Date: Sat, 29 Jan 94 04:30:06 PST 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 #26 To: tcp-group-digest TCP-Group Digest Sat, 29 Jan 94 Volume 94 : Issue 26 Today's Topics: Circuit Cellar help Linux & Radio (2 msgs) Linux: another question... RE - mtu,mss,window TCP MSS and TCP WIN settings 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: 28 Jan 94 11:29:34 EST (Fri) From: Terry Bell Subject: Circuit Cellar To: TCP-Group@ucsd.edu Send email to 'info@circellar.com' for info on how to request files via email. -- Terry Bell N8HSP tbell@n8hsp.nshore.org n8hsp@n8hsp.ampr.org 44.70.4.10 ------------------------------ Date: Sat, 29 Jan 1994 02:10:14 EST From: "Eric Todd Budinger" Subject: help To: tcp-group@ucsd.edu -- Eric T. Budinger Dan's Domain 201-586-1223 budinger@ds.gen.nj.us Ham Central SysOp ericbud@ritz.mordor.com 1-201-398-4619 (voice) n2koj@w2xo.pa.usa.na 1-201-205-2134 (digital) ------------------------------ Date: Fri, 28 Jan 94 15:28:07 GMT From: Alan Cox Subject: Linux & Radio To: tcp-group@ucsd.edu Slackware is probably the best choice - nobody uses the latest gcc much yet because the libraries and tools dont match and the C++ part of gcc 2.5.x is a real problem at the moment. For radio info and Linux mail linux-activists-request@niksula.hut.fi X-Mn-Admin: join HAMS (or it might be HAM) Hardly anyone uses it ------------------------------ Date: Fri, 28 Jan 1994 17:24:28 -0500 From: "Brandon S. Allbery" Subject: Linux & Radio To: tcp-group@ucsd.edu In your message of Fri, 28 Jan 1994 15:28:07 GMT, you write: +--------------- | For radio info and Linux | | mail linux-activists-request@niksula.hut.fi | X-Mn-Admin: join HAMS | (or it might be HAM) | | Hardly anyone uses it +------------->8 Lots of us *read* it; however, it seems that the tcp-group, nos-bbs, and ka9q-unix mailing lists are used in preference to the Linux-Activists HAMS channel. (This may be because Mail-Net is so klunky. Someday someone will get smart and replace it with listserv...) There was a brief flurry of activity a few months ago that revealed a large number of readers. ++Brandon -- Brandon S. Allbery kf8nh@kf8nh.ampr.org bsa@kf8nh.wariat.org "MSDOS didn't get as bad as it is overnight -- it took over ten years of careful development." ---dmeggins@aix1.uottawa.ca ------------------------------ Date: Fri, 28 Jan 94 14:08:22 MET From: Pierpaolo Pernici Subject: Linux: another question... To: tcp-group@ucsd.edu Hello fm IW5DHE, Pierpaolo... Could someone tell me the best-choice between the linux-distributions !? I thought that Slackware was the best-choice but I've seen that in the dist-files there isn't the latest version of GCC... ;^) Is there a mailing-list for Amateurs_with_Linux !? (rgr Wampes, JNOS, Ping-Pong conversd, etc... etc..) Shold I install JNOS or Wampes !? Excuse me for all my questions... All the best, Pierpaolo -- ---------- ---------- Pierpaolo Pernici (IW5DHE) == Radio-GW Team - Pisa == Surface Address: Internet: iw5dhe@amsat.org Post Office Box nr.8 AmprNet: iw5dhe@gw.iw5dam.ampr.org I-56030 S.Pietro Belvedere (PI) Italy AX25-NET: iw5dhe@iw5cmm.#pi.ita.eu ---------- ---------- ------------------------------ Date: Fri, 28 Jan 94 12:14:46 EST From: crompton@NADC.NADC.NAVY.MIL (D. Crompton) Subject: RE - mtu,mss,window To: tcp-group@ucsd.edu Terry, I am sending this to tcpgroup - your return addresses ans cc's to were a mile long. The NOS (JNOS) manual has a section towards the rear that describes in detail the relationship of these parameters. In summary - Paclen is used for ax25/netrom connects MTU is used for TCP/IP connects and is the maximum data (and header) size that will be sent on an interface. MSS is a negotiated value between the server/client - it uses the lowest of the two and cannot be larger than MTU. This however does not mean that intermediate hops could have lower allowed thruput - this is what the "dicovery" techniques that Phil and other have been talking about is suppose to adjust for. The window is the maximum outstanding data - I.E. data in the pipeline. Typical values used here are based on our 2400 baud 2 meter circuit. MTU - 512 MSS - 472 window - 944 paclen 256 Paclen is limited by the maximum that MOST netrom sites can handle. For NOS to NOS connections this could be made larger - X1x sites allow a larger paclen - but be careful here as you may run into trouble. This is a good example of why mss/window and paclen per interface values make sense. Strictly speaking though you should be able to set your MTU per interface and set the