Sun-Spots-Request@RICE.EDU (William LeFebvre) (05/12/88)
SUN-SPOTS DIGEST Tuesday, 10 May 1988 Volume 6 : Issue 81 Today's Topics: Re: SLIP for Sun Re: Non-printing use of "lpr" queueing mechanism: MDQS Need suggestions ... quick SunOS 4.0 "features" multi-level domain names and more on SunOS 4.0 Need help with YP Server setup 3/60 as a server? ESTALE? dvitool? TeXsun? Send contributions to: sun-spots@rice.edu Send subscription add/delete requests to: sun-spots-request@rice.edu Bitnet readers can subscribe directly with the CMS command: TELL LISTSERV AT RICE SUBSCRIBE SUNSPOTS My Full Name Recent backissues are available via anonymous FTP from "titan.rice.edu". For volume X, issue Y, "get sun-spots/vXnY". They are also accessible through the archive server: mail the request "send sun-spots vXnY" to "archive-server@rice.edu" or mail the word "help" to the same address for more information. ---------------------------------------------------------------------- Date: Thu, 05 May 88 12:25:22 MST From: lyndon%ncc@uunet.uu.net Subject: Re: SLIP for Sun You should also consider obtaining the latest release of the internet code from 4.3BSD. It contains a much improved SLIP driver, and is already set up for installation on binary-only Sun's. The complete source code was posted to comp.bug.4bsd.fixes, and is also available via ftp from ucbvax and uunet. ------------------------------ Date: Thu, 5 May 88 13:01:52 PDT From: rusty%math.Berkeley.EDU@cartan.berkeley.edu Subject: Re: Non-printing use of "lpr" queueing mechanism: MDQS You might look into the MDQS software package. It was done by (I think) Doug Kingston at BRL. If you have access to the internet you can get it via ftp from vgr.brl.mil (192.5.23.6). MDQS stands for "Multi Device Queueing System". It provides a "many to many" mapping between devices and queues (as opposed to the one to one mapping that the BSD lpr provides). For example, with it you can have 1 printer queue drained by several printers. It's a very powerful generalized queueing system and as sort of a proof of how general it is they also supply a "batch" system where the "device" is the shell and the input is shell scripts. I used MDQS at a previous job (where compute intensive jobs were the norm) to provide a system whereby users each had their own queue of jobs, so that everyone had an easy way to ensure that they only had one cpu-bound job running at a time. ------------------------------ Date: 6 May 88 20:03:15 GMT From: sunybcs!fredonia!mazumdar@decvax.dec.com Subject: Need suggestions ... quick We are currently running 4.3BSD on a VAX-11/750. In an attempt to add a second cpu we are investigating Sun's. The Sun representative has suggested the following configuration Sun 3/140 server 516D 374 Mb disk 60Mb cartidge 16 terminal MUX + OS etc., Since this would be our first Sun system we would like to hear about the experiences of other sites running Suns in a multiuser environment before we make our decision. We are a four year college and our use would be solely for teaching. The problem is that our proposal is due in a week. So a fast response would be appreciated. Please include your telephone number and your address if your site is geographically close to ours. If you are a university/college we would really like to hear from you. Thanks. Jin Mazumdar (uucp:) ...decvax!sunybcs!fredonia!mazumdar >>> The following are for historical interest only <<< Dept. Of Math and C. S. State University of New York College at Fredonia Fredonia, N.Y. 14063 (716) 673 3459 [[ I don't know if this will get out in time to help you....but I'll try. --wnl ]] ------------------------------ Date: Thu, 5 May 88 14:41:26 EDT From: flanagan@lnssun2.tn.cornell.edu (Douglas Flanagan) Subject: SunOS 4.0 "features" Here at Cornell we have a University Software Support Agreement with Sun which lets us upgrade all of the Sun machines on campus to the latest version of SunOS. As the contact person at Cornell for this contract, I receive the new tapes and docs etc. Recently I received two very interesting letters from Sun. One was short but it's purpose was to tell me to expect SunOS 4.0 "in the weeks to come". Because the updates to the manuals are going to be "a complete set of revised documentation", they will be shipped separately from the tapes and may not arrive at the same time. So, it does appear that Sun is gearing up to start shipping 4.0 very soon. The other letter was titled "When Should I Upgrade to SunOS 4.0?" This one had a few suprises. It says: With the release of SunOS 4.0, Sun will update Sun FORTRAN and Sun Pascal compilers. Sun Fortran 1.1 includes VAX?VMS FORTRAN 4.0 extensions and Sun Pascal 1.1 includes global optimization and numerous other feature enhancements. Sun will ship these new compilers as separate products, rather than with the SunOS operating system. The original Berkeley derived compilers (f77 and pc) will no longer be available as part of SunOS. Well, I can see why they might not want to bother with f77 and pc any more, especially with all of the changes to the libraries. And maybe these compilers really are better (that would seem likely). And, under our software support agreement, they will ship us these compilers for free. BUT: 1) We will now have to pay an extra $50/month/language/one architecture plus $25/month/language/additional architecture for software support for these compilers. Sun does not have any special University software support agreement for these unbundled products -yet. That's one reason I am writing this. Support for these languages will now cost us as much as it does for the rest of SunOS. 