Sun386i (RoadRunner) Frequently Asked Questions This FAQ dfescribes the Sun386i computer, common problems, solutions, software available, hardware available, and almost anything else people on the sun386i mailing list have discussed over the past few years. Please feel free to make your own contributions and help fix any bugs. Last Updated: August 12, 1994/Stephen Houser (houser@usm.maine.edu) 0-0. Table of Contents 1-0. General Information about the FAQs and the Sun386i 1-1. Where is the FAQ? 1-2. Who do I ask questions? 1-3. What is a sun386i, RoadRunner, etc. 1-4. Where did the 386i get the name 'RoadRunner'? 2-0. Software for the Sun386i 2-1. Anonymous FTP sites for the Sun386i 2-2. SunOS 4.0.2 Upgrade, Where can I get it? 2-3. SunOS 4.0.2 Problems installing from tape 2-4. Synopsis of available 'free' software 2-5. GCC 2.2.2 2-6. GCC 2.5.8 2-7. GhostScript 2.6.1 2-8. GDB 4.7 (does not work) 2-9. GNU EMACS 18.58 2-10. PERL version 4.036 2-11. XWindows version 11 release 5 2-12. XWindows version 11 release 6 2-13. GnuPlot 3.5 2-14. HP LJ III printcap 2-15. PPP (ppp-sunos4.1.pl6.tar.Z) on a 386i 2-16. SNAP Cannot Create New Users 2-17. Removing NIS YellowPages 3-0. Hardware for the Sun386i 3-1. NVRAM Problems and Solutions (lots of problems) 3-2. NVRAM Suppliers (Sun Battery) 3-3. Memory Expansion 3-4. 486 upgrade for the 386i 3-5. Large SCSI disks 3-6. Running around topless (i.e. without a framebuffer/monitor) 1-0. General Information about the Sun386i ----------------------------------------------------------------------- 1-1. Where is the FAQ ----------------------------------------------------------------------- I (Stephen Houser) maintain the FAQ and try to post it to the= list semi-regularly (i.e. every so often). If you are really desperate look in the ftp sites listed in this copy of the FAQ. If you are really really desperate and cannot find it there...= I have anonymous ftp setup on my own (working machine) with the current= FAQ and perhaps a few other files (as I compile them, and before the other sites get them). ftp://cxc1.msu.edu/386i/list_archive Michigan State University Charles Henrich http://rs560.msu.edu/~henrich ftp://sahsun.usmacs.maine.edu/pub/sun386i (temporary site) University of Southern Maine Stephen Houser http://csir1.usmacs.maine.edu/~houser I'm also working on an HTML version of the FAQ so you can perhaps find it via my home page URL given above. ----------------------------------------------------------------------- 1-2. Who do I ask questions? ----------------------------------------------------------------------- There is a mailing list for Sun 386i users maintaned by; forgotten@some.unknown.site.drat! In other words, I forgot, please tell me, my mail to ...-request seems to bounce To get added to the list (or subtracted) send mail to: sun-386i-request@itc.yorku.ca -- bounces?!? For general things, like postings, and all that send mail to: sun-386i@itc.yorku.ca -----------------------------------------------------------------------= =20 1-3. What is a sun386i, RoadRunner, etc. =20 -----------------------------------------------------------------------= =20 It's an Intel 80386-based UNIX computer produced for a short while by Sun Microsystems. There are 2 standard models the 150, and= the 250. Sun386i 150 - A 20Mhz 80386? Sun386i 250 - 25Mhz 80386 CPU, 80387 Co-processor, 327 or 91 Mb=20CDC Wren= IV Hard disk, 1 Rs232 port, 1 Centronics printer port, standard= =20 (AUI) Ethernet port, 3 Sun (??) bus expansion slots, 5 PC AT ISA bus expansion slots. =20 ----------------------------------------------------------------------- 1-4. Where did the 386i get the name 'RoadRunner'? ----------------------------------------------------------------------- ... under construction ... 2-0. Software for the Sun386i ----------------------------------------------------------------------- 2-1. Are there any anonymous FTP sites for the Sun386i? ----------------------------------------------------------------------- Here are the URLs of the popular sites that I know of. They contain both binaries and archives of the mailing list: ftp://cxc1.msu.edu/386i Michigan State University Charles Henrich http://rs560.msu.edu/~henrich ftp://ftp.primus.com/pub/sun386 Primus Control (who maintains it??> ftp://usc.edu/archive/usenet/systems/sun/sun-386i (maintainer?) ----------------------------------------------------------------------- 2-2. SunOS 4.0.2 Upgrade; Where can I get it? ----------------------------------------------------------------------- ftp://cxc1.msu.edu/386i/software/sunos4.02 INSTALL README diskette.