MSS/window much larger and let it adjust. I.E. set you mss to the header size less than the maximum MTU on your system. If using ethernet - you might set MTU - 1500 on the ethernet port MTU - 512 on a 2400 baud local port MTU - 256 on a 1200 baud local port etc. and set MSS= 1470, window = 2940 In practice this is suppose to work - I am not sure if there may not be some problems in NOS. Phil may want to comment on this as I believe he wrote the description of it in the manual (way back) and he certainly knows the inter-workings of it in NOS. Doug ------------------------------ Date: Fri, 28 Jan 1994 11:38:30 -0800 From: Phil Karn Subject: TCP MSS and TCP WIN settings To: goldstein@carafe.tay2.dec.com >Raj describes that as a classical fallacy, albeit one with much >political clout. You can't solve congestion with bandwidth, just >move it. Now the backbone is fat but congestion occurs at the low- >bandwidth egress points ("on-ramps"). ATM nets will really see this. Agreed. Another effect of higher bandwidth is to remove the "usual" indication of congestion - increased delay. Since most of the packet switches haven't scaled their queue sizes proportionately to their link speeds, there's relatively little variation in queueing delay nowadays. Almost all of it is now speed-of-light delay, at least on the NSFNet backbone. You see little change as load increases until packets suddenly start dropping. This argues even more strongly for the DECbit scheme or something like it, since there's less that a TCP can use to infer that congestion is occurring. >It's simpler than that. The congestion window is in the receiver, >not the sender. All you need to do is lower the advertised window >size, not echo back the bit. That's an advantage of TCP over, say, Yes, of course. Why didn't I think of that? >That spare bit just sort of begs to be used for this, doesn't it? >I think somebody once suggested it at IETF but got nowhere, but then >I only know that as hearsay and I don't even remember from whom. It was probably me. I tried to talk this up at the IETF some years ago, but didn't get a lot of interest for some reason. Maybe this is something we could implement first in amateur radio, get a lot of experience with it, and then go to the IETF. You tend to get taken more seriously if you have more than just an idea, but an actual demonstration that your ideas work and work well. Phil ------------------------------ Date: Fri, 28 Jan 1994 06:54:25 -0800 (PST) From: resmana - petra christian university To: tcp-group@ucsd.edu SUBSCRIBE mailing list Resmana ------------------------------ Date: 28 Jan 94 12:56:13 GMT From: uucp@attmail.com To: tcp-group@UCSD.EDU >From UCSD.EDU!owner-tcp-digest Fri Jan 28 04:30:02 PST 1994 remote from internet Received: by gw1.att.com; Fri Jan 28 07:55:31 EST 1994 Received: from localhost by ucsd.edu; id EAA17272 sendmail 8.6.4/UCSD-2.2-sun Fri, 28 Jan 1994 04:30:06 -0800 for tcp-digest-list Received: from localhost by ucsd.edu; id EAA17266 sendmail 8.6.4/UCSD-2.2-sun Fri, 28 Jan 1994 04:30:04 -0800 for tcp-group-ddist Message-Id: <199401281230.EAA17266@ucsd.edu> Date: Fri, 28 Jan 94 04:30:02 PST From: internet!UCSD.EDU!tcp-group (Advanced Amateur Radio Networking Group) Errors-To: TCP-Group-Errors@UCSD.EDU Reply-To: internet!UCSD.EDU!TCP-Group Precedence: Bulk Subject: TCP-Group Digest V94 #25 To: internet!UCSD.EDU!tcp-group-digest TCP-Group Digest Fri, 28 Jan 94 Volume 94 : Issue 25 Today's Topics: GPIB and packet jnos gateway MTU, TCP MSS, TCP WIN New NET/Mac (hamradio TCP/IP) for the Macintosh TPC MSS and TCP WIN per port 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, 27 Jan 94 16:33:29 CST From: garnier@sol.med.ge.com (Dave Garnier x7-4286) Subject: GPIB and packet To: tcp-group@ucsd.edu *>FYI - *> *>The Circuit Cellar BBS has a parallel to GPIB adapter info and SW *> *>GPIBLPT.ZIP *I tried to look for it with archie but could not find any references. *Could you put it somewhere where I could ftp this file from. Could some kind soul post the fone number to this group? Thanks much! dave garnier wb9own ge medical systems, milwaukee garnier@sol.med.ge.com ------------------------------ Date: Thu, 27 Jan 1994 08:22:14 -0500 From: Gary Skuse Subject: jnos gateway To: TCP-Group@UCSD.Edu I am running jnos on a dos box which is connected to a landline by a serial port and to our University's extended LAN via an ethernet connection. I would like to set up a gateway between the two so that I can call in by telephone and get out over the Internet. I can