Sun-Spots-Request@Rice.edu (William LeFebvre) (09/04/88)
SUN-SPOTS DIGEST Friday, 2 September 1988 Volume 6 : Issue 216
Today's Topics:
Re: my pwd/getwd problem
Re: Ethernet "balancing" when serving many clients
tar filename length (2)
Rolo tool 2.0
Mathematica Information
groups of more than one line.
Bomb dbx
Caveats and Questions about "format"
sun-source/ofiles.shar
Help Needed On CD-ROM drive
Symbolic Computing Environment for Lucid Lisp?
Ethernet cards - SUN negligence, other vendors?
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, 1 Sep 88 23:37:03 EDT
From: stpstn!aad@philabs.philips.com (Anthony A. Datri)
Subject: Re: my pwd/getwd problem
I couldn't get my machine to umount the #$$#$% partition, so I booted it
standalone, and lo and behold! screwed up protections on the mount point
for the partition. Changed that, booted multiuser, and it's been fine.
Many thanks to Mr. Moderator for his explanation of the underlying cause.
Anthony A. Datri,SysAdmin,StepstoneCorporation,stpstn!aad
------------------------------
Date: Fri, 02 Sep 88 08:33:52 EST
From: mckay@harbor.ecn.purdue.edu (Dwight D McKay)
Subject: Re: Ethernet "balancing" when serving many clients
bob@allosaur.cis.ohio-state.edu (Bob Sutterfield) writes:
>ND only runs on the first interface, the one on the CPU board (ie0).
Not true. We have it running on ie1 as well. The only thing you must
remember to do is start a rarpd for each interface. No other
modifications are required.
--Dwight Mckay, ECN Workstation Software Support
--Purdue University, Engineering Computer Network
--Office: MSEE 104f, Phone: (317) 494-3561
--ARPAnet: mckay@harbor.ecn.purdue.edu, Usenet: ...rutgers!pur-ee!mckay
------------------------------
Date: Thu, 1 Sep 88 22:51:56 EDT
From: stpstn!aad@philabs.philips.com (Anthony A. Datri)
Subject: tar filename length (1)
I *think* that tar is specified to allow 128 character filenames --
certainly a number close to that. [[ As has been said recently, it is
exactly 100 characters. --wnl ]] I also believe that tar is a very old
program -- many implementations have a switch for reading/writing V6
compatible tapes, presumably indicating that the change took place with
V7, which is still a few years back. As I remember, V7 had the 14
character filenames (like 2.9bsd and hp/ux (grumble)), and 128 characters
probably worked reasonably well then. Now, with 900 MB drives and
programs like Interleaf that encourage long filenames, it doesn't cut it.
I've got a few disked 3/50's to back up in addition to our server, and I
don't like using dump because it makes me dump the whole partition -- the
user files are just in /usr, and tar doesn't like a lot of my users'
Interleaf files. I've been using dump (sigh), but what I may do is modify
a tar program (PDTAR would be a good one to try) to use a long filename
buffer. I haven't looked at the sources, but it can't be *that* hard to
do. Of course, your tapes are then unreadable without the hacked tar, but
you can solve that by keeping a standard-tar tape containing your code
with your backups. This is what I plan on doing.
While a single-file backup of everything would certainly be convenient
from an administrative standpoint, if it gets 98% of the way through and
your Fujitsu drive hands you a "xt0: hard error, er=1d", you're screwed.
And, unless you've got a big tape like an Exabyte, you also are limited by
the single-volume limitation of most tar programs. I think the AIX man
page for tar implied something of multi-volume tar archives, though...
