Sun-Spots-Request@RICE.EDU (William LeFebvre) (03/08/88)
SUN-SPOTS DIGEST Sunday, 6 March 1988 Volume 6 : Issue 25 Today's Topics: Re: lpd printing to a TCP terminal server attached printer Re: rcp and rlogin hang in 3.4EXPORT, fix Re: Multi-process debugging using Dbxtool Re: Diskless CAD Stations 3.5 UPGRADE URGENT Bug fix for Sun 4-3.2FCS SUN rasterfiles -> QUIC WORM drives HP LaserJet Troff filter Stuck in caps lock fix Problems with Sun 3/50 and X Needed: info on TOOL-TOOL Sun 4 info requested ping doesn't work with subnets and 3.4.2? Questions about 256 entry colormap GKS implementations? 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 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@rice.edu". ---------------------------------------------------------------------- Date: Thu, 25 Feb 88 12:18:26 EST From: Andrew Hodgson <munnari!mimir.dmt.oz.au!ash@uunet.uu.net> Subject: Re: lpd printing to a TCP terminal server attached printer The easiest way to print through a terminal server or any other non local device is to write the accounting filter and make it send the file to the remote device. As an example we have a imagen laser printer on our ethernet which uses IP both TCP and UDP. The termcap entry is as follows imaprint|generic Imagen printer:\ :ff=\00:lo=lock:mx#0:pl=60:pw=80:px=2016:py=2624:rw:\ :sb=Imagen 8/300 Laser Printer:S:sh: imaenet|generic Imagen ethernet printer:\ :cf=/usr/lib/imagen/icif.e:\ :df=/usr/lib/imagen/idvi.e:\ :gf=/usr/lib/imagen/igraph.e:\ :if=/usr/lib/imagen/ipf.e:\ :nf=/usr/lib/imagen/idimp.e:\ :rf=/usr/lib/imagen/ifort.e:\ :tf=/usr/lib/imagen/icat.e:\ :vf=/usr/lib/imagen/imp.e:\ :tc=imaprint: ip|mezz_ip|imagen|Imagen 8/300 Laser Printer:\ :lp=/dev/null:hn=ip:rt=/usr/spool/lpd.ip/resfonts:qu#300:\ :af=/usr/adm/lpd.ip-ac:ah=mimir:lf=/usr/adm/lpd.ip-errs:\ :sd=/usr/spool/lpd.ip:tc=imaenet: Note the lp device is /dev/null so that lpd just sends it to dev/null and the actual communications is done by :if: which makes a TCP/IP connection to the printer. ------------------------------ Date: Wed, 24 Feb 88 23:40:17 PST From: rob@presto.ig.com (Rob Liebschutz) Subject: Re: rcp and rlogin hang in 3.4EXPORT, fix > Since we have installed SUNOS 3.4 (EXPORT) on our SUNs we expierience the > following problem: > When starting an rcp to or from a remote hosts which is not a SUN (in our > case either a VAX 11/750 BSD4.3 or a Gould PN9000 UTX 2.0 == Bsd4.3) rcp > copies a few blocks (the exact block count varies) and then hangs. ... I've seen similar problems with ftp. There is a 0 size window bug in 3.4 tcp which causes this problem. I've not tried it yet, but the fix is supposed to be to install 3.5 which is shipped to sites with software support only if you ask for it. Another solution would be to get the 3.4 tcp source that's floating around the net and link that into your kernel. ------------------------------ Date: Thu, 25 Feb 88 09:12:51 PST From: weiser.pa@xerox.com Subject: Re: Multi-process debugging using Dbxtool Yes, you can debug multi-processes with dbx, but you have to do it yourself. If you do a fork, then an exec, just have your program remember the child pid use that to attach a new dbx. (In the system we're developing here there is some code to do this automatically for certain fork/exec's done explicitly). If you do a fork, but not an exec, dbx will not be able to attach to the new child unless your executable file has type OMAGIC. (See 'man ld' for how to make a file have OMAGIC). We do this here--works fine. -mark ------------------------------ Date: Thu, 25 Feb 88 13:09:48 PST From: jeff@tc.fluke.com (Jeff Stearns) Subject: Re: Diskless CAD Stations If you have enough RAM to keep your code and data resident in memory without paging, then a network between you and the disk doesn't