , then do cp temp /dev/fd0c
Note that using 'diskcopy' in DOS emulation mode won't work since the DOS
and Unix formats are different.
Since we pay something like $6 each for these high density 3-1/2" disks I
ended up spending a good chunk of the $234 I originally saved. My advice:
buy the 3-1/2" distribution format and save yourself an evening's worth of
formatting and copying.
- Glenn Roberts, MITRE Corporation
gfr%wolfgang@gateway.mitre.org
------------------------------
Date: Thu, 20 Oct 88 10:12:13 EDT
From: steve@umiacs.umd.edu (Steven D. Miller)
Subject: SunOS 4.0 netstat brokenness
It is perhaps arguable that this is somehow more "right" than the 4.3BSD
code. The 4.3BSD code notes that there are more mask bits in use than
expected, and guesses at the subnet mask, assuming that a class A or class
B network is likely to be subnetted on 8-bit boundaries, and that a class
C is likely to be subnetted on 4-bit boundaries. It keeps widening the
netmask by the assumed boundary width until <net address> & ~netmask == 0,
and thus it does not deal properly with, say, a class B network with a
25-bit wide netmask.
It seems to me that something in netstat should get the interface
configuration, stuff the netmasks for each interface somewhere, then
compare the destination network addresses to those masks in order to pick
the right one. In order for the names to come out right, though, I
suspect that getnetbyaddr() would need some hacking. (How important are
the "a.b.c", "a.b", and "a" formats to inet_addr(), anyway?)
I'd rather have the 4.3BSD behavior than nothing, but then again, I'm
sitting on a class B network subnetted on 8-bit boundaries, so I'm
prejudiced. Anyway, netstat -n -r probably tells you the right things, in
an admittedly less reasonable format.
-Steve
Spoken: Steve Miller Domain: steve@mimsy.umd.edu UUCP: uunet!mimsy!steve
Phone: +1-301-454-1808 USPS: UMIACS, Univ. of Maryland, College Park, MD 20742
------------------------------
Date: Wed, 19 Oct 88 23:22:40 PDT
From: wolf@ssyx.ucsc.edu (Mike Wolf)
Subject: Problems with pc/nfs + IBM worm disk
We are trying to get pc/nfs up and running at the same time as an IBM worm
disk drive. For some reason, pc/nfs won't install when the worm device
driver is installed in config.sys. When autoexec.bat gets to the nfsrun
command, nfsrun returns the error:
PC-NFS already installed.
We can't find a way to stop this behavior. No other device has a problem
with the IBM drive. Has anyone seen this or a fix to it?
- Michael Wolf
wolf@ssyx.ucsc.edu
...!ucbvax!ucscc!ssyx!wolf
------------------------------
Date: Thu, 20 Oct 88 02:06:09 -0500
From: stevens%antares@anl-mcs.arpa
Subject: 88k for Sun 3/4 any news?
Does anyone have any info regarding addon processors based on the MC88100
or Hypermodule for Sun3' or 4's I suspect this would be some kind of VME
based card with some local memory in the spirit of the Transputer cards
floating around ?
Please reply to me directly. Thanks..
--Rick Stevens
------------------------------
Date: Thu, 20 Oct 88 10:01:42 -0400
From: Timothy A. Grunwald <fed!mqws1!m1tag00@uunet.uu.net>
Subject: What is autocall?
In a recent posting Craig Sarazin asked for improvements to calentool's
notification procedure. I use calentool and it works fine. I also use
monthtool which has significant advantages to calentool. The problem I
have is I don't know how to use its notification procedure.
>From the man page:
The .monthtool File
The reminders are kept in a file in the home directory
called the .monthtool file. This file is kept in a
format similar to that used by autocall, used to warn
the user of impending appointments.
What is the autocall facility and how do I use it?
Thanx for your help.
Tim Grunwald
Federal Reserve Board
21st & C st N.W. DC 20551
uunet!fed!m1tag00
------------------------------
End of SUN-Spots Digest
***********************