Sun-Spots-Request@RICE.EDU (Vicky Riffle) (08/26/87)
SUN-SPOTS DIGEST Tuesday, 25 August 1987 Volume 5 : Issue 36 Today's Topics: Administrivia: archive server Re: File server troubles Re: Sun Disks Please save me from embarrassment, use the latest 'sdi' game Re: Strange ftp problems memory, disks, and maintenance? Sharable object libraries on a Sun-3? How do Suns hold up to heavy usage? NFS to support versions? 'zs1: silo overflow' messages? SIMMS for Sun 3/60? DECNET MAIL Server/Client for SUN using SUNLINK/DNA? Finite element analysis software to be offered on SUN workstations Emacs Icon (2) ---------------------------------------------------------------------- Date: Tue, 25 Aug 87 15:30:00 CDT From: William LeFebvre <phil@Rice.edu> Subject: Administrivia: archive server Well, I may regret making this announcement, because I know that it will be followed by a tremendous flood of mail to our site, but I believe that our archive server is now ready for use. To access the archive server, send mail to the address "archive-server@rice.edu", or the Usenet equivalent: "...!rice!archive-server". This server is essentially the same one used by Brian Reid for his mod.recipes archives at decwrl, and I thank him for making it available to us. This server is an automatic mail-responding program. It will (hopefully) answer your mail message with an appropriate response. To learn more about the use of the server, send a message to the above address that merely contains the word "help" (on a line by itself). If it can figure out a return address (which it should be able to do in almost every case), it will send out an answer within an hour after receiving your request. If you have problems with the server and want to send mail to a real person, use the address "archive-management@rice.edu". I'll keep a close eye on things for a few weeks and see how things work out. William LeFebvre ------------------------------ Date: Thu, 20 Aug 87 20:35:14 EDT From: dpz@styx.rutgers.edu (David P. Zimmerman) Subject: Re: File server troubles Bryan@sierra.stanford.edu says: When all 3 disks are being accesses, the file server crashes. Because of our user load, it has been crashing about once every hour. And, as you would expect, this often results in fsck being run and files lost. - Does anyone else have a similar file server configuration? If so, are you experiencing this kind of thing? - What can I do until the vapor-ware materializes? We have a 3/180 file server here that has 3 Eagles on two controllers, with a fair amount of access on each. It runs on SunOS 3.2, and gives us no problems. Tonight I'm going to remove one of the disks; then we will be running 2 Eagles, one per controller. Sun seems to think the problem only occurs when there are 2 Eagles on one controller. Any other ideas? We plan to buy another server, but that will take a while. An Eagle is a terrible thing to waste. I find it interesting that you have these problems on all three disks, since one of them actually has a controller all to itself. Because of this, I think that Sun's reasoning may be faulty. I would also rule out kernel problems, since we don't do anything here with the kernel that would make a difference with resource map allocation (which tells me that those rmfrees are a result of something else). Maybe the interrupt level switches on the drive controllers are giving the trouble? When we first got our 3/280, it used to crash on large NFS transfers until we took it apart and found that its Ethernet controller and disk controller were stepping on each other at the interrupt level, which a flick of some config switches