Anthony A. Datri,SysAdmin,StepstoneCorporation,stpstn!aad
------------------------------
Date: Thu, 1 Sep 88 19:53:54 EDT
From: Bennett Todd <bent!bet@mcnc.org>
Subject: tar filename length (2)
man 5 tar reveals that the problem isn't specifically a misfeature of any
particular implementation of tar(1); the tar archive format specifies a
header block for each file, and the name slot in that header block is 100
bytes long. So, while you could hack PD-tar to read and write a tar-like
format with a longer filename spec, the resulting archives wouldn't be
compatible with any standard tar. That might or might not defeat the
purpose of using tar; if you were to view a suitably-hacked PD-tar as a
new utility strictly for Exabyte backups, and give it a new name (and keep
tar(1) around of course) then that might be a simple way to get the backup
tool you want. Me, I like dump. I use the following script:
#!/bin/sh
# Initialize
IFS='
'
table=/etc/dumpables # filesystems to dump
TAPE=/dev/nrsmt0 export TAPE # device to dump to (mt(1) uses $TAPE)
wrfile="bdd of=$TAPE bs=126b" # command to send a file to tape
echo "Rewinding tape and dumping header"
mt rewind # set up tape
shar `which $0` $table | $wrfile
egrep -v '^[ ]*(#.*)?$' <$table | \
sed -e 's/#.*$//' \
-e 's/[ ][ ]*/ /' \
-e 's/[ ]*$//' | \
while read machine partition command
do
echo "Dumping $machine:$partition"
rsh $machine $command $partition </dev/tty | \
$wrfile >/dev/tty 2>&1
done
mt weof
mt rewind
which looks prettier with tabstops every 4 spaces (I call it
/etc/takedump). It works off a database /etc/dumpables which looks like
this:
# database of filesystems to be dumped over the network
#
# machine partition command
server1 /dev/rndl0 /etc/dump 0dsf 6250 50000 - # client 1
server1 /dev/rndl1 /etc/dump 0dsf 6250 50000 - # client 2
server1 /dev/rndl2 /etc/dump 0dsf 6250 50000 - # client 3
server1 /dev/rndl3 /etc/dump 0dsf 6250 50000 - # client 4
server1 /dev/rxy0a /etc/dump 0dsf 6250 50000 -
server1 /dev/rxy1g /etc/dump 0dsf 6250 50000 -
server1 /dev/rxy0g /etc/dump 0dsf 6250 50000 -
server1 /dev/rxy0d /etc/dump 0dsf 6250 50000 -
server1 /dev/rxy1a /etc/dump 0dsf 6250 50000 -
server1 /dev/rxy1b /etc/dump 0dsf 6250 50000 -
server2 /dev/rxy0a /etc/dump 0dsf 6250 50000 -
server2 /dev/rxy0g /etc/dump 0dsf 6250 50000 -
(and like that for all the other machines)
The command to take the dump is specified in the database rather than in
the script because we have one machine (a stellar -- really slick) which
is System V based, and on which dump has a different name (because of a
name conflict). I decided to generalize the thing once and for all.
Everybody on our net who can handle rsh gets backed up onto one tape. I
love it. Oh, the shell script and database are written as a sharchive
first on the backup tape, so that I don't have to label the suckers with
anything more than a date. I am really truly lazy.
-Bennett
bet@orion.mc.duke.edu
P.S. dump/restore is my favorite for backups; they are most often used to
recover files that users have inadvertently deleted, and restore i (the
interactive option) is beautiful for picking files out of a dump.
------------------------------
Date: Thu, 1 Sep 88 20:46:19 MDT
From: roberts%studguppy@lanl.gov (Doug Roberts @ Los Alamos National Laboratory)
Subject: Rolo tool 2.0
I got rolo.shar.1 & rolo.shar.2 from titan, but I suspect that there is
something wrong with both shar files: (Getting them again in binary mode
didn't help.)
rolo.shar.1:
Inspecting for damage in transit...
Ouch [diff of wc output]:
5c5
< 1003 4285 24846 cards.c
---
> 1003 4285 24845 cards.c
9c9
< 1856 7497 48665 total
---
> 1856 7497 48664 total
rolo.shar.2:
Inspecting for damage in transit...
