Date: Fri, 23 Sep 94 04:30:03 PDT From: Advanced Amateur Radio Networking Group Errors-To: TCP-Group-Errors@UCSD.Edu Reply-To: TCP-Group@UCSD.Edu Precedence: Bulk Subject: TCP-Group Digest V94 #210 To: tcp-group-digest TCP-Group Digest Fri, 23 Sep 94 Volume 94 : Issue 210 Today's Topics: gateway software? How to Join or Leave (2 msgs) Hubris Mail failure TCP retry time problem? wn940921 : where can I find it? Send Replies or notes for publication to: . Subscription requests to . Problems you can't solve otherwise to brian@ucsd.edu. Archives of past issues of the TCP-Group Digest are available (by FTP only) from UCSD.Edu in directory "mailarchives". We trust that readers are intelligent enough to realize that all text herein consists of personal comments and does not represent the official policies or positions of any party. Your mileage may vary. So there. ---------------------------------------------------------------------- Date: Thu, 22 Sep 1994 21:56:54 -0800 (PDT) From: jmorriso@bogomips.ee.ubc.ca (John Paul Morrison) Subject: gateway software? To: gateways@mpg.phys.hawaii.edu, tcp-group@ucsd.edu What software are people using on their gateways? I'm having terrible luck with the stability of KA9Q, Jnos etc. Did you compile your own? Do you have a huge patch file you keep around to apply to new versions of NOS? I don't care what flavour you're using; if it's stable I wan't to try what you're using. I need a *stable* tcp/ip router and gateway (ip over ip encap) for ethernet to 56kbps and 1200bps packet. Just tonight I was using rcp to copy 15 megs of a directory tree over the 56kbps link; but then I opened an ftp connection to a Sun; when I closed the ftp connection, the JNOS router bit the dust (right in the middle of the rcp! arg!). Rhetorical question: why can Suns, SGIs, Linux boxes etc. run for *months* doing routing (and a million other things), and NOS up times are measured in days at best? thanks --------------------------------------------------------------------------- BogoMIPS Research Labs -- bogosity research & simulation -- VE7JPM -- jmorriso@bogomips.ee.ubc.ca ve7jpm@ve7jpm.ampr.org jmorriso@ve7ubc.ampr.org --------------------------------------------------------------------------- ------------------------------ Date: Thu, 22 Sep 1994 15:33:57 -0500 (CDT) From: ssampson@sabea-oc.af.mil (Steve Sampson) Subject: How to Join or Leave To: tcp-group@ucsd.edu One of the most irritating problems of a listserver is that people don't remember how they got in, or the instructions for getting out. For this particular group, you would probably check the masthead of the tcp-group digest. Of course that won't help, because the masthead is wrong - always has been. The simple fact is that most listservers use the user name of listserv Those that don't are non-standard, and should probably be avoided if you have a poor memory. The tcp-group is archived on the ucsd.edu machine. So the full address is listserv@ucsd.edu Now here's the fun part. Just send the word help In the message. Don't include your damned 12 line signature block with stupid graphics either. If a member see's that another member is having trouple they can also unsubscribe them. delete-all joe@blow.edu -- Steve ------------------------------ Date: Thu, 22 Sep 1994 20:21:22 -0700 From: brian@nothing.ucsd.edu (Brian Kantor) Subject: How to Join or Leave To: tcp-group@ucsd.edu, ssampson@sabea-oc.af.mil In article <9409222033.AA08892@sabea-oc.af.mil> you write: >particular group, you would probably check the masthead of the tcp-group >digest. Of course that won't help, because the masthead is wrong - always >has been. Steve is so sure of