[1-8] Contains images of eight floppies used to upgrade both the runtime operating system and the development system. On of the disks is= a replacement diagnostics disk, that does not seem to boot? Get the README file along with the INSTALL and diskette.* files.= It describes step-by-step how to make floppies from the images and= how to install the upgrade. You will need to have a working SunOS= 4.0.1 system to do the upgrade. ----------------------------------------------------------------------- 2-3. SunOS 4.0.2 Problems installing from tape ----------------------------------------------------------------------- From the July 1990 Software Technical Bulletin (Sun publication), SunOS Sun386i Patch - page 39-40: ------------------------------------------------------------------ Patch for the loadc command for those running SunOS Sun386i 4.0.2. Symptom: Clusters upgraded by SunOS Sun386i 4.0.2 cannnot be= loaded from tape using the loadc command. Problem: For those running SunOS Sun386i 4.0.2, the following= error message is encountered when using the load or the loadc commands to load any clusters modified as part of the SunOS Sun386i 4.0.2 release. This occurs for those= using tape media only, and does not occur when using the diskette media. "The wrong tape is in the drive." The above error message always occurs, and after the user has correctly inserted the SunOS Sun386i 4.0.2 release tape which= the system requested. This problem occurs after reinserting the requested tape. Without the below workaround to this problem, the user cannot= load the following clusters: advanced_admin base_devel man_pages config networking_plus proflibs sysV_commands sunview_devel Also, loading either or both of the below 'special' global names will fail: appl devel Workaround: Use the below procedure to work around the loadc problem (ie. patches the program). 1. Log in as root or change to superuser (represented by the sharp character, #, below). Then issue the commands shown in the following steps. 2. # mount -o remount /usr <-- make /usr writeable 3. # cd /usr/lib/load 4. # cp load load.old 5. # sed -e '361s/)/|*"$1 $NAME"*)/' load.old > load 6. # sum load 7. If the checksum from the sum command is as shown below, 61115 14 then enter the commands shown next. # rm load.old # sync; sync; reboot 8. If the checksum from the sum command is not as shown below, 61115 14 then enter the following command # cp load.old load and return to step 5 above. After rebooting, the system should now allow the clusters listed above to be loaded without further difficulty. ----------------------------------------------------------------------- 2-4. Free software ----------------------------------------------------------------------- Here is a short list of current versions that work on the RoadRunner, and perhaps are avaliable in binary form for anonymous ftp from= some nice person. Many programs compile without incident, those that= cased us problems are noted, and have seperate compilation directions.= For the most part many GNU programs do work, allthough they usually= need a little tweaking. If you don't see what you like ask on the list and let's get more programs added to this list. Most of these that have been checked were done so with the GNU= gcc compiler. Using the distributed Sun cc compiler may not work (especially with X windows). GCC 2.5.8 (C/C++ compiler) GCC 2.2.2 (C compiler) BASH 1.12 (shell replacement, like ksh) EMACS 19.24 (text editor/operating system) GhostScript 2.6.1 (PostScript interpreter) PERL 4.036 (text processing languange) PATCH 2.1 (file patcher) GZIP 1.2.4 (compression) ZOO 2.0.1 (compression) XMODEM 3.9 (file transfer) ZIP 1.0 (PC-compatible compression) UNZIP 5.0 (PC-compatible compression) Oleo 1.5 (spreadsheet) ISpell 4.0 (spell checker) GnuPlot 3.5 (graph plotting program) TeX 3.1415 (C version 6.1) (text formatting system) LaTeX2e 1994/06/01 patch level 3 Dvipsk 5.55a (TeX drivers for PostScript) Xdvik 1.8 (Tex previewer for X Windows) XWindows 11 release 5 patchlevel 25 (X Window system) XMosaic 2.1 (WWW client for X) XV 3.00 (Graphic viewer for X) XView 3.20 (Window manager for X) XWindows 11 release 6 (work in progress) Here is a list of some programs that DO NOT WORK on the RoadRunner. There may be people working on getting them to go, post to the= list with any information or insight that might make them work! GDB 4.2 (see notes below, does not work?) If you have compiled some of these or others and had to jump through hoops, post it so we can include the directions here. ----------------------------------------------------------------------- 