Ouch [diff of wc output]:
7c7
< 1325 4938 32778 panel.c
---
> 1325 4938 32774 panel.c
11c11
< 1741 6835 50869 total
---
> 1741 6835 50865 total
--Doug
Douglas Roberts
Los Alamos National Laboratory
(505)667-4569
dzzr@lanl.gov
[[ Oh phooey. I forgot to fix that. NOTE: THERE IS NOTHING WRONG WITH
ROLO. All that is missing are a few tabs. The program will still compile
and run correctly. I meant to fix up the totals in the shar files and
didn't. I will do that now. If you get this message when unpacking the
rolo files, don't worry. Just forge ahead. --wnl ]]
------------------------------
Date: Fri, 2 Sep 88 10:33:24 CDT
From: wri!windsor@uxc.cso.uiuc.edu (Eric Windsor Blankenburg)
Subject: Mathematica Information
[[ For those of you who read Usenet news: --wnl ]]
I have posted some up-to-date information about Mathematica for
Macintoshes, Suns and other machines on comp.newprod. as well as on
news.announce.important
-- Stephen Wolfram
------------------------------
Date: Fri, 2 Sep 88 07:55:48 EDT
From: Ronald G Minnich <rminnich@super.org>
Subject: groups of more than one line.
natints!brian writes:
> The problem, evidently, is that YP doesn't deal with group entries that
> consist of more than one line. For example,
>
> group1:*:50:brian,foo,foo2
> group1:*:50:lulamae,jimbob,marylou
>
> The users brian, foo and foo2 will actually be assigned to the group1
> group. The users lulamae, jimbob and marylou will not.
This appears to be a problem with makedbm. I tried the following:
duper> makedbm - /tmp/junk
key1 testing
key2 this
key1 sh*t
duper> makedbm -u /tmp/junk
YP_LAST_MODIFIED 0589204475
YP_MASTER_NAME duper
key2 this
key1 testing
duper>
So makedbm just throws away lines with identical keys, doesn't complain,
no indication of any problem.
Awful.
ron
------------------------------
Date: Tue, 2 Aug 88 10:50:20 EDT
From: brickman@cme-durer.arpa (Jonathan E. Brickman)
Subject: Bomb dbx
>The following code bombs dbx, although it runs fine by itself.
>#include <stdio.h>
>#include <suntool/sunview.h>
>
>main()
> {
> int i;
> Frame BaseFrame;
>
> BaseFrame = window_create(0,FRAME,0);
> printf("Type an int ");
> scanf("%d",&i);
> }
I know one possible bomb source. You're doing a scanf BEFORE the output
buffer has been cleared, i.e., before a '\n' has been printfed. Don't
know why dbx would bomb and ordinarily wouldn't, but maybe dbx does some
secondary bufferring of its own.
||Jonathan E. Brickman
------------------------------
Date: Thu, 1 Sep 88 13:14:49 EDT
From: Mike Jipping <jipping@cs.hope.edu>
Subject: Caveats and Questions about "format"
Folks:
This is about using "format" under SunOS 4.0. We are running 4.0 on our
3/160 server, (serving a set of 3/50s) with a 280 MB drive on a Xylogics
controller. Recently, we have had to repartition the disk -- twice: once
to reconfigure space after initial setup (we didn't expect all the
additional disk space!) and once to increase space to /usr.
The first time, after searching the manuals and calling Sun support, I
felt ready! I didn't change the root partition (xy0a), but I moved other
partitions around, so I planned on losing mount, umount, and newfs. Not
to worry! Moving these to the root partition, I could find them and still
redo the other filesystems (Sun Technical Support suggested this method).
I eventually used a combination of MUNIX (completely memory resident
SunOS) and format to make the new partitions. I then attempted to restore
the file systems...and ran into problems. Based on this experience I have
some comments/questions:
(1) The scheme worked fine to a point. I was able to use mount and
umount to manipulate filesystems, but newfs bombed. The reason? mount
umount are not dynamically linked and newfs is. Thus, if you are
tinkering with /usr and lose /usr/lib/{ld.so,libc.so.0} (like a disk
repart surely would) you are out of luck! I did this the first time,
and had to go through suninstall to reinstall 4.0!
(2) The second time, I adjusted the size of the /usr partition (xy0g),
but this time copied the required parts to another partition, mounted
/usr on that partition, remade and restored xy0g, then remounted
everything and restored the second partition. It worked -- but what a
pain!
(3) Is there another way to do this? If newfs was not dynamically linked,
everything should work as expected, right?