2) Anyone who purchases a new machine which is shipped with SunOS 4.0 will now have to pay extra if they want FORTRAN or Pascal. I would not be complaining if Sun *replaced* f77 and pc with their own compilers as part of SunOS. But what they are doing is taking them *out* of SunOS and making you buy them separately. 3) Doesn't this look like a bad trend towards unbundling SunOS? It would a disaster if SunOS became what the System V releases for the 3B2 are like. It would become much more difficult to manage and use such a system. It's one of the things I hate about DEC and AT&T. I'm afraid the AT&T marketing critters are getting too much say in how Sun does things. The other thing that I'm now wondering about is NeWS. I had the impression, mostly from postings to usenet groups I guess, that SunOS 4.0 would come with NeWS as the windowing system and have Suntools and X11 implemented on top of it. Well, this letter makes no mention of it. All it says about new window system enhancements is "Industry standard text-menu terms and menu-selectable edit functions", and "mailtool with multiple reply/compose windows, incremental reading of new mail, and hierarchical folders". So, I guess NeWS will always be a separate product? Will Sun eventually drop suntools from SunOS and make us all buy NeWS? When will we be able to get a release of NeWS which includes X11 and suntools? Of course the above comments are my own opinion and I do not speak for Cornell, although I know many of our systems people share these concerns. My only source of information is this letter which I received, plus a few phone calls to Sun's Rochester, NY office. If I have my facts wrong or have mis-interpreted the letter, please correct me. The other items in the "Feature" list for 4.0: State-of-the-art virtual memory architecture Shared library facility File mapping (treating a file as part of virtual memory) Support for 64 open files per process Lightweight process library NETdisk support for diskless clients Resizable swap area for diskless clients Automatic mounting of remote file systems Secure RPC Secure NFS System V Release 3.0 Base System interface Complete System V STREAMS interface File System Re-organization suninstall utility Secure system capabilities (meets "C2" requirements) On-line disk formatting utility Optimizing compilers (Application CPU performance is increased by up to 20% due to global opt. in C comp.) -Douglas Flanagan Systems Programmer, Lab. of Nuclear Studies, Cornell U., Ithaca, NY 14853 flanagan@lnssun2.tn.cornell.edu ------------------------------ Date: Thu, 5 May 88 13:59:41 PDT From: roode@orc.olivetti.com (David Roode) Subject: multi-level domain names and more on SunOS 4.0 A fix for sendmail.cf to handle this appared in Sun Spots several months back. Here is the fix extracted from our sendmail.cf: ##### special local conversions S6 R$*<@$*$=D>$* $1<@$2LOCAL>$4 convert local domain R$*<@$*$=D.$=U>$* $1<@$2LOCAL>$5 or full domain name # test for domain.universe directly # $=U should work for olivetti.com but doesn't so use $U directly R$*<@$D.$U>$* $1<@LOCAL>$2 or full domain name R$*<@$*$=D.$U>$* $1<@$2LOCAL>$4 or full domain name You add the last two lines following the existing appearance of the first two. The problem this solves is that a destination host like pisa.orc.olivetti.com wasn't being recognized as being in the local domain when the local domain was orc located withing domain olivetti.com . SOME THINGS COMING IN SUNOS 4.0 At Monday's meeting of the S.F. BAYSLUG, SunOS developers mentioned that the sendmail coming out with SunOS 4.0 would have an improvement to make this kludge unnecessary. (Or so I interpreted their statement.) Also, it will neatly handle delivering all mail from the file server so that all clients can share a common set of mailboxes and none need run sendmail. Recipients will see the mail as coming from the file server and will be able to have replies delivered even if the particular client happens to be down. Also an Internet name server is going to be supported in SunOS with somewhat better integration with Yellow Pages. SunOS 4.0 is supposed to begin the distribution cycle next week, and they have already been running it on 1000 machines in house. Everyone has been mailed a letter announcing certain layered products for which support under 4.0 will be delayed. Virtual memory has been completely re-done in 4.0, with many effects, most pleasant. There are no more texts, so no more limit on the number that can be active, i.e. no text table full problem. There's also no more (nominally 10%) allocation of main memory to file buffers so any memory not needed for other things is used for file buffering. This general dynamicism of allocation improves performance of the typical single user diskless client. Trade-offs were made to use more memory and CPU to reduce I/O requirements. This means 4mb is an absolute minimum for reasonable operation of SunOS 4.0, which only affects Sun 2's. Swap space requirements