make much difference. If you must page a modest amount, expect a noticable but acceptable slowdown. If your workstation must page steadily, or if it must *swap*, expect to become very unpopular with others who share the network with you. Your response time (and theirs) may also increase by an order of magnitude. If your application's appetite for data is modest, don't expect NFS I/O to be a significant bottleneck. Be advised that reading data via the NFS is several times faster than writing data; if your application writes several hunderd kilobytes or more, the NFS could become a significant factor. A diskless Sun running release 3.4 should have *A MINIMUM OF 8 MEGABYTES* to run SunWindows without paging. Expect this to change with the soon-to-be- nnounced release 4.0. (Bad news: no software release from Sun has ever consumed less RAM than its predecessor.) To your basic 8Mb, add enough RAM to hold the estimated working set for your application. This may be 1-2Mb for code and many Megabytes for data. Our experience suggests that diskless Suns with 16Mb should be good performers with most current CAD software. The key to diskless performance is to ensure that you rarely page and never swap. Jeff Stearns Domain: jeff@tc.fluke.COM Voice: +1 206 356 5064 UUCP: {uw-beaver,decvax!microsof,ucbvax!lbl-csam,allegra,sun}!fluke!jeff Snail: John Fluke Mfg. Co. / P.O. Box C9090 / Everett WA 98206 ------------------------------ Date: 24 Feb 88 02:46:39 GMT From: cyrus@hi.unm.edu (Tait Cyrus) Subject: 3.5 UPGRADE Again I am in the process of upgrading my SUN 3/160 from one release of SunOS to another (3.4 to 3.5). As before, when I ran SUN's UPGRADE shell script, I found some possible changes to the UPGRADE script which would make it run (better). It appears that 3 of my suggestioned 4 changes (and who knows how many other peoples) were made to the 3.5 version of the script with the exception of how it finds which clients are enabled and the nd partition numbers those clients use. The problem I had was that I made /etc/nd.local "look" pretty (used tabs and such to line the different fields up). This caused UPGRADE to bomb because it expected spaces to be between fields. Also, 2 of our 4 clients were commented out (the ones using nd0 and nd2). The UPGRADE script "assumes" that the partition numbers start at 0 and are sequential. Below are the diffs plus the added file needed to "clean" up the UPGRADE script to make it work with any possible format of /etc/nd.local and any possible sequence of nd partition numbers. I don't know if anyone cares (especially SUN since 4.0 won't have nd to deal with) so you might only want to be aware of the possibility of problems if /etc/nd.local has been "hacked" on. *** UPGRADE.org Mon Feb 15 21:31:37 1988 --- UPGRADE Tue Feb 16 00:20:29 1988 *************** *** 542,605 **** fi # # Create CLIENTLIST from /etc/nd.local ! # Looks ugly but handles all possible cases ! # nd.local format ! # user hostname hisunit mydev myoff mysize mylunit maxpkts ! # ! endsym=$ ! ind=0 ! CLIENTLIST="" ! while true; do ! grep " $ind $endsym" $NDLOCAL > /dev/null 2>&1 ! case "$?" in ! 0) ! set `grep " $ind $endsym" $NDLOCAL` ;; ! 1) ! grep " $ind $endsym" $NDLOCAL > /dev/null 2>&1 ! case "$?" in ! 0) ! set `grep " $ind $endsym" $NDLOCAL` ;; ! 1) ! grep " $ind$endsym" $NDLOCAL > /dev/null 2>&1 ! case "$?" in ! 0) ! set `grep " $ind$endsym" $NDLOCAL` ;; ! 1) ! grep " $ind [0-9]$endsym" $NDLOCAL > /dev/null 2>&1 ! case "$?" in ! 0) ! set `grep " $ind [0-9]$endsym" $NDLOCAL` ;; ! 1) ! grep " $ind [0-9]$endsym" $NDLOCAL > /dev/null 2>&1 ! case "$?" in ! 0) ! set `grep " $ind [0-9]$endsym" $NDLOCAL` ;; ! 