(4) Finally, the documentation for format needs work! The SysAdmin
manual hints (p. 181) that MUNIX may be necessary for certain tasks,
but does not say which ones! I know it makes sense that for tinkering
with the disk you are running off of you might need something like
MUNIX, but even the Technical Support guy couldn't tell me whether I
needed it for a repart or not [note: I even moved format (not
dynamically linked) to the root partition and unmounted every effected
partition, but format refused to run without MUNIX!].
Does anyone have comments on this? I would think that a repartition is
likely once in a while, especially around new system installation time.
And with all the hornblowing Sun has been doing about the 4.0 format
program, I would think this would be easier!
-- Mike
Mike Jipping
Hope College
Department of Computer Science
jipping@cs.hope.edu
P.S. Just a last thought...try using the instructions in the SysAdmin
manual for restoring a damaged "/usr". It explicitly dictates the use of
/usr/etc/newfs... with both the program and the libraries on the missing
partition.
------------------------------
Date: Fri, 02 Sep 88 11:28:02 EDT
From: Daniel R. Ehrlich <ehrlich@blitz.cs.psu.edu>
Subject: sun-source/ofiles.shar
I recently retrieved sun-source/ofiles.shar from titan.rice.edu, but have
not been able to get it to compile successfully on a Sun 4/260S running
SunOS 4.0. Has anyone else been able to do this? If so what do I need to
do to make it work? Thanks in advance.
Dan Ehrlich <ehrlich@blitz.cs.psu.edu>
The Pennsylvania State University
Department of Computer Science
University Park, PA 16802
------------------------------
Date: Fri, 2 Sep 88 11:53:32 CDT
From: keane arase <kean@sphinx.uchicago.edu>
Subject: Help Needed On CD-ROM drive
*** H E L P ! ! ! ***
Does anyone out in Netland know the diffence between a Toshiba XM-2100A
CD-ROM Drive and a XM-2000A CD-ROM Drive?
I am trying to install one of these CD-ROM devices on a local Sun. I've
got the driver, and have installed it, and everything seems to check out
OK, but can't win arbitration over the drive.
The unfortunate thing about it is, the hardware I have is a XM-2100A
drive, and the driver I have is for a XM-2000A drive. Was this just
*another* typo in the driver docs? Or is there a difference between these
drives?
(I hate coming into the middle of a project :-()
I have called Toshiba, but all I get are Answering Machines that don't
return calls. If anyone out there can help, it would be *very*
appreciated.
Thanks!
Keane Arase
Systems Programmer
University of Chicago
kean@sphinx.uchicago.edu
syskean@uchimvs1.uchicago.edu
312/702-2556
------------------------------
Date: 2 Sep 88 13:18:07 GMT
From: mit-amt!turk%mit-amt.MEDIA.MIT.EDU@eddie.mit.edu (Matthew Turk)
Subject: Symbolic Computing Environment for Lucid Lisp?
I'm looking for info about the Symbolic Computing Environment for Lucid
lisp that I believe Sun just came out with. What does it do? Supposedly
it makes the Sun Lisp environment much more Symbolics-like, which would be
great.
Mail to me, I'll summarize if folks are interested. Thanks,
Matthew Turk
turk@media-lab.media.mit.edu
------------------------------
Date: Fri, 2 Sep 88 14:28:59 EDT
From: "Pawel Stefanski" <stefan@gmuvax2.gmu.edu>
Subject: Ethernet cards - SUN negligence, other vendors?
We ordered our second Ethernet card in the middle of April, with the
promise of 30 days delivery. We are still waiting for it today. Is it a
single case, or a symptom of something more common, caused by 'SUN rapid
growth'?
Also, does anyone know about other possible vendors, who offer fully
compatible cards (for SUN-3) and are more reliable?
Pawel A. Stefanski, Phone (703)764-6057, (703)323-2713,
(stefan@gmu90x.gmu.edu OR stefan@gmuvax2.gmu.edu )
Machine Learning Laboratory, Department of Computer Science,
George Mason University, Fairfax, VA 22030.
------------------------------
End of SUN-Spots Digest
***********************