Sun-Spots-Request@Rice.edu (William LeFebvre) (11/19/88)
SUN-SPOTS DIGEST Wednesday, 16 November 1988 Volume 7 : Issue 17 Today's Topics: ALM-2 problems The ALM-2 patches A bug with the yppasswd command problems with ie1 on 4/280 Problems with multiple TTY windows More strange `rshd` behavior on SunOS 4.0 Network problems and finger-pointing lpd account doesn't work (Sun3/3.2) 8mm 2.3 GB Video Backup Device Query A cheap SMD/NFS server? CDC Wren V on a Sun? Sun-2 w/more than 4Meg? Lear Siegler 4105/7 termcap? 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: Tue, 8 Nov 88 14:19:37 BST From: Colin Walls <ctw%CNS.UMIST.AC.UK@cunyvm.cuny.edu> Subject: ALM-2 problems We too have had ALM problems since we moved to SunOS 4.0, though not the ones mentioned by Nadim Tohme. 1. On exit from Emacs using a terminal connected to the ALM-2 the terminal will hang and requires a control-C to clear the terminal and allow typing again. Also attempting to stop Emacs with a control-Z and restart with an ``fg'' command causes the terminal to hang. 2. We have an asynchronous FTP attached to an ALM port. Since the installation of 4.0 we have had to reduce the speed of the line from 9600 to 600 Baud. The reason for this is that the we get constant checksums due to whole characters being lost. We do not get bits out of characters being lost. 3. Using kermit through an ALM line can make the line go into ``history'' mode. That is the character echoed to the screen is not the one that has just been typed, rather it is one that was typed about a dozen characters before. 4. Cat'ing a large file will occasionally hang in the middle of the file, typing a control-C is necessary to restart it. Also, typing a control-S to stop the file being listed and then control-Q to restart it does not work. The file can be stopped from listing and cannot be restarted again. We have had one lot of patches from Sun that appears to cure problem 3, the others are still with us. Ah well, it will soon be Christmas and Santa Claus may bring us a nice new ALM driver. Colin Walls Janet: Colin@uk.ac.umist DARPA: Colin@umist.ac.uk Uucp: ukc!umist.ac.uk!Colin Tel: 061-236-3311 x 2626 International: +44-61-236-3311 x 2626 ------------------------------ Date: Tue, 8 Nov 88 17:32:19 PST From: root@helios.ucsc.edu (De Clarke x2630) Subject: The ALM-2 patches About the ALM-2 patches: at last we received sufficient patches to make our ALM work properly. It was a war story, but now all 16 ports handshake correctly. Unfortunately, the patched kernel no longer handshakes correctly with ttya (and presumably b) on the cpu board; ctrl-s ctrl-q and 'more' are broken, the brokenness consisting of extreme character loss. So the story isn't over yet on the serial patches. Just thought y'all might like to know... de voice: 408-429-2630 fax: 408-429-2730 email: root@helios.ucsc.edu root@portal.bitnet ------------------------------ Date: Tue, 8 Nov 88 21:59 N From: <DIMITRI@CGEUGE51.BITNET> Subject: A bug with the yppasswd command We have found a bug (?) with the yppasswd change. I don't know if it has been reported before. If yes, then can somebody tell me if there is patch (binary distribution)? The bug has to do the matching of user login names: if a user login name is made up from just the first characters of another login name, then the yppasswd program is confused with the short login name. That is, if we have a login "abcdef" and another "ab", the user ab will have problems in changing his passwd. The yppasswd program will try to match his old passwd with the one of abcdef! If, there is no passwd set for abcdef, (or they happen to have the same!!) then (here come the nice point!) the passwd entry of user ab will be replaced with the passwd etry of user abcdef with the new passwd set. Thus, user ab will disappear!! Here is a description for reproducing the bug. Environment: SUN UNIX 3.5 On the YP master we have: the passwd file for YP in /usr/adm/passwd the /etc/passwd file redirects to the yp server. Set-up for the bug: Passwd entries in the master yellow pages passwd file (for us /usr/adm/passwd) t123::9123:1900:test for bug:/u1/test:/bin/csh t1::9100:1900:test for bug:/u1/test:/bin/csh Entries in the /etc/netgroup file (,t1,cuisunnet) (,t123,cuisunnet) Update the YP server. Execution: Login as t1. run % yppasswd Changing yp passwd for t1 Old yp passwd: <cr> New passwd: test123 Retype new passwd: test123 yellow pages passwd changed on cuisuna % whoami Intruder alert! % Have a look at the master YP passwd file. You see t123::9123:1900:test for bug:/u1/test:/bin/csh t123:tfEcEikzG.8Dg:9100:1900:test for bug:/u1/test:/bin/csh User t1 no longer exits! Other details: If user t123 did had a passwd, then t1 will not be able to change its passwd. The yppasswd program tries to match it with that of t123. Note that we have also a t1234 user with his passwd entry before t123 and t1, then it his passwd entry that yppasswd will use. Fix: Well, we do not have sources, but I think that the problem