1) ! break ;; ! esac ;; ! esac ;; ! esac ;; ! esac ;; ! esac ! if [ "$1" = "version" ] ; then ! shift 2 ! fi ! CLIENT=$2 ! case "$1" in ! "#user" ) ! ;; ! "user" ) ! mkdir /${CLIENT} 2> /dev/null ! CLIENTLIST="${CLIENTLIST} ${CLIENT}" ;; ! * ) ! ;; ! esac ! ind=`expr ${ind} + 1` ! done ! for CLIENT in $CLIENTLIST ! do ! set `grep "user $CLIENT 0" $NDLOCAL | sed s,\/," ",g` ! ndl=ndl${8} cd /dev /dev/MAKEDEV ${ndl} 2> /dev/null done --- 542,553 ---- fi # # Create CLIENTLIST from /etc/nd.local ! # ! for CLIENT in `awk -f get_client_names < $NDLOCAL` ! mkdir /${CLIENT} 2> /dev/null ! CLIENTLIST="${CLIENTLIST} ${CLIENT}" ! set `grep $CLIENT $NDLOCAL | awk '$3 == "0" { print $0 }'` ! ndl=ndl${7} cd /dev /dev/MAKEDEV ${ndl} 2> /dev/null done -------- below is file "get_client_names" --------------------------- # # This awk script expects /etc/nd.local as input. The output produced # is a list of names of client machines that are enabled in /etc/nd.local. # Note: Unlike the original way SUN did it, this version does not care # if the partition numbers are sequential or not. # # nd.local format # 'user' hostname hisunit mydev myoff mysize mylunit maxpkts # # # only look at lines that have 'user' in them # $1 ~ /user/ { # # just make sure that 'user' was the first thing in the line # (i.e. not commented out) and that the nd partition number # is positive. If the partition number is negative, then # it is refering to either the swap area or the pub area which # we don't want. If it is positive, then it is a real client # partition number. # if( $1 == "user" && $7 >= 0 ) printf("%s ",$2) } -------- above is file "get_client_names" --------------------------- -- W. Tait Cyrus (505) 277-0806 University of New Mexico Dept of Electircal & Computer Engineering Parallel Processing Research Group (PPRG) UNM/LANL Hypercube Project Albuquerque, New Mexico 87131 e-mail: cyrus@hc.dspo.gov ------------------------------ Date: Wed, 2 Mar 88 11:32:36 PST From: S.C.Blair FLIC Admin(512)331-3426 <ascway.UUCP!scb@spar-20.spar.slb.com> Subject: URGENT Bug fix for Sun 4-3.2FCS For those of you who've been @#$%'d by the fact that the Sun4-3.2FCS won't allow you to NFS to the Sun4 from the Sun3's, here's the fix: To do this, all you need is *1* machine on your network running SUNO/S3.5: cd /usr/etc cp rpc.mountd rpc.mountd.3.4 rcp(from sun3.5 machine)/usr/etc/rpc.mountd rpc.mountd.3.5 ps ax | grep mountd kill the bugger cp rpc.mountd.3.5 rpc.mountd kill '/etc/nfsd >x<' and re-fire it off... un-comment the machine(SUN4) in your fstab, and do a 'mount -a'. The Sun4 node will then appear JUST LIKE A NORMAL MOUNT (which it is). I have my machine running 3.5, but most of my other machines are running 3.4 (don't ask why!!). Then things will be fine...(diskless and diskfull Sun3's-I don't have any Sun2). Steve Blair Schlumberger Technology Corp.-Austin, Texas Disclaimer: The above is working fine on 30+ Sun3's running 3.4 O/S. If you're running a lower release, good luck. Your Sun4 must be running SunOS Release Sys4-3.2(which mine is). Otherwise YOU'RE ON YOUR YOUR OWN and *I'M* not responsible.....(works for me fine!!)scb ------------------------------ Date: Wed, 24 Feb 88 15:08:54 EST From: nagasamy@caip.rutgers.edu (Vijay Nagasamy) Subject: SUN rasterfiles -> QUIC Here is the code for converting SUN rasterfiles to QMS QUIC code. Sorry, there is no man page for this program, however there is sufficient help available in the program itself. (Try rfquic2 -h). Uucp: {..., ames, cbosgd, harvard, moss, siesmo}!rutgers!caip!nagasamy Internet: nagasamy@caip.rutgers.edu US Mail: Vijay Nagasamy, Department of Mechanical & Aerospace Engineering, Rutgers University, P.O. Box 909, Piscataway, NJ 08855. Voice: Off: (201)-932-2824 Home: (201)-878-1818 [[ Stored in the archives as "sun-source/rfquic2.shar". It is 9809 bytes in length. --wnl ]] ------------------------------ Date: Wed, 24 Feb 88 07:44:38 pst From: cygnet!will@sun.com (Will Nelson) Subject: WORM drives We are using WORM drives on sun 3's. These are of the 12" variety, using the SCSI interface, and of course, our high performance jukeboxes. We currently have drivers for Hitachi and Optimem drives. These were originally purchased from SUN, and have been enhanced to operate better in the jukebox environment. Will Nelson uucp: {ucbvax!decvax!decwrl}!pyramid!oliveb!cygnet!will Cygnet Systems, Inc. 601 West California Avenue Sunnyvale, CA 94086-4831 (408) 773-0770 ------------------------------ Date: Thu, 25 Feb 88 14:41:52 EST From: darkstar!brian@uc.msc.umn.edu (Brian Utterback) Subject: HP LaserJet Troff filter I have received a number of requests for the LaserJet troff filter mentioned in the printcap entry I submitted to the archives. The filter was written by Sverre Froyen and is available in the comp.sources.unix(?) archives. Anyway, I think that I got them from uunet. I think they were called "lcat". Anyway, if you have trouble finding them, I will send you a copy. Brian Utterback |UUCP:{ihnp4!cray,sun!tundra}!hall!blu Cray Research Inc. |ARPA:blu%hall.cray.com@uc.msc.umn.edu One Tara Blvd. #301 | Nashua NH. 03062 |Tele:(603) 888-3083 ------------------------------ Date: Thu, 25 Feb 88 16:32:31 EST From: Timothy.Freeman@theory.cs.cmu.edu Subject: Stuck in caps lock fix A response to Aaron Konstam's note in the 25 Feb 88 Sun-spots: It seems that my Sun 3/160 sometimes misses an up transition on the shift keys. When I'm running X, the problem goes away after I press both shift keys at once. Sometimes the meta keys (labelled left and right) get "stuck" too; this goes away when I press both meta keys at once. I don't know whether these workarounds would work if I were not running X. ------------------------------ Date: Tue, 23 Feb 88 23:40:14 EST From: sjc@purdue.edu (Steve Chapin) Subject: Problems with Sun 3/50 and X I've been having a minor problem with my Sun 3/50 and X windows. I'm sure the answer is obvious, but it happens rarely enough that I haven't bothered to track it down, and just often enough to keep me grumbling. There is some magic keystroke sequence that causes X to interpret everything typed as a capital letter. I can never figure out what it is, because I always type a few strokes before I realize I've done it. (It's ctrl-something). Anyway, I'd like to know if there's a corresponding keystroke that will restore the keymap. Thanks for any help... Steve Chapin ARPA: sjc@gwen.cs.purdue.edu UUCP: ...!purdue!sjc ------------------------------ Date: Wed, 24 Feb 88 13:52:51 EST From: gotham!ursa!future!raj@sun.com (Raj) Subject: Needed: info on TOOL-TOOL Those of you that have struggled with Sunview know how hard it it to put together a complex arrangement of buttons, sliders etc on a screen, I mean in terms of placement etc. Also trying out different sizes, fonts etc take a lot of time. What is needed here is a TOOL-TOOL - a tool that will let you interactively specify where you want the panels, buttons, sliders, canvasses etc., and when you hit "go", it will either write the code for you or atleast give the parameters to use in the calls to window_create. I was told by our Sun-rep that sun is working on such a tool, but with low priority. When we know that high priority items such as 4.0 are delayed an year, you can imagine when the tool-tool would be out. He also told me that IDE had such a tool. Apparently someone at IDE worked on it for his masters thesis. I would be interested in obtaining information, source, or binary of such a tool-tool if any of you out there have it. You can email me at sun!gotham!ursa!raj. ------------------------------ Date: 24 Feb 88 17:10:07 GMT From: wucs1!wucs2!krf@uunet.uu.net (Kevin Fenster) Subject: Sun 4 info requested We are