figure out how to telnet but I can't ftp. Has anyone done this? I would appreciate any suggestions anyone has to offer. Eventually I would like to do something similar at home so that I can access my ethernet network at home via the telephone from work. tnx and 73, ka1njl ------------------------------ Date: Thu, 27 Jan 1994 10:31:25 -0800 (PST) From: uswnvg!tconboy@uunet.UU.NET (Terry A. Conboy) Subject: MTU, TCP MSS, TCP WIN To: uunet!ucsd.edu!tcp-group@uunet.UU.NET It seems that there is a minor confusion about these parameters. My (probably meager) understanding is that your MTU determines the biggest datagram that you send (allowing for 40 bytes of TCP/IP overhead) and that MSS and WIN control what is sent TO you. In other words, if you are sending mail or files to another host, your MSS and WIN have no effect on what you send out to them. I have a PC with a 1200 bps KISS TNC and a 2400 bps SLIP link. I set the MTU on the TNC to 296 since the radio path is sometimes marginal to some stations I communicate with. I set the MTU on the SLIP to 1064 to allow big datagrams to be sent to my computer at work over a 2400 bps dialup. I set TCP MSS to 1024 and TCP WIN to 2048 to allow up to 2 datagrams to be outstanding without an ACK. This works well in practice, since the congestion window (Cwind) typically gets adjusted down to the MTU-40 of the station at the other end of the radio link very quickly. So for me, having MSS or WIN per port isn't very important. However, it does require that the station at the other end of a radio link be intelligent in setting his MTU on the radio port. I've gleaned this from reading the docs, watching lots of traces, and looking at the 'socket nnn' displays. If there's a flaw in my reasoning, let me know. 73, Terry -- Terry Conboy email: tconboy@uswnvg.com U S WEST NewVector Group packet: n6ry@n7ipb.wa.usa 3350 - 161st Ave SE, MS 571 office: (206) 450-8388 Bellevue, WA 98008 fax: (206) 450-8399 ------------------------------ Date: Thu, 27 Jan 1994 10:13:03 +0100 From: adam@igg.tno.nl (Adam van Gaalen PA2AGA) Subject: New NET/Mac (hamradio TCP/IP) for the Macintosh To: ham-digital@ucsd.edu, tcp-group@ucsd.edu The Netherlands, January 25, 1994. Hello dear reader, Today I distributed the file NET_Mac2.3.35.sea.hqx... In this version of NET/Mac I implemented the following: - New parameter for 'perform' command - New: 'reboot_on_bad_date' } These new commands were implemen- - New: 'sourcewhendone' } ted to make it easier to maintain - New: 'resetsmtpto' } an unattended slip-telephone-link - Improved the generation of dynamic aliases - Some more 'enhancements' for q-and-d AppleTalk-fix - Some more 'enhancements' for q-and-d fix This version obsoletes all versions of info-mac/comm/radio-netmac in the Sumex-Aim archives. The new NET/Mac has been uploaded to ucsd.edu, to the directory /hamradio/packet/tcpip/incoming. If it's not there (anymore), then look at /hamradio/packet/tcpip/mac. Kind regards, Adam PA2AGA (e-mail: adam@igg.tno.nl ) ( or: pa2aga@igg.tno.nl for letters only, NO BIG files here) ------------------------------ Date: Thu, 27 Jan 94 07:04:23 CST From: Jack Snodgrass Subject: TPC MSS and TCP WIN per port To: tcp-group mailling list Someone posted a message or sent me a note saying that they had fixed JNOS to allow a MSS and WINDOW value to be asinged to each port. I forgot who it was, but they said that they'd post the changes this week. I'm anxios to see the code. I've tried setting the MTU on the port to less than the TCP MSS and I'm not happy with the results. Stuff seems to go smoother when I just reduce the MSS for the entire system rather than reducing the MTU for a single port. I see packets being broken up into mtu size pieces but NOS doesn't seem to re-assemble them and acknowledge them all the time. I'm not familiar enough with the process to figure out what's wrong, but I did see the same group of MTU sized packets be sent several times and never ack'd by the receiving system. Anyway... I'd like to mod the code to use a MSS and WINDOW based on the port. At least in my enviornment, stuff does not change from port to port so that should not cause a problem. Thanks. 73's de Jack - kf5mg Internet - kf5mg@kf5mg.ampr.org - 44.28.0.14 AX25net - kf5mg@kf5mg.#dfw.tx.usa.noam - home (817) 488-4386 Dialup - kf5mg@tcet.unt.edu - work (looking for) ------------------------------ End of TCP-Group Digest V94 #25 ****************************** ------------------------------ End of TCP-Group Digest V94 #26 ******************************