Date: Fri, 2 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 #191 To: tcp-group-digest TCP-Group Digest Fri, 2 Sep 94 Volume 94 : Issue 191 Today's Topics: encapsulation encapsulation and port #'s Encapsulations TCP-Group Digest V94 #189 TCP-Group Digest V94 #189 (NOS sizeups as a router) 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, 01 Sep 94 09:47:20 From: kz1f@RELAY.HDN.LEGENT.COM Subject: encapsulation To: tcp-group@ucsd.edu There are two rfc's that cover ipip encapsulation, rfc1226 and rfc1241, I believe it is 1241 that is authored by our very own Brian Kantor and deals with AX25 specifically. The other deals with generic ipip encap. I don't know to what extent the NOS implementation maps to these, except the port number is mentioned in the rfc as 93, not 94. Walt ------------------------------ Date: Thu, 01 Sep 94 10:09:30 From: jks@giskard.utmem.edu Subject: encapsulation and port #'s To: tcp-group@ucsd.edu So could some one unconfuse me (direct mail please) as to the history of RFC's on this and the current port 94 to port 4 switch. Jack KD4IZ (901) 448-6242 ------------------------------ Date: Thu, 1 Sep 1994 12:46:09 +0200 (BST) From: iialan@iifeak.swan.ac.uk (Alan Cox) Subject: Encapsulations To: tcp-group@ucsd.edu I raised the issue of IP encapsulation over AX.25 and possible mapping strategies in amongst some examples on the IPng discussion list. This lead to the suggestion someone ('how about you ?') ought to write this up and submit it. Now I'm happy to do this and post drafts onto here but I don't want to step on any feet, and it seems people like Phil or Brian ought to be the ones who have first option on doing this and also probably have a few better ideas 8) Thoughts ? Alan ------------------------------ Date: Thu, 1 Sep 1994 20:41:15 -0700 From: Phil Karn Subject: TCP-Group Digest V94 #189 To: jmorriso@bogomips.ee.ubc.ca >You've got to be kidding. I need to run KA9Q because it has an 'encap' >interface (tunneling IP over IP) for gateways, and I can easily hang a >KA9Q router. For example, a wayward RPC program can go nuts sending >out udp packets to portmap on a machine on the other end, instant >sieze up of the NOS router. Or (so I'm told) run xmaze through a KA9Q >box that's routing to a 56kbps tcp/ip packet network, one or two mazes >get drawn and the NOS box dies. Maybe NOS just doesn't like UDP. This doesn't make any sense, at least not if NOS is just routing the UDP packets. A router doesn't even look beyond the IP headers of the packets it routes. My NOS router at home stays up for months at a time, despite heavy daily use (from my BSDI box) for just about every protocol in the suite - Telnet, FTP, rlogin, X, NFS, DNS, NNTP, SMTP, you name it. Maybe the problem is that people are trying to add too much application cruft to NOS instead of using it as the simple low-end router it was designed to be. Phil ------------------------------ Date: Fri, 2 Sep 1994 02:17:06 -0500 From: "Milton D. Miller II" Subject: TCP-Group Digest V94 #189 (NOS sizeups as a router) To: karn@qualcomm.com >From mailfail@UCSD.EDU Thu Sep 1 23:59:40 1994 >Date: Thu, 1 Sep 1994 20:41:15 -0700 >From: Phil Karn >To: jmorriso@bogomips.ee.ubc.ca >CC: TCP-Group@UCSD.EDU >In-reply-to: (jmorriso@bogomips.ee.ubc.ca) >Subject: Re: TCP-Group Digest V94 #189 >>You've got to be kidding. I need to run KA9Q because it has an 'encap' >>interface (tunneling IP over IP) for gateways, and I can easily hang a >>KA9Q router. For example, a wayward RPC program can go nuts sending >>out udp packets to portmap on a machine on the other end, instant >>sieze up of the NOS router. Or (so I'm told) run xmaze through a KA9Q >>box that's routing to a 56kbps tcp/ip packet network, one or two mazes >>get drawn and the NOS box dies. Maybe NOS just doesn't like UDP. >This doesn't make any sense, at least not if NOS is just routing the >UDP packets. A router doesn't even look beyond the IP headers of the >packets it routes. My first response to the original posting is "ohh... he is talking about the problem that NOS never discards incoming packets, but instead buffers them up until they get routed and sent out. Even if there is not enough memory left to do the necessary copies to send a packet out." problem. One way to help reduce the problem is to limit the number of interrupt pool buffers to the reasonable burst rate for the incoming wire. The burst gets used up, but the rest are dropped until the queue is refreshed during the normal timer processing. Of course, not compiling the unneeded services helps also :-). milton -- Milton Miller KB5TKF miltonm@bga.com ------------------------------ End of TCP-Group Digest V94 #191 ******************************