replacing a number of Vax 11/750s starting this summer, and are looking at various machines to put in their place. One of the machines which we are looking at is the Sun 4 line, specifically the Sun 4/280 Server. Our goal is to support 100 simultaneous users for the next academic year. Preferably these 100 users could still get something done. Right now we try (but fail) with 3 750's running 4.3BSD Unix. The type of work done is basically instructional computing within an Engineering School, which is primarily compiles for various languages (including ADA), document prepartion, circuit simulation, circuit layout, symbolic manipulation, finite element analysis, etc. Here's the question: How many users can we support on a Sun 4/280? The config guides for a Sun 3/280 says approx. 20 diskfull workstations running NFS. My feeling is that it takes more power to support a person tied to the machine with a regular dumb terminal, than a workstation (missing some cpu crunch). This seems to indicate that a Sun 3/280 can support fewer than 20 persons? Am I all confused? Is there anyone running 3/280's as timesharing machines (i.e. non-workstation)? I realize that a 4/280 has more cpu kick, but it uses the same I/O; a bottleneck? Any help, comments, pointers to this would be helpful. We are also looking at the Sequent Symmetry series and the Encore Multimax. I realize these are completely different machines, but their price structures are also a little different. Ooops!! Yes, I realize that we can't support all 100 users off of a single sun. It may take 4 or 5 to do it. That's what I am trying to find out. Thanks in advance. Kevin Fenster Washington University ECL, St. Louis 314-889-6429 UUCP: krf@wucs1.UUCP or ..!uunet!wucs1!krf "Someday it will be krf@wucs2.wustl.edu" ------------------------------ Date: Wed, 24 Feb 88 19:00:14 EST From: starner@prc.unisys.com (Mark Starner) Subject: ping doesn't work with subnets and 3.4.2? Since upgrading to 3.4(.2) and subnetting a class B subnet into a class C subnet, ping doesnt seem to work through our gateways. I can ping someone on my own network, but cant ping the "other" side of the file server (gateway) or any other machines in our subnet. I can ping outside of our network fine. Pinging from Non-Sun systems into any of our machines works fine. (ie, from our vax to a workstation through a file server gateway) Anyone have any ideas? solutions? Thanks Mark Starner (215) 648-7382 Unisys/Paoli Research Center Paoli, PA starner@prc.unisys.com ------------------------------ Date: Wed, 24 Feb 88 19:46:08 EST From: rayk@sbcs.sunysb.edu (Raymond T Kreisel) Subject: Questions about 256 entry colormap Does anyone know if it is possible to have a 256 entry colormap for a canvas and have this same 256 entry colormap associated with panels that are in the same frame as the canvas ? In all of my attemps, on Sun 3/110's and Sun3/160's, the panels always keep the mono colormap, which came from the inital SunView null frame. thanks in advance, ray Ray Kreisel CS Dept., SUNY at Stony Brook, Stony Brook NY 11794 UUCP: {allegra, philabs, pyramid, research}!sbcs!rayk ARPA-Internet: rayk@sbcs.sunysb.edu CSnet: rayk@suny-sb ------------------------------ Date: Thu, 25 Feb 88 13:21:09 EST From: palmer@ncifcrf.gov Subject: GKS implementations? We are interested in acquiring a GKS implementation for our Suns. What commercial products are out there? I'd be especially interested to hear personal experiences with various systems. Please mail directly to me. Thanks. -Tom Thomas C. Palmer NCI Supercomputer Facility c/o PRI, Inc. Phone: (301) 698-5797 PO Box B, Bldg. 430 Uucp: ...!uunet!ncifcrf.gov!palmer Frederick, MD 21701 Arpanet: palmer@ncifcrf.gov ------------------------------ End of SUN-Spots Digest ***********************