himself. Unfortunately, he's wrong. The masthead of the tcp-digest clearly states to send mail to tcp-group-request@ucsd.edu, and by golly, that works for me! But then, Steve knows better than I do. Hey, I just wrote the software, the masthead, and run the system. So do it his way. Or as the instructions say. Your choice. Gosh, I wish >I< were young enough to know everything. Hubris, I think they call it. - Brian ------------------------------ Date: Fri, 23 Sep 1994 04:32:32 -0500 (CDT) From: ssampson@sabea-oc.af.mil (Steve Sampson) Subject: Hubris To: tcp-group@ucsd.edu > Hubris Isn't that what the fags at ucsd smear their butt with? Well you should pat yourself on the back on the fine and wonderful list server you "wrote". I suspect it's probably the only thing you ever finished... -- Steve ------------------------------ Date: Thu, 22 Sep 94 13:47:00 edt From: Adminstrator Subject: Mail failure To: dayhub!3445a!ucsd.edu!TCP-Group@prowler.daytonoh.NCR.COM User mail received addressed to the following unknown addresses: GPPGDAYTON/GPPGPOST/lshannon ------------------------------------------------------------------------------ Return-Path: Received: from prowler.daytonoh.ncr.com by gppgpost.daytonoh.ncr.com id <2E81C331@gppgpost.daytonoh.ncr.com>; Thu, 22 Sep 94 13:47:29 edt Received: by prowler.daytonoh.ncr.com; 22 Sep 94 13:30:34 EDT Received: by dayhub.DaytonOH.NCR.COM; 22 Sep 94 13:30:23 EDT Received: by 3445a.DaytonOH.NCR.COM; 22 Sep 94 13:30:46 EDT id AA780255289 Thu, 22 Sep 94 13:34:49 EST Date: Thu, 22 Sep 94 13:34:49 EST From: TCP-Group@ucsd.edu Message-Id: <9408227802.AA780255289@WPDSMTP.DaytonOH.NCR.COM> To: tcp-group-digest@ucsd.edu Subject: TCP-Group Digest V94 #209 Received: by ccmail from 3445a.DaytonOH.NCR.COM >From ncrhub1!ucsd.edu!owner-tcp-digest@dayhub.DaytonOH.NCR.COM X-Envelope-From: ncrhub1!ucsd.edu!owner-tcp-digest@dayhub.DaytonOH.NCR.COM Received: by 3445a.DaytonOH.NCR.COM; 22 Sep 94 13:27:45 EDT Received: by dayhub.DaytonOH.NCR.COM; 22 Sep 94 13:27:01 EDT Received: from ncrgw1 by ncrhub1.NCR.COM id as24192; 22 Sep 94 12:23 EDT Received: by ncrgw1.NCR.COM; 22 Sep 94 10:42:15 EDT sendmail 8.6.9/UCSD-2.2-sun Thu, 22 Sep 1994 04:30:09 -0700 for tcp-digest-list Received: by ucsd.edu; id EAA26135 sendmail 8.6.9/UCSD-2.2-sun Thu, 22 Sep 1994 04:30:07 -0700 for tcp-group-ddist Message-Id: <199409221130.EAA26135@ucsd.edu> Date: Thu, 22 Sep 94 04:30:06 PDT From: Advanced Amateur Radio Networking Group Errors-To: TCP-Group-Errors@UCSD.EDU Reply-To: TCP-Group@ucsd.edu Precedence: Bulk Subject: TCP-Group Digest V94 #209 To: tcp-group-digest@ucsd.edu TCP-Group Digest Thu, 22 Sep 94 Volume 94 : Issue 209 Today's Topics: Mail failure UNSCRIBE wn940921 memory and things Send Replies or notes for publication to: . Subscription requests to . Problems you can't solve otherwise to brian@ucsd.edu. Archives of past issues of the TCP-Group Digest are available (by FTP only) from UCSD.Edu in directory "mailarchives". We trust that readers are intelligent enough to realize that all text herein consists of personal comments and does not represent the official policies or positions of any party. Your mileage may vary. So there. ---------------------------------------------------------------------- Date: Wed, 21 Sep 94 13:34:00 edt From: Adminstrator Subject: Mail failure To: dayhub!3445a!ucsd.edu!TCP-Group@prowler.daytonoh.NCR.COM User mail received addressed to the following unknown addresses: GPPGDAYTON/GPPGPOST/lshannon ------------------------------------------------------------------------------ Return-Path: Received: from prowler.daytonoh.ncr.com by gppgpost.daytonoh.ncr.com id <2E806E8E@gppgpost.daytonoh.ncr.com>; Wed, 