2-5. GCC 2.2.2 ----------------------------------------------------------------------- Gcc 2.2.2 is an older version of the GNU Gcc compiler. The directions are kept here because you will need an 'older' version such as= this one to compile some of the newer versions of Gcc. For some reason= the Sun supplied cc compiler loops indefinately when compiling newer versions of Gcc. Gcc 2.2.2 compiles without any problems on Sun386i with SunOS= 4.0.2: ./configure sun386i make LANGUAGESc make stage1 make CC"stage1/gcc -Bstage1/" CFLAGS"-g -O" make stage2 make CC"stage2/gcc -Bstage2/" CFLAGS"-g -O" make compare make CC"stage2/gcc -Bstage2/" CFLAGS"-g -O" install make install-fixincludes After you have done all this, you can remove (erase) the source= tree for gcc-2.2.2 and begin compiling a newer version of gcc. GCC Pointers... --------------- See GCC 2.5.8 =20 ----------------------------------------------------------------------- 2-6. GCC 2.5.8 ----------------------------------------------------------------------- To compile newer versions of Gcc (like 2.5.8) you will need an= older version of GCC (like version 2.2.2). For some reason the compile= of 2.2+ hangs when using the Sun supplied compiler. Gcc 2.2.2 is= the last version of Gcc know to compile with the Sun supplied cc compiler. ./configure sun386i make LANGUAGESc make stage1 make CC"stage1/gcc -Bstage1/" CFLAGS"-g -O" make stage2 make CC"stage2/gcc -Bstage2/" CFLAGS"-g -O" make compare make install CC"stage2/gcc -Bstage2/" CFLAGS"-g -O" Somehow 'setjmp.h' appears to get munged. I fixed it with the following changes to the private copy GCC keeps in: '/usr/local/lib/gcc-lib/sun386i/2.5.8/include/setjmp.h' --------------------------- CUT HERE ---------------------------- 47,53d46 < typedef int jmp_buf[_JBLEN]; < < /* < * One extra word for the "signal mask saved here" flag. < */ < typedef int sigjmp_buf[_JBLEN+1]; < 74a68,74 > > typedef int jmp_buf[_JBLEN]; > > /* > * One extra word for the "signal mask saved here" flag. > */ > typedef int sigjmp_buf[_JBLEN+1]; --------------------------- CUT HERE ---------------------------- GCC Pointers... --------------- ftp://cxc1.msu.edu/386i/software gcc-2.5.8.sun386i.bin.filelist gcc-2.5.8.sun386i.bin.tar.gz Sun386i version of GCC version 2.5.8. Compiled with GCC= 2.4.5. gcc-2.5.8.sun386i.setjmp-fix 'setjmp.h' fix file (including comments on the problem) ftp://prep.ai.mit.edu/pub/gnu/ gcc-2.5.8.tar.gz GCC 2.5.8 sources as distributed from GNU. ----------------------------------------------------------------------- 2-7. Ghostscript 2.6.1 ----------------------------------------------------------------------- Yes, you can get GhostScript 2.6.1 to work on the Road Runner! First you should get all the fixes that have been put out by the GhostScript maintainers (from prep.ai.mit.edu or some other GNU site). Apply these fixes with patch. Then you will need to fix= the file 'gdevmem.h' with the following patch: --------------------------- CUT HERE ---------------------------- *** gdevmem.h.~1~ Wed May 26 21:52:04 1993 --- gdevmem.h Mon Dec 13 13:39:30 1993 *************** *** 119,124 **** --- 119,125 ---- (byte **)0, /* line_ptrs (filled in by mem_open) */\ 0, /* invert (filled in for mono) */\ 0, (byte *)0, /* palette (filled in for color) */\ + 0, /* target */\ 0 /* memory_procs */\ } --------------------------- CUT HERE ---------------------------- If you don't use the patch you will mos likely get something that= looks like the following, when trying to display fonts (here using the= example program 'prfont.ps'): $ gs prfont.ps Initializing... done. Ghostscript 2.6.1 (5/28/93) Copyright (C) 1990-1993 Aladdin Enterprises, Menlo Park, CA. All rights reserved. Ghostscript comes with NO WARRANTY: see the file COPYING for= details. Loading Times-Roman font from ptmr.gsf... 199192 187947 0 done. GS>/Times-Roman DoFont Segmentation fault The rumor is the GhostScript people were notified. As of June= 10, 1994 when I fetched fix-04 this problem was still there. Perhaps= they will eventually fix it in the offocial sources. GhostScript Pointers... ------------------------ ftp://cxc1.msu.edu/386i/software ghostscript-2.6.1.sun386i.bin.tar.gz Sun386i version of GhostScript version 2.6.1, patched up= to fix #04, and 'gdevmem.h' fix. Compiled with GCC 2.5.8. ghostscript-2.6.1.gdevmem-fix 'gdevmem.h' fix file (including comments about the problem) =20ftp://prep.ai.mit.edu/pub/gnu/ ghostscript-fonts-2.6.1.tar.gz GhostScript fonts for use with GhostScript 2.6.1. ghostscript-2.6.1.tar.gz GhostScript 2.6.1 source code as distributed from