fixed. dpz ------------------------------ Date: Fri, 21 Aug 87 14:56:17 EDT From: Root Boy Jim <rbj@icst-cmr.arpa> Subject: Re: Sun Disks Tim Bray (tbray@watsol.waterloo.edu) says: The New Oxford English Dictionary Project is grappling with the problem of putting the whole dictionary, rather than just pieces, on line. The dictionary is > 400 Mb in size. We want it in one file. We also plan Why not use 26 files (or more), one for each letter of the alphabet? Big files get accesses slowly because of indirect blocks. to have several different kinds of indices with a total size somewhat greater than that of the raw data. (Not in the same file with the dictionary). We currently have a 3/160 with some Eagles. We will likely be moving to 4/xxx next year unless somebody else catches up in the price/performance race. It seems clear that it is a bad idea to try to solve this problem with Eagles from sun. They are too small and too expensive. So, what's the solution? We are open to all sorts of radical ideas including WORMs. Well, okay, if you want to have it all in one file, why not just use an eagle *as a raw device*? That's right, (cd /dev;ln rxy1c oxford). Refer to the dictionary as /dev/oxford. Yow! Am I defined yet? Adv(Thanks)ance. Adv(Your Welcome)ance. (Root Boy) Jim Cottrell <rbj@icst-cmr.arpa> National Bureau of Standards Flamer's Hotline: (301) 975-5688 ------------------------------ Date: Thu, 20 Aug 87 23:15:35 PDT From: weiser.pa@xerox.com Subject: Please save me from embarrassment, use the latest 'sdi' game It has come to my attention that there are old copies of my 'sdi' game floating around out there. If your 'sdi' doesn't drop rocks with the middle button, or popup a pie menu when you press 'things to read', then please grab the new version with tons of changes. (The new version was first announced around April or so.) Available by anonymous ftp from mimsy.umd.edu, file sdi.shar.Z, or from net.sources.games. And thanks for all those postcards from satisfied users! -mark ------------------------------ Date: 24 Aug 87 11:57 EST From: mclach%lishakill.tcpip@ge-crd.arpa Subject: Re: Strange ftp problems We too have been having problems with ftp trying to make data transfers (get, put, ls etc). But the problem has been limited to two machines, one on a sub-net and the other on our main net, both are nd clients of the same gateway machine, while other machines on both nets are not experiencing any problems. We are running SUNos 3.2. It became possible for us to determine more specifically the circumstances under which the problem occurs when the same error message occurred in code written locally and executed on these two machines. The bind system call fails with bind: can't assign requested address when using the AF_INET/SOCK_STREAM domain and requesting the system to generate a port number (by specifying 0 (zero) in the sin_port field of the sockaddr_in structure). The problem surfaced when we replaced the ND75 (with several mods) kernel with the GENERIC kernel to bring up a 3/50. We have only experienced the problem when running the GENERIC kernel (installed in /pub.MC68020/vmunix). This problem has been reported to SUN, and we are awaiting a reply. Daniel R. McLachlan ARPA: mclachlan@ge-crd.arpa UUCP: seismo!rochester!steinmetz!desdemona!mclachlan US MAIL: GE Corp R&D, PO Box 43, Bldg 37 Rm 527, Schenectady, NY 12345 ------------------------------ Date: 12 Aug 1987 1519 PDT From: Keith Lantz <lantz@gregorio.stanford.edu> Subject: memory, disks, and maintenance? I've been putting together the facilities for the new Olivetti Research Center in Palo Alto, and before I sign on the dotted line, thought I'd get second (or third or fourth...) opinions on the following items: 1. Has anyone (else) had any experience yet with the Xylogics 752 controller and/or the CDC 9720 series 8" disks? As far as I know, Xylogics and CDC both just started shipping, BUT rumor has it Sun intends to go with both products in the near future. 