are reduced. Page at a time sharing and copy-on-write are features of the new VM. Performace was improved 15% for diskless 3/50's and 25% for diskless 3/260's. Overall performance was a wash, but nearly everyone will see some benefits and some disbenefits. Perhaps the area suffering somewhat is multiuser configurations, but I theorize that addition of memory will alleviate this considerably. Shared libraries are a new feature. The disk space needed by the operating system files in general has been reduced as a result. Client implementation has been overhauled totally to use NFS instead of ND. To boot from a non-Sun NFS server, a client needs only TFTPBOOT support, a new RPC service dealing with configuration information, and standard NFS even a very old implementation. On Sun's servers, client space now lives inside a filesystem, hence inside a usable partition. Within that filesystem, sharing can take place between different clients. Swap space occurs to regular files, though they generally are pre-allocated. Allocations can be changed so long as space is available. One person described nearly running out of swap space, allocating a new swap file, doing a swap-on, and saving himself. To re-size an active swap file, only that client need be halted. Several OS filesystems formerly directly under the root have been moved under /usr to increase sharing by clients. Lightweight processes are another new feature, as well as automounting of remote filesystems (and autodismounting) and secure networking enhancements. There is increased System V support. There is online disk formatting, compartmentalized documentation sets and more. It slices, it dices, it chops, it grinds... [[ Hurry, this is a limited time offer. Not available in any store. Call before midnight tonight.... --wnl ]] David Roode Olivetti Research Center (415) 496-6243 ------------------------------ Date: Fri, 6 May 88 10:37:20 pdt From: pkb@wyse.com (Praveen Bhatia) Subject: Need help with YP Server setup I need to set up the following. Can anyone tell me how to do it. I need to have yp_server1 to provide 'services' for say passwd & group files, and yp_server2 for the other files (ethers, aliases,....). Thus when a yp_client needs to lookup the passwd or group data bases, the request goes to yp_server1 (directly or indirectly) and all other requests go to yp_server2 (directly or indirectly). Any help would be appreciated. Thanks. Praveen ------------------------------ Date: Thu, 5 May 88 17:51:23 ADT From: David Trueman <dalcs!david@uunet.uu.net> Subject: 3/60 as a server? Someone around here without news access is considering using a 3/60 with one or two 327 MB disks as a server for 4-6 diskless 3/50s. I have my doubts about the preformance of this configuration (to be used for software development). Is there anyone using a similar configuration who could comment on this? Thanks. UUCP {uunet utai watmath}!dalcs!david CDN david@cs.dal.cdn INTERNET david%dalcs@uunet.UU.NET ------------------------------ Date: Tue, 3 May 88 19:51:43 PDT From: ulysses!csw@ucbvax.berkeley.edu Subject: ESTALE? Running SunOS 3.2 on 3/280s and 3/60s Can someone please explain how and why one can gets the error ESTALE, "Stale NFS file handle"? I have an application that consists of two parts. One part runs on a file server constantly updating files by using rename(2) to move new files into well known places. Another part of the application runs on nfs clients looking at these well known files to see if their status has changed. If the status has changed, it rereads the files. The problem is that sometimes a stat(2) on the client on the well known file fails with errno=70. Since I'm using rename() on the server I expected the file to never be unavailable! Can anyone offer suggestions for how I might avoid getting the ESTALE error? Chris Warth ATT Bell Laboratories Murray Hill NJ {ihnp4|ucbvax}!ulysses!csw ------------------------------ Date: Thu, 5 May 88 14:05:29 MDT From: Michael Wester <wester@aleph0.unm.edu> Subject: dvitool? Does anyone have the source for dvitool and/or any better TeX previewer tools? We do have the source for dvisun (from a TeX distribution tape) but that isn't as nice as dvitool for which we only have an old (but working) binary? Michael Wester --- wester@aleph0.UNM.EDU (Internet), wester@unmb (BITNET) Department of Mathematics, University of New Mexico, Albuquerque, New Mexico [[ NOTICE: dvitool and the rest of the VorTeX software is NOT PUBLIC DOMAIN! It is also NOT REDISTRIBUTABLE! You must sign a license to legally obtain a copy. We have the sources to dvitool, but we cannot redistribute them. If you want more information about dvitool and VorTeX (the project of which dvitool is a part), send e-mail containing your postal address to "dist-vortex@ucbvax.Berkeley.edu". And, yes, dvitool is very nice. --wnl ]] ------------------------------ Date: Mon, 2 May 88 11:20:16 EDT From: formtek!pen@idis.UUCP (Philip E. Nickerson, Jr.) Subject: TeXsun? Where can I acquire a copy of TeXsun sources? -Phil Philip E. Nickerson,Jr. |UUCP {pitt,psuvax1}!idis!formtek!pen (412)937-4900|(800)FORMTEK| decvax!formtek!pen |Snail Formative Technologies, Inc., Foster Plaza VII | 661 Andersen Dr., Pittsburgh PA 15220 ------------------------------ End of SUN-Spots Digest ***********************