GNU. ghostscript-2.6.1.fix-01.gz ghostscript-2.6.1.fix-02.gz ghostscript-2.6.1.fix-03.gz ghostscript-2.6.1.fix-04.gz GhostScript 2.6.1 source code fixes as distributed from= GNU. ----------------------------------------------------------------------- 2-8. GDB 4.7 (does not work) ----------------------------------------------------------------------- When compiling gdb-4.7 i get the following errors on files gdb/solib.c and gdb/sun386-nat.c (all others files compiled fine) cc -c -g -I. -I. -I./../include solib.c "solib.c", line 261: rtc_sp undefined "solib.c", line 264: N_COMM undefined "solib.c", line 269: n_un undefined "solib.c", line 269: n_strx undefined "solib.c", line 272: member of structure or union required "solib.c", line 272: cannot recover from earlier errors: goodbye! *** Error code 1 cc -c -g -I. -I. -I./../include sun386-nat.c "sun386-nat.c", line 208: unknown size "sun386-nat.c", line 209: unknown size "sun386-nat.c", line 214: PTRACE_GETREGS undefined "sun386-nat.c", line 214: inferior_pid undefined "sun386-nat.c", line 215: PTRACE_ARG3_TYPE undefined "sun386-nat.c", line 216: PTRACE_GETFPREGS undefined "sun386-nat.c", line 219: unknown size "sun386-nat.c", line 221: f_st undefined "sun386-nat.c", line 221: FP0_REGNUM undefined "sun386-nat.c", line 222: member of structure or union required "sun386-nat.c", line 222: cannot recover from earlier errors:= goodbye! *** Error code 1 >From what i could guess, in the first case the errors are due= to the non standard format of sun386 a.out files (coff ``mixed''= with a.out at least in the headers) but i could not find a satisfactory solution (except commenting out the affected code). The second one in did not completely understand. Apparently the is ``not'' included because of the #define _EXEC_= in tm-sun386.h. I tried to compile gdb because I had problems compiling gcc-2.3.3: first i got a warning from as when compiling cse.c (first pass, compiling with stock compiler): cc -c -DIN_GCC -g -I. -I. -I./config cse.c Assembler: cse.c aline 16087 : Warning: Table overflow: some optimizations= lost (Labels) Then it hanged when first using the generated cc1 to compile libgcc2.a it seems that cc1 is stuck in a cpu-bound loop: if [ -f libgcc2.ready ] ; then true; else touch libgcc2.ready;= fi rm -f tmplibgcc2.a ; ar rc tmplibgcc2.a ${name}.o; rm -f ${name}.o; done _muldi3 < Here it HANGS!!! (cc1 never= terminates) Any help would be GRATELY appreciated. [Unresolved - Stephen] ----------------------------------------------------------------------- 2-9. GNU EMACS 18.59 ----------------------------------------------------------------------- I had no problems compiling GNU Emacs on the 386i. I used GCC= 2.3.2 running on SunOS 4.0.2. Here is a simplifed explanation of how= to do it... - Get the compressed file (emacs-18.59.tar.z) and expand it into= a directory to compile under. [Note: .z is different than .Z,= the small one uses GNU's gzip program not the UNIX compress program] $ zcat emacs-18.59.tar.z | tar xvf - - Edit the makefile and change any directories you do not like... (I changed /usr/local/emacs to /usr/local/lib/emacs, cause= I did not want to clutter up my /usr/local directory) LIBDIR/usr/local/lib/emacs BINDOR/usr/local/bin MANDIR/usr/local/man - If you want an optimized editor, edit ./src/ymakefile and specify it by changing the CFLAGS line=20to reflect that.... change: CFLAGS C_DEBUG_SWITCH ... to: CFLAGS C_OPTIMIZE_SWITCH ..... - Copy ./src/config/h-dist to ./src/config.h and edit it to include 386i include files. $ cp src/config.h-dist src/config.h These lines replace the #include lines in config.h #include "s-sunos4-0.h" #include "m-sun386.h" You may want to check other settings in config.h for XWindow capability, that is if you want it. #define HAVE_X_WINDOWS - Make the sucker... (or just do the next step and install at= the same time) $ make all - Install it.... $ make install That's all there is to it. My version appears to work fine, I= have used it for editing, sending mail, etc. I have not tried some= of the news reading functions, etc. So you are on your own there, I imagine it works. GNU EMACS Pointers... --------------------- ----------------------------------------------------------------------- 2-10. PERL version 4.036 ----------------------------------------------------------------------- I give up, what do I have to tweak to get perl-4.036 to compile= on my sun386i? It dies looking to link with h_errno - In the Configure script. When you are prompted for additional libraries enter: "-ldbm -lm -lresolv" - When you run the tests, it'll complain