2. I am being led to believe I really "need" 16 Mbytes of memory to feel "comfortable" in a typical r&d (programming) environment. Anyone care to argue that less will do nicely (especially since going with 12 Mbytes would save $2000 per 3/60), or provide any information as to how to get the necessary SIPPs for less from someone other than Sun? 3. Question #2 in the context of a workstation at home. Specifically, if I was willing to give up a bit of "comfort", might I be happy with a stand-alone Sun 3/50? How much is "a bit"? Do I really want something with more memory at home as well? 4. Any terrible things to say about the various second-source memory shops, e.g. Helios, Custom Memory, or Counterpoint? 5. I am being led to believe that, because file servers are disk-bound, it really doesn't pay to buy anything but a 3/180S as a file server--IF all it will be used for is a file server (i.e. not as a compute server, gateway, etc.). However, I believe all these statements were made in the context of servers with Multibus-based rather than VMEbus-based controllers. Do/will VMEbus-based controllers change anything; i.e. does the effective throughput improve markedly even though the maximum transfer rate of the disk itself remains the same? On the other hand, if a server is used as a general-purpose backend communications server (UUCP, etc.) as well as file server, is it then cost-effective to upgrade to a 280S (or beyond)? 6. Finally, I have many conflicting accounts as to whether Sun will install and/or maintain second-source peripherals. My latest reading from Sun (the third conflicting reading to date, albeit from different people) is that they WILL maintain second-source disks if and only if a) you let them install the disks--for a fee ranging from $850 to $1500, depending on who you believe--and b) you buy at least one disk from them--although 1 of 3 sources at Sun thought this wasn't necessary. Care to share your experiences in this regard? Thanks in advance, Keith P.S. ANYONE considering buying Sun equipment should be sure to get hold of the latest edition of the "Configuration Guide" from their friendly sales rep. It answers a LOT of questions (including addressing the issue of 180S vs. 280S), although unfortunately not those presented here! ------------------------------ Date: 19 Aug 87 16:13:05 GMT From: nosc!logicon!sloane@sdcsvax.ucsd.edu (Gary Sloane) Subject: Sharable object libraries on a Sun-3? We are running sun-3 computers with Sun OS version 3.2, soon to be version 3.4 if our salesman ever comes through. So much for being on Sun Micro's "priority" list for an OS upgrade. (Sorry, had to get the flame out). We have a situation where we must load several tasks into memory and not allow them to swap. Memory is limited, so it is desirable to cut down the size of the processes as much as possible. All of them use several libraries in common: libm.a, libc.a, libI77, U77, and F77.a, etc. Is there any way that these libraries can be loaded into memory once and be shared, instead of having each process keep it's own copy of the needed libraries? Also, when you link to a library (in the default method) does the entire library object get attached to the process, or does the loader only draw into the linkage the particular sections of