21 Sep 94 13:34:06 edt Received: by prowler.daytonoh.ncr.com; 21 Sep 94 13:30:09 EDT Received: by dayhub.DaytonOH.NCR.COM; 21 Sep 94 13:29:48 EDT Received: by 3445a.DaytonOH.NCR.COM; 21 Sep 94 13:26:24 EDT id AA780168631 Wed, 21 Sep 94 13:30:31 EST Date: Wed, 21 Sep 94 13:30:31 EST From: TCP-Group@ucsd.edu Message-Id: <9408217801.AA780168631@WPDSMTP.DaytonOH.NCR.COM> To: tcp-group-digest@ucsd.edu Subject: TCP-Group Digest V94 #208 Large message has been converted into an attachment. ------------------------------ Date: Wed, 21 Sep 1994 21:11:15 -0400 (EDT) From: Admin Subject: UNSCRIBE To: uunet!aquin!men2a!uunet.uu.net!ucsd!tcp-group UNSCRIBE ------------------------------ Date: Wed, 21 Sep 94 14:54:51 EST From: BARRY TITMARSH Subject: wn940921 memory and things To: TCP-GROUP , wnos-group Just to confirm unless informed otherwise. the memory problem on this version of wnos is cleared. and the latest uploaded .exe's do contain netrom. however I compiled both in 286 mode on the compiler. Barry. ------------------------------ End of TCP-Group Digest V94 #209 ****************************** ------------------------------ Date: Thu, 22 Sep 94 07:27:00 -0000 From: mikebw@bilow.bilow.uu.ids.net (Mike Bilow) Subject: TCP retry time problem? To: nos-bbs@hydra.carleton.ca [Replying originally to a post addressed to the nos-bbs@hydra.carleton.ca list] On 94 Sep 20 at 10:52, Dr. Gareth D. Blower wrote: DGDB> rto = backoff(tcb) * (4 * tcb->mdev + tcb->srtt); DGDB> Where does that factor of 4 get plucked from? It means DGDB> that when we calculate the retry time, we give the mean DGDB> deviation of the RTT 4 times the weight of the RTT itself. * * * DGDB> If we do a PING to a particular station, we will see DGDB> quite a small mean deviation, but that's because we send DGDB> one packet at time. When we are doing an FTP, we may send, DGDB> all in one go, any number of packets, depending on the DGDB> negotiated window size. If we send one packet, and then DGDB> two contiguous packets, and then three contiguous packets, DGDB> the RTT will vary. For one packet, it will be DGDB> approximately RTT. If two packets are sent contiguously, DGDB> the RTT for the first packet will be approximately 1.5 DGDB> RTT. If three packets are sent contiguously, the RTT for DGDB> the first packet will be approximately 2 RTT What you say has quite a lot of validity, but I think the reasoning was most likely to concoct a quick and dirty mechanism to prevent the retry timer from expiring on one packet while others are still being sent. This is an inherent problem on a half-duplex channel, since the RTT can only be measured on a packet-by-packet basis. It might be reasonable to extend Karn's Algorithm to say that we will only use an RTT measurement taken from an acknowledgement of the last transmitted frame. If we did this, then we would never get any RTT measurements unless we drop into stop-and-wait mode somehow, either through the use of the congestion window or some other cause. You might ask Phil Karn about this directly, although I believe that he has traditionally expressed objection to any linear backoff implementation as opposed to exponential backoff. -- Mike ------------------------------ Date: Fri, 23 Sep 94 11:19:52 utc From: iw1cfl@ik1qld-10.ampr.org Subject: wn940921 : where can I find it? To: tcp-group@ucsd.edu Hello all, where can I find wn940921 sources other than ucsd.edu? The problemi is that here the connectionf with ftp.ucsd.edu is slow, so if I can connect another host maybe i can download it easily. Best 73 Mike ------------------------------ End of TCP-Group Digest V94 #210 ******************************