about a missing program. Copy that missing program by hand to (presumably) /usr/local/bin, and re-run the tests. These two hacks were the only things I had to do to install perl-4.036 on my 386i with SunOS 4.0.2. PERL Pointers... ---------------- ----------------------------------------------------------------------- 2-11. XWindows version 11 release 5 ----------------------------------------------------------------------- XWindows 11 release 5 can be compiled on the roadrunner. A few importlant things should be said first; 1. Use GCC, the sun compiler will not work; Edit 'site.def' and define HasGcc to be "YES" #define HasGcc YES Edit 'sun.cf' and define the right SunOS 4.0.2, and other things. 2. Xload is messed up, it cannot find some 'kvm' routines. You= can hand edit the make file and add -lkvm on the SYS_LIBRARIES= line, and it should compile fine. SYS_LIBRARIES -lkvm 3. When you start you make, use -DNOSTDHDRS for the make; make World BOOTSTRAPCFLAGS"-DNOSTDHDRS" I have compiled almost every patchlevel of X11R5 up to patch #26 with this method and have had no problems other than those=20 mentioned. X11R5 Pointers... ----------------- ftp://cxc1.msu.edu/386i/software X11R5.26.sun386i.bin.tar.filelist X11R5.26.sun386i.bin.tar.Z X11R5 binaries for sun386i (and a filelist). =20 X11R5.26.sun386i.man.tar.filelist X11R5.26.sun386i.man.tar.Z X11R5 manual pages for X11R5. =20 ----------------------------------------------------------------------- 2-12. XWindows version 11 release 6 ----------------------------------------------------------------------- Problems building shared libraries with GCC caused my build to= die,=20 and I have not had the time to look into the problem. =20 Anyone get it to work yet?=20 X11R6 Pointers... ----------------- ---------------------------------------------------------------------- 2-13. GnuPlot 3.5 ---------------------------------------------------------------------- GnuPlot has 2 problems compiling on the 386i: =20 1. Header file inclusion is really messed up 2. The 386i enforces non-writeable data segments =20 Patches to version 3.5 of GnuPlot are available. The patches include= =20 commenting out the redundant (and erroneous) #includes of time= =20 structures, and a fix to gnuplot_x11 (where they try to sscanf()= a=20 constant character string). =20 ... under construction, the file below may not be available= ... GnuPlot Pointers... ------------------- ftp://sahsun.usmacs.maine.edu/pub/sun386i gnuplot-patches.tar.Z Patches to GnuPlot source files to make them compile=20 correctly on the sun386i. =20 ---------------------------------------------------------------------- 2-14. HP LaserJet Printcap Entries ---------------------------------------------------------------------- Printcap Entry: --------------------------- CUT HERE ---------------------------- lp|LaserJetIII:\ :pw#80:\ :lp/dev/pp0:\ :sd/var/spool/(null):\ :br#9600:\ :fs#06030:\ :fc#0300:\ :of/usr/lib/hplaserjet: --------------------------- CUT HERE ---------------------------- hplaserjet filter script: --------------------------- CUT HERE ---------------------------- #!/bin/csh -f # # @(#)hplaserjet 1.1 88/10/27 Sun Microsystems Inc 1988 # # # This is the output filter form use with HP laserjets I and= II. # It is called via the of entry in the generic_hp printcap= entry # All it does is send the escape sequence to the printer, so= that: # CR is mapped to CR # LF is mapped to CR LF # FF is mapped to CR FF /usr/bin/echo -n 'k2G' #enable wraparound /usr/bin/echo -n 's0C' /usr/bin/cat if ($status 0) then exit 0 else exit 1 endif --------------------------- CUT HERE ---------------------------- Note that I've substituted and for the actual characters in the script. Edit the actual two characters in each line before trying to use the script, of course. [Rick Emerson] ---------------------------------------------------------------------- 2-15. PPP (ppp-sunos4.1.pl6.tar.Z) on a 386i ---------------------------------------------------------------------- Here is a summary of the steps I had to go through to bring up= PPP on my 386i. 1) PPP Software: I used the ppp-sunos4.1.pl6.tar.Z package and installed it as= is. 2) setting up the modem: I have a Multitech II V32bis/V42bis modem, the only non-default set-up I required was to get it to use hardware RTS/CTS flow control. 