the library it needs? Gary Sloane ...!nosc!logicon!uucp ------------------------------ Date: Thu, 20 Aug 87 19:03:36 EDT From: Hal Birkeland <hkbirke@media-lab.media.mit.edu> Subject: How do Suns hold up to heavy usage? How well do Suns survive long-term, heavy usage? I have heard rumors of power supplies turning into fireballs, keyboards self destructing, and mouse problems. Have these and other problems been fixed? There is a possibility that Suns will be purchased but these durability questions get in the way. Please respond by e-mail to save the net. -- hal hkbirke@mit-amt.mit.edu hkbirke@media-lab.media.mit.edu ------------------------------ Date: Thu, 20 Aug 87 23:42:37 PDT From: weiser.pa@xerox.com Subject: NFS to support versions? I have heard a rumor that future versions of NFS from Sun will support versions. I presume, or hope, that versions means there is a way to do things like request the most recent version of a file, or write a new version which is atomically locked so that no one else can accidently end up writing that version as well. These semantics are probably implementable, but slow at best, with the current NFS definition. Can anyone shed light on this rumor? Thanks. -mark ------------------------------ Date: Fri 21 Aug 87 17:19:25 N From: Niall Mansfield <MANSFIEL%EMBL.BITNET@wiscvm.wisc.edu> Subject: 'zs1: silo overflow' messages? Running SunOS 3.3 on a 3/260, sometimes I get lots of messages on the console saying zs1: silo overflow arriving every few minutes. The zs(4) entry says (more or less) that this means the silo is overflowing, which doesn't get us very far. What does it really mean, i.e. what can I do about it -- is it a dodgy terminal line, or what? (I reckon it's something like that -- I've tightened up all the terminal connections everywhere, and the messages stop coming.) Thanks, Niall European Molecular Biology Laboratory Meyerhofstrasse 1, 6900 Heidelberg, W. Germany (06221) 387-0 [switchboard] (06221) 387-247 [direct] ------------------------------ Date: Sun 23 Aug 87 13:07:53 PDT From: David Roode <ROODE@bionet-20.arpa> Subject: SIMMS for Sun 3/60? If these are the same as the Mac II and Mac Plus/SE 1 Megabyte SIMM, then there are already 3rd party suppliers--check in a Mac magazine. If I recall correctly, then Sun is already slightly less than Apple regarding list price. Does anyone know if these SIMMs are interchangeable? ------------------------------ Date: 24 Aug 87 15:31:20 GMT From: allen@pyr.gatech.edu (P. Allen Jensen) Subject: DECNET MAIL Server/Client for SUN using SUNLINK/DNA? I am trying to write a Client/Server program for interfacing to the DECNET mail server on a SUN using the SUNLINK DNA software. I am able to establish a logical link ($QIO IO$_ACCESS or DNA SES_LINK_ACCESS) but do not know the protocol. I have a few questions: 1) Has anyone already done this and are you willing to share the software ? 2) Where do I go to get information on the protocol that the DECNET MAIL server uses ? 3) Why didn't SUN provide this with the SUNLINK/DNA package ? Thanks, P. Allen Jensen Department of Civil Engineering - Georgia Institute of Technology Atlanta Georgia, 30332-0355 (gatech!gitpyr!allen) USA, Earth ------------------------------ Date: Wed, 19 Aug 87 16:33:27 PDT From: marleen@sun.com (Marleen Martin McDaniel) Subject: Finite element analysis software to be offered on SUN workstations MSC/NASTRAN FINITE ELEMENT ANALYSIS SOFTWARE TO BE OFFERED ON SUN-3 AND SUN-4 WORKSTATIONS MOUNTAIN VIEW, CA -- August 18, 1987 -- Sun Microsystems, Inc. and The MacNeal-Schwendler Corporation (MSC) announced today an agreement to convert the