is in the string matching with the login names (the loop is exiting when all the characters of the active login name have been matched, without checking if the matched name has more characters) For now, what we do, is to place the shorter names first in the passwd file. Has anybody else found this bug? dimitri Dimitri Konstantas | UUCP : mcvax!cernvax!cui!dimitri Centre Universitaire d'Informatique | EARN : dimitri@cgeuge51.bitnet Univeristy of Geneva | EAN : dimitri@cui.unige.ch ------------------------------ Date: Tue, 8 Nov 88 14:32:21 PST From: cam@ucbarpa.berkeley.edu (Carol Martin) Subject: problems with ie1 on 4/280 The ie1 interface on one of our 4/280s will pulse hundreds of (what appear to be) broadcast packets onto the net in a fraction of a second. This happens at long and regular intervals. A neighboring 4/280 exhibits no irregular behavior of ie1. It's not the ethernet board itself or the transceiver, nor does it seem to be the software. The rev levels on the two 4/280 cpus are different: the offending one is a rev. 13, the other is a rev 12. Anyone seen anything like this? carol ------------------------------ Date: 8 Nov 88 22:41:06 GMT From: folta@tove.umd.edu (Wayne Folta) Subject: Problems with multiple TTY windows Summary: I had a friend who sepnt a day trying to get an application working with multiple TTY windows...The note is very obscure, the symptoms are that All of the windows' output goes to the last one opened (including any echoing of input). Wayne Folta (folta@tove.umd.edu 128.8.128.42) ------------------------------ Date: Tue, 8 Nov 88 17:28:42 EST From: seth@sol.ctr.columbia.edu Subject: More strange `rshd` behavior on SunOS 4.0 No, sorry, I don't know the answer, but I have a few more cases of what you are describing. [Re: Strange 'rshd' behavior..] Yes, you may be assured that host.equiv and .rhosts are set to what the Sun manuals appear to want. 1) Root on Sun (running 3.5) to any Sun running 4.0 Gives "Address already in use" - when non-root, works fine - when going to Suns running 3.5, works fine 2) Root on Sun (running 4.0) to io.ctr.columbia.edu running 4.0 Gives "Protocal error: closing connection" dumps core - when non-root, works fine - when going to other Suns that apparently have the same programs and files, works fine - I think its a 3/160 (most others are 50/60/110) with a color moniter attached. Seth Robertson Systems Manager seth@ctr.columbia.edu Telecommunications Research Laboratory uunet!columbia!ctr!seth Center for Telecommunications Research (212) 854-5619 Columbia University ------------------------------ Date: Tue, 8 Nov 88 17:11:20 PST From: root@helios.ucsc.edu (De Clarke x2630) Subject: Network problems and finger-pointing Finger-Pointing Madness Strikes Again! OK, this is the background: server: 4/280S 32MB running SunOS 4.0 (with ALM patches) named helios server: 4/110P14 8MB running SunOS 4.0 (SCSI shoebox) named phaeton client: 4/110 mono 8MB named phoebus helios -------- [DEC LB100] ----------|------------------ ThickNet | [DELNI-AA] ||||||| | | [DEMPR-AA] | ||||||| | ThinNet phaeton-----------phoebus--| phoebus boots from helios, but it mounts phaeton's /u (user filesystem) as its own. Only user directories are obtained from phaeton, in other words; the OS and all is from helios. There are several segments coming off the DEMPR, supporting diskful and diskless Suns; there are several MicroVAX II's on the DELNI. Problem: intermittently (but frequency increases until next reboot) phoebus hangs on rlogin or telnet from other TCP/IP hosts. Immediately after reboot, all is well; then the problem begins to recur; after about 6 hours your chances of rlogin are worse than 50/50. Behaviour varies interestingly with use of telnet vs rlogin and with the host from which it's attempted (bsd, SunOS, CMU telnet): sometimes the hang happens at the login prompt; or after login and motd, but before the first shell prompt; or after *the first shell command*. Sun says nothing useful, alas. After a month or so of email back and forth they point the finger at DELNI-AA, saying that they've heard that there's a ROM FCO for DELNI's and that it's probably the DELNI interacting with the 4/110. Unfortunately there are 8 other Suns, 5 of which are 4/110's, on that DEMPR cascaded from the DELNI, and they all work. And DEC says, "What ROM FCO?" Has anyone seen this problem? Do you have a clue? Please write to me if you do; I don't peruse the news often enough to rely on seeing your answer there. HELP! De Clarke voice: 408-429-2630 fax: 408-429-2730 email: root@helios.ucsc.edu system%bigvax.dnet@portal.ucsc.edu system@portal.bitnet ps -- sunspots is great and thank you rice for hosting it! ------------------------------ Date: Tue, 8 Nov 88 17:10:21 EST From: hsi!stepstone.com!aad@uunet.uu.net (Anthony A. Datri) Subject: lpd account doesn't work (Sun3/3.2) I've got an Imagen with the following printcap: lp|lpe:\ :lpd=/usr/lib/lpd:\ :lp=/dev/null:\ :hn=rolf:\ :rt=/usr/spool/lpe/resfonts:\ :qu#300:\ :af=/usr/adm/imprint-accnt/lp:\ :lf=/usr/spool/lpe/ilog:\ :st=/usr/spool/lpe/status:\ :sd=/usr/spool/lpe:\ :tc=imaenet: /usr/adm/imprint-accnt/lp never gets touched. Am I doing something wrong? ------------------------------ Date: Tue, 8 Nov 88 17:18:53 EST From: slcpi!poseidon!escott@uunet.uu.net (E. Scott Menter) Subject: 8mm 2.3 GB Video Backup Device Query Hi there. I know this has probably been beaten to death, but I've been off the list for a couple of months, so forgive me. We're evaluating a few Exabyte-style tape drives. One from Perfect Byte (Exabyte drive), one from Peripheral Devices (Delta Microsystems drive), and one from Unisun (??). Anyway, the latter is much cheaper than the rest, presumably because they provide no software at all (the first two provide their own drivers, and in the case of Peripheral, also a dd-like program that does buffering, called "bdd"). My question: anybody know under why I should pay over $1,000 more for a custom driver? Thanks. Anybody's who's interested in the results of my (informal) evaluation, has just to ask. Oh, and I support a SunSpots SICP (Special Interest Cocktail Party) at SUG. 8^) Scott Menter Sr. Network Administrator Shearson Lehman Commercial Paper, Inc. uunet!slcpi!escott ------------------------------ Date: Mon, 7 Nov 88 16:29:39 PST From: felix!arcturus.starwars.COM!dav@hplabs.hp.com (David L. Markowitz) Subject: A cheap SMD/NFS server? I have a manager who is looking for a cheap way to add fast NFS disks to his network. We want something fast enough to serve lots of reasonably powerful machines (386i/250's, 3/260's, etc) and feel that SMD is necessary (convince me otherwise, please!) for acceptable performance. We don't want to buy another 260/280 type machine, though. The cost is just too high (after adding a BIG disk). Don't tell me to save money on 3rd party disk, controller (a must for performance) and memory - we plan to do so. We were disappointed to find that the 4/110 can't do VME DMA. Isn't there a small machine out there that can host a Xylogics 753 or Ciprico 3200? We don't need 12 slots - three would do nicely. SunOS would be preferred, but something else similar that can speak NFS might be considered... Another possibility would be to add SMD disks to an old 3/160 he has. Would it be fast enough with a good controller? I know Sun recommends that servers be faster than their clients, but this won't be doing and OS/swap serving, only data (but lots of data). David L. Markowitz Rockwell International ...!sun!sunkist!arcturus!dav dav@arcturus.UUCP ------------------------------ Date: Fri, 28 Oct 88 17:37:07 PDT From: scubed!megatek!jay@harvard.harvard.edu (Jay Schlegel) Subject: CDC Wren V on a Sun? Anybody have experience formatting/analyzing/using a CDC Wren V disk (94221-190) on a Sun? I've been unable to successfully format and analyze using 4.0 format or 3.5 diag (on a sun3). As this disk has variable sectors per track, it's not possible to feed diag (or format) correct information regarding sectors per track. I also fear that the embedded controller may not be MD21 compatible. Any *HELP!!!* would be appreciated, thanks, -*jay Jay Schlegel --> Megatek Corp. --> jay@megatek or --> ucsd! hplabs!hp-sdd!megatek!jay -- If you can dream it, you can do it! -- W. Disney ames!scubed! -- What he said! ^^^ -- Jamaican Bobsled Team ------------------------------ Date: Wed, 9 Nov 1988 09:31-EST From: David.Maynard@k.gp.cs.cmu.edu Subject: Sun-2 w/more than 4Meg? I've never heard of anyone putting more than 4 Meg of memory in a Sun-2 system. For one thing, that is the physical limit when using a 2/120 card cage with 1 Meg boards. Is it possible to get more memory (say 8 Meg) by using a larger card cage and/or 3rd party 2 and 4 Meg boards? With the larger boards, do you still get 0-wait state performance due to the RAS/CAS trick? Also, has anyone seen the larger memory boards appearing on the resale market? I've heard some people give the argument that it isn't economical to put more than 4 Meg. But, it's got to cost less than replacing the monitor, disk controllers, cabinet, etc. with a Sun-3. Besides, if the Sun-2 breaks, I have a chance at fixing it. --- David P. Maynard (dpm@cs.cmu.edu) Dept. of Electrical and Computer Engineering Carnegie Mellon University Pittsburgh, PA 15213 --- Any opinions expressed are mine only. I haven't asked the ECE department or CMU what they think. --- ------------------------------ Date: Wed, 9 Nov 88 11:23:51 EST From: Loki Jorgenson <loki@physicsa.mcgill.ca> Subject: Lear Siegler 4105/7 termcap? In trying to setup the termcap for a Lear Siegler 7107 color graphics terminal, I find idiosyncrasies which defy simple correction. Rather than spend my time struggling with it, does anyone have one handy? Email to me or root@physicsa.mcgill.ca would be appreciated. Loki Jorgenson loki@physicsa.mcgill.ca Physics, McGill University Montreal Quebec CANADA ------------------------------ End of SUN-Spots Digest ***********************