3) dialing up: at work we have a Telebit netblazer, with a couple of dial-back modems. I use a combination of a shell script and tcl/expect= to dial up work, wait for the dial-back, run ppp (and then kill= the now spurious tip process which eats lots of CPU). Here's the shell script, ppp-start is the expect script and all= the rest is just messing around to kill the tip process(es) I use= to communicate with the modem, but which get confused and gobble= CPU when ppp is actually running. --------------------------- CUT HERE ---------------------------- #!/bin/sh find() { ps -ax | egrep " $1 " | egrep -v "egrep" |\ sed 's/^\([ 0-9]*\) .*/\1'/ } PPP`find ppp` TIP`find tip` if [ ! -z "$PPP" -o ! -z "$TIP" ]; then echo "ppp or tip already running, ppp: $PPP, tip: $$TIP" exit 1 fi ppp-start & while true; do PPP`find ppp` if [ ! -z "$PPP" ]; then break; fi sleep 30 done PIDS`find tip` echo "Killing tip processes: $PIDS" kill $PIDS exit 0 --------------------------- CUT HERE ---------------------------- Here's the expect script: --------------------------- CUT HERE ---------------------------- #!/usr/local/bin/expect spawn tip UNIX-mtechII expect "*connected*" send "ATZ\r" expect "OK" send "ATX1\r" expect "OK" # use RTS/CTS send "AT&E4\r" expect "OK" send "ATDT(your phone number)\r" for {} 1 {} { =20 expect "*Where:*" break } send "\r" expect "NO CARRIER" send "AT&E4\r" expect "OK" send "ATX1\r" expect "OK" set timeout 60 expect "*RING*" exec ppp mru 1500 192.5.254.52:192.5.254.47 /dev/cua0 --------------------------- CUT HERE ---------------------------- I'm not sure the second set of AT&E4 and ATX1 are really necessary, they got put in when I was debugging and never got taken out! As you've probably guessed this scripts are modelled on those supplied Karl Fox in the ppp release I used. I just preferred= to use expect in place of his chat program. So far, so good. IP routing, or put another way, now the real work/fun starts! My 386i, dredd, has two ip interfaces - ie0 (which is broken 'cos= I don't have an ethernet or a transceiver at home) and ppp0, after booting sunos gives me two routes: Destination Gateway Flags Refcnt Use Interface 192.5.254.0 dredd U 21 1480 ie0 loopback localhost U 24 181 lo0 The first of these will never work, cos there's no ethernet, but= all is well since I don't try and talk to anything else over the ethernet - or so I thought, removing it appears to trash yp!? Karl's suggestions for routed use two routes in the /etc/gateways file: net 0.0.0.0 gateway gatekeeper.ansa.co.uk metric 1 passive host dredd gateway dredd metric 0 passive on startup, routed then does the following: installs these routes and *deletes* 192.5.254.0 via ie0 it's a good job it deletes 192.5.254.0 via dredd over ie0 'cos= it won't work. Now here comes the snag, as soon as I try and use ppp0, routed re-installs my new routes (from /etc/gateways) and also re-installs the ie0 (or forgets to remove it if it is added by ip). Once the ie0 route is back, all my data vanishes down my ie0 interface which is good approximation to /dev/null. Or rather= all data destined for machines on the same subnet as me (i.e. at work) gets trashed, I can talk to the rest of the internet (see end). So, the solution I adopted is to avoid using routed and manaully install the routes I require as follows: /usr/etc/route -n add 0.0.0.0 gatekeeper.ansa.co.uk 1 /usr/etc/route add dredd dredd 0 /usr/etc/route -n delete 192.5.254.0 dredd This works and when running ppp adds its own route: gatekeeper.ansa.co.u dredd UH 0 0 ppp0 and at this point I can access the rest of the internet and all= the machines at work. I suspect routed/IP gets confused because my home machine is on= the same network (i.e. 192.5.254) as all the other's at work. Karl's setup at morningstar has all of the PPP machines on a differnet subnet. Given that IP has no way of telling that my machine is special (other than via its IP address) I'm happy to beleive this= is the case. Ideally I'd like to move our PPP machines onto a differnet subnet and then be able to use routed. Can anyone offer any advice as to whether this is a correct understanding of what is happening? The fun was experimenting with things like removing the 192.5.254.0 via dredd over ie0 route before adding the new routes and then= trying to use yellow pages. I even managed to blow away my X server a= couple of times by trashing its network from underneath it - a good job= I wasn't installing this for anyone else! [Cosmos Nicolaou] PPP Pointers... --------------- ---------------------------------------------------------------------- 2-16. SNAP Cannot Create New Users ---------------------------------------------------------------------- I can no longer create users with SNAP or from the login screen.