MSC/Nastran finite element software package to run on Sun-3 and Sun-4 technical workstations. The agreement will give mechanical engineers another powerful computer-aided engineering and design (CAE/CAD) system on Sun platforms for applications in many industries, including automotive, aerospace, biomedical and heavy manufacturing industries. Sun offers a broad range of mechanical CAE/CAD application software for its Sun-3 and Sun-4 families through key OEMs such as Computervision, Prime Computer, Schlumberger and Applicon, and through more than 220 leading third-party CAE/CAD products. The NASTRAN product joins Sun's third party program, Catalyst and will be listed in Sun's Catalyst catalog, which includes more than 1,100 hardware and software products. NASTRAN is a large scale, general purpose digital computer program that solves a wide variety of engineering analysis problems utilizing the finite element method. Analysis capabilities include static and dynamic structural analysis, material and geometric non-linearity, heat transfer, aero-elasticity, acoustics, electromagneticism and other types of field problems. MSC/Nastran's competitive advantages include functionality, performance, reliability, client support and service. The NASTRAN product will take particular advantage of the new Sun-4's floating point performance and memory options that allow expandability to 128 megabytes. "We're pleased to offer NASTRAN on a workstation with the level of price/performance of the Sun-4 supercomputing workstation," said MSC's president, Dr. Joseph Gloudeman. "We anticipate that extremely large problems can be analyzed very quickly for a system of this price and size." The MacNeal-Schwendler Corporation, headquartered in Los Angeles, markets computer software products for mechanical computer-aided engineering (MCAE) applications. The company's principal product is MSC/NASTRAN, a sophisticated computer program used by designers and engineers to analyze the strength characteristics and dynamic response of structures and products. MSC/NASTRAN is marketed world-wide and enjoys a very large installation base. MSC stock is traded on the American Stock Exchange under the symbol MNS. Sun-3 and Sun-4 are trademarks of Sun Microsystems, Inc. UNIX is a registered trademark of AT&T. NASTRAN is a registered trademark of the National Aeronautics and Space Administration. MSC/NASTRAN is an enhanced, proprietary version developed and maintained by The MacNeal Schwendler Corporation. FOR MORE INFORMATION Greg Xenakis Press Relations (415) 691-6543 Karen Ruppert The MacNeal-Schwendler Corporation Marketing Communications (213) 259-3830 ------------------------------ Date: Fri, 21 Aug 87 11:55:30 PDT From: alan@sun.com (Alan Marr, Graphics) Subject: Emacs Icon It does indeed exist. I cooked this up from the illustration in the "Unix Emacs" manual by James Gosling, 1983. James was amused to see it and approved of it. /* Format_version=1, Width=64, Height=64, Depth=1, Valid_bits_per_item=16 */ 0xFFFF,0xFFFF,0xFFFF,0xFFBF,0xFFFF,0xFFFF,0xFFFF,0xFE7F, 0xFFFF,0xFFFE,0xBFFF,0xF8FF,0xFFFF,0xFFFE,0x9FFF,0xE1FF, 0xFFFF,0xFFFE,0x1FFF,0xC7FF,0xFFFF,0xF555,0x5FFF,0x2FFF, 0xFFFF,0xD555,0x6FFC,0x5FFF,0xFFFF,0x5553,0xAFF9,0x3FFF, 0xFFFF,0x5557,0xAFF4,0x7FFF,0xFFFF,0x4007,0xAFC8,0xFFFF, 0xFFFD,0x1FEF,0xAF93,0xFFFF,0xFFFD,0xBFDF,0x5E47,0xFFFF, 0xFFFE,0x7FCE,0x38CF,0xFFFF,0xFFF6,0x7FF1,0x671F,0xFFFF, 0xFFFB,0x7FFF,0x9B3F,0xFFFF,0xFFFB,0x7EFF,0xECFF,0xFFFF, 0xFFFC,0xFEFF,0xF5FF,0xFFFF,0xFFF6,0xFEFF,0xFBFF,0xFFFF, 0xFFF9,0xFEFF,0x1DFF,0xFFFF,0xFFF9,0xFDFE,0xCDFF,0xFFFF, 0xFFFD,0xFBFD,0x86FF,0xFFFF,0xFFFD,0xFBFB,0xC37F,0xFFFF, 0xFFFB,0xF7F8,0x3F7F,0xFFFF,0xFFFB,0xF7FF,0xDFBF,0xFFFF, 0xFFFB,0xEFFF,0xE7DF,0xFFFF,0xFFFB,0xEFFF,0xFFE7,0xFFFF, 0xFFBB,0xEFFF,0xF7FB,0xFFFF,0xFF8B,0xEFFF,0xF7FD,0xFFFF, 0xFF63,0xEFFF,0xF7FE,0x7FFF,0xFF69,0xEFFF,0xF7FF,0xBFFF, 0xFEEE,0x77FF,0xEFFE,0xDFFF,0xFEDE,0xB7FF,0xEFFF,0x6FFF, 0xFF1D,0x8FFF,0xDFFF,0xB7FF,0xFFDB,0xB7FF,0xDFFF,0xF7FF, 0xFE27,0x79FF,0x3FFF,0xFBFF,0xF9C9,0x747C,0xFFFF,0xCBFF, 0xFBBE,0x6DBB,0xF03E,0xFDFF,0xFBBE,0x1B8F,0xCFFE,0xC6FF, 0xFBBE,0xDB67,0xFFFF,0x46FF,0xF7BD,0xE369,0xE01F,0x6EFF, 0xF44D,0xEADC,0x1AE7,0xB9FF,0xF3F1,0xF4DB,0x7DFB,0xFBFF, 0xF7FE,0x4B1B,0x7EFC,0xF7FF,0xF7FF,0x9B96,0xFEFF,0x2FFF, 0xF7FF,0xBDE6,0xFF77,0x5FFF,0xEFFF,0xDDF9,0xFF68,0xFFFF, 0xEFFF,0xDEFE,0xFF6F,0xFFFF,0xEFFF,0xE1FE,0xFF77,0xFFFF, 0xEFFF,0xEEFF,0x7F77,0xFFFF,0xDFFF,0xEEFF,0x7FB7,0xFFFF, 0xDFFF,0xEEFF,0xBFFF,0xFFFF,0xDFFF,0xEDFF,0xBFFF,0xFFFF, 0xDFFF,0xC1FF,0xDFFF,0xFFFF,0xBFFF,0xDDFF,0xDFFF,0xFFFF, 0xBFFF,0xDBFF,0xEFFF,0xFFFF,0xBFFF,0xD7FF,0xEFFF,0xFFFF, 0xBFFF,0xB7FF,0xF7FF,0xFFFF,0xBFFF,0x8FFF,0xFBFF,0xFFFF, 0xBFFF,0xBFFF,0xFBFF,0xFFFF,0xBFFF,0xBFFF,0xFDFF,0xFFFF, 0x7FFF,0xBFFF,0xFDFF,0xFFFF,0x7FFF,0x7FFF,0xFEFF,0xFFFF, 0x7FFF,0x7FFF,0xFEFF,0xFFFF,0x7FFF,0x7FFF,0xFF7F,0xFFFF Alan Marr. ucbvax!sun!alan [[ Now in the archives on "titan.rice.edu" as "sun-icons/am-emacs.icon". ]] ------------------------------ Date: Fri, 21 Aug 87 20:07:20 EDT From: dpz@paul.rutgers.edu (David P. Zimmerman) Subject: Emacs icon Hi, looking for an Emacs icon? My friend Andy Gaynor (aka Silver) did this one up, and I was so impressed with it that I trashed mine and snarfed his. dpz /* Format_version=1, Width=64, Height=64, Depth=1, Valid_bits_per_item=16 */ 0xFFFF,0xFFFF,0xFFFF,0xFFFF,0xFFFF,0xFFFF,0xFFFF,0xFFFF, 0xEAAA,0xAA30,0x0102,0xAAAB,0xD555,0x54F0,0x0601,0x5557, 0xEAAA,0xAB8C,0x2600,0xAAAB,0xD555,0x5383,0x1C00,0x5557, 0xEAAA,0xA502,0xB000,0x2AAB,0xD555,0x6D02,0xE004,0x1557, 0xEAAA,0x9904,0x400C,0x0AAB,0xD555,0x1304,0x4218,0x0557, 0xEAAA,0x8208,0x4010,0x02AB,0xD554,0x8210,0x40E0,0x2157, 0xEAA9,0x8200,0x4280,0x10AB,0xD555,0x0600,0x8380,0x0057, 0xEAB2,0x0401,0x8640,0x842B,0xD56D,0x0406,0x9CC1,0x6017, 0xEAA0,0x1418,0x8802,0x800B,0xD560,0x0C00,0x8005,0x0007, 0xEA00,0x3800,0xE005,0x0003,0xD400,0x6801,0x400C,0x0003, 0xEE06,0x2801,0x8000,0x0003,0xD18A,0xD802,0x0000,0x0003, 0xE147,0x9006,0x0060,0x0003,0xE262,0x10F8,0x00A0,0x0003, 0xE4A1,0x1008,0x0560,0x0003,0xE358,0x3010,0x0340,0x0003, 0xE2AC,0x2010,0x0060,0x0003,0xDD56,0x2020,0x1000,0x0003, 0xEAAB,0x2042,0x2800,0x0003,0xD555,0xA380,0x3000,0x0003, 0xEAAA,0xE080,0xC000,0x0003,0xD555,0x4129,0x4000,0x6003, 0xEAAA,0xC213,0x0000,0x4003,0xD555,0x422E,0x0001,0xA003, 0xEAAA,0xDC04,0x0001,0xCC03,0xD555,0xC618,0x0004,0x9203, 0xEAAA,0x8B00,0x000A,0x0203,0xD555,0x8D80,0x4026,0x0403, 0xEAAB,0x9AC0,0x4110,0x1803,0xD555,0xB560,0x4009,0x6003, 0xEAAB,0x2AB3,0xC042,0x8003,0xD555,0x555D,0x8221,0x4003, 0xEAAB,0x6AAA,0x8586,0x8003,0xD557,0xD555,0x0698,0x0007, 0xEABF,0xEAAB,0x0120,0x000B,0xD543,0xB555,0x8040,0x0017, 0xEABF,0xEAAA,0xC048,0x002B,0xD567,0x5555,0x6030,0x0057, 0xEAA7,0x6AAA,0xB000,0x00AB,0xD567,0x5555,0x5800,0x0157, 0xEACE,0x2AAA,0xAC00,0x02AB,0xD58C,0x1555,0x5600,0x0557, 0xEB0C,0x0AAA,0xAB00,0x0AAB,0xD7FF,0xFD55,0x5580,0x1557, 0xEFFF,0xFEAA,0xAAC0,0x2AAB,0xD7FF,0xFD55,0x5560,0x5557, 0xEFFF,0xFEAA,0xAA90,0xAAAB,0xD7FF,0xFD55,0x5531,0x5557, 0xEBFF,0xFAAA,0xAB4A,0xAAAB,0xD57F,0xD555,0x548D,0x5557, 0xEAAA,0xAAAA,0xAC0A,0xAAAB,0xD555,0x5555,0x5415,0x5557, 0xFFFF,0xFFFF,0xFFFF,0xFFFF,0xFFFF,0xFFFF,0xFFFF,0xFFFF [[ Now in the archives on "titan.rice.edu" as "sun-icons/ag-emacs.icon". ]] ------------------------------ End of SUN-Spots Digest ***********************