= The Sun386 claims I should speak with my system administrator concerning the publickey file. Well. I AM the system administrator, and the manuals don't help much. The publickey seems like it's in fairly good shape. The= books also suggested I should use newkey. Done already - and with no results. Yes I have had this problem.. The way I found out what was= causing it was trial and error and many sleepless nights.. :) OK.. Have you edited ANY of the files that snap needs to change, so that a new user can be added.. These files include, passwd, group, etc... If you used either edit or vi on the files snap will no longer WANT to change or add to the files.. The way I found to get around all of this.. is to use sunview= and then the edit file option/program. Edit the passwd and group file and then save them back to the disk. For SOME UNKNOWN reason snap wants to see something on the= bottom of the file or the way the file has been saved before it will attempt to add or delete users... Hope this helps... Its the way I fixed snap on my machine... [Brett] ---------------------------------------------------------------------- Forget SNAP ! Setting up new users is trivial. 1. Edit /etc/passwd with "vipw". Add a line for the new user,= with no encrypted password. Make sure UID numbers are correct (i.e.= same as that user has on other machines wich may cross-mount the= user's home directory.) 2. Go to ~users, run the script "copy_home", suitably modified= to suit your needs (mine does a chown -R on the new user's home directory, among other things.). 3. Put a logical link for the new user into the /exports directory system if that user wants to mount his or her filesystem remotely. Put into /etc/exports. Run exportfs -a. (This could very easily be put into copy_home). 4. Edit /etc/aliases if mail is involved. Run "newaliases" (though= I think sendmail updates the compiled aliases files automatically anyway). 5. Log in as the new user, set up a temporary password with "passwd", test the account (e.g. mail). Note that the model users' home directory in ~users/defaults can= be edited to your heart's content. This procedure is much better in my view than running SNAP. And= you could automate quite a bit of it in the "copy_home" script. If SNAP is the only reason you keep YP running, get rid of both= ASAP! You're much better off learning about this stuff than being kept= in ignorance by SNAP. The 386i is quite a nice UNIX box without all= this YP/SNAP stuff intended once upon a time to quell the fears of UNIX-phobes in the business world. [Stefan Mochnacki] ---------------------------------------------------------------------- 2-17. Removing NIS YellowPages ---------------------------------------------------------------------- Temporarily remount your /usr partition write-enabled and rename /usr/etc/ypbind and /usr/etc/ypserv to something else (see Ch.14,= =20 p.369 of "System and Network Administration", Part No. 800-1733-10, Rev. A, 9 May 1988). Addresses and netmasks may need to be more explicitly stated in "ifconfig" statements and the like, and corresponding "if" and= =20 "fi" statements need to be commented out. [Stefan Mochnacki] 3-0. Hardware for the Sun386i ----------------------------------------------------------------------- 3-1. NVRAM Problems and Solutions (lots of problems) ----------------------------------------------------------------------- A Bad, or nearly bad NVRAM/TIMER chip causes many, very strange,= =20 problems on the 386i. The most common problems are listed below: * On power up, after the Auto booting in progress.... it conplains that "ethernet is not properly connected, boot can not continue." Usually a disk address and loading vmunix message will appear. * A claim by the system during bootup that something is newly connected to the ethernet port. =20 * Later in the bootup, because it has falsely concluded that there's something connected there, additional tweaking is done on the port. It is this that leads to the "ie0: init failed: no intr:" message. * Most conspicuously, volatility of the non-volatile RAM (NVRAM). * With the system up, it is easy to examine the content of NVRAM, using maybe od or some such. The timer is in the last few words. You should see it counting on successive examinations if it's working ok. * The system will die a strange, tortured death (beginning around six hours after bootup by one account). One by one, windows= would die or partially die, until there was nothing left. These problems indicate you will need to replace the NVRAM/TIMER.= =20 The exact part number is MK48T02B25 2Kx8 Zeropower (TM) / Timekeeper= =20 (TM) RAM by SGS-Thomson Microelectronics. It fits in a socket= on the=20 CPU board (no soldering required). After it's installed its content= =20 have to be initialized, of course. It's possible to boot up with= =20 garbage in most of the bytes, after which a procedure can be used= to=20 initialize all of it. The procedure and some helpful utilities have been posted to the= 386i=20 list, they are collected in SUN-NVRAM.TXT on some 386i ftp sites. NVRAM Pointers... ----------------- ftp://sahsun.usmacs.maine.edu/pub/sun386i/sun-nvram.txt Copy of NVRAM utilities created by John E. Kreznar. =20 ----------------------------------------------------------------------- 3-2. NVRAM Suppliers (Sun Battery) ----------------------------------------------------------------------- Solar Systems, phone #'s: 800-253-5764 if you are in the States or 206-869-9354. This is an aftermarket product and cost approximately $150 US= last year. They ship overnight COD. ... A good (cheaper!) supplier would be nice if people have them... ----------------------------------------------------------------------- 3-3. Sun386i Memory Expansion ----------------------------------------------------------------------- "It looks like the 386i SIMMs are 80 ns, whereas the spec for= the PC is 100 or 120ns. Does anyone know if these 386i simms will work= OK in the PC?" Yes; they should be OK. I have swapped SIMMs between my 386 PC= and the 386i model 250 several times. My 386i runs OK with 100ns SIMMs. [John ] Mine also works fine with PC SIMMs [Stephen Houser] ----------------------------------------------------------------------- 3-4. Is there a 486 upgrade for the 386i? ----------------------------------------------------------------------- Nope. The Cyrix was rumored to work, but no one will admit they have= one running. There were also rumors of Sun producing a 486 upgrade motherboard, to which one person replied: I have definately seen some (actually running) albeit within= Sun's offices. Apparently they made a few dozen prototypes and actually shipped a couple to really power desparate customers before= the Sparc project killed the whole intel based line. The product= was offered only as a CPU board swap for existing systems. As far= as I know the ones made all had some problems above and beyond what goes on with a normal 386i which may explain why nobody claims= to be running one today. [TEG@vicor.com] ----------------------------------------------------------------------- 3-5. Large SCSI disks... ----------------------------------------------------------------------- Larger drives will work fine on a Sun. You are still limited to 2GB filesystems (i.e partitions) by SunOS= =20 and Solaris. The only "gotcha" is that you cannot low level format= =20 this drive with the format program. The sd device driver times= out=20 before the format is complete and slams a scsi bus reset which= then=20 aborts the format on the drive leaving it in a unuseable state.= As=20 long as you don't format the drive you should be safe. So get= a pre-formatted drive, or format it on another Sun (Sparc?) that= does not have this problem. Here are some disks people are using in their RoadRunners (capacities are 'formatted' capacity): Seagate, ST702N (Wren V), 702MB (really 600MB) Have only used 400MB of it so far (in one partition). Used the SPARC format.dat entry. ST43400N (Elite 3), 3.15 Gig Have only been able to use 75 sectors per cylinder= =20 giving 2 Gig formatted capacity. The drive has= a variable number of sectors per cylinder [75-109?] Micropolis, 1908-15, 1381MB CDC, Wren IV, 330 Meg Wren V, 600 Meg Wren VI, 600 Meg =20 Fujitsu, 2623SA, 680 Meg M2263SA, 670 Meg M2266SA, 1.2 Gig M2624FA, 520 Meg Disk Drive Pointers... ---------------------- ftp://sahsun.usmacs.maine.edu/pub/sun386i sun-disks.txt Copy of 'format.dat' entries for large SCSI disks. = =20 ----------------------------------------------------------------------- 3-6. Running around topless (i.e. without a framebuffer/monitor) ----------------------------------------------------------------------- ... under construction ... --------------------------------------------------------------------- This is the end of the FAQ. Stephen Houser This is the end of the FAQ. Stephen Houser --========================_21701170==_ Content-Type: text/plain; charset="us-ascii" Stephen ______________________________________________________________________ Stephen Houser (N1QWF) http://csir1.usmacs.maine.edu/~houser University of Southern Maine Academic Computing Services 144 Luther Bonney Hall BITNET: houser@portland 96 falmouth Street Internet: houser@usm.maine.edu Portland, Maine 04103 AT&T: (207)780-4588