Path: hermes.fwi.uva.nl!fwi.uva.nl!sun4nl!EU.net!uunet!uunet.ca!uunet.ca!io.org!usenet From: kmedcalf@io.org (Keith Medcalf) Newsgroups: comp.os.os2.networking.tcp-ip Subject: Re: Routing with SLIP & Ethernet Date: 11 Oct 1994 13:17:50 GMT Organization: Keith Medcalf and Associates Lines: 86 Distribution: world Message-ID: <37e39u$4hh@ionews.io.org> References: <36ioic$t4u@geraldo.cc.utexas.edu> <2e965dde.45585452454d45@extreme.mayn.sub.de> Reply-To: kmedcalf@io.org (Keith Medcalf) NNTP-Posting-Host: zealot.net2.io.org X-Newsreader: IBM NewsReader/2 v1.02 In <2e965dde.45585452454d45@extreme.mayn.sub.de>, pilgrim@extreme.mayn.sub.de (Bjoern "Pilgrim" Karlowsky) writes: >Hi Kevin! >: ************************* ************************* >: * 128.97.97.10 (lan0)*===============* 128.97.97.30 (lan0)* >: * * ethernet * * >: * 128.23.35.20 (sl0) * * * >: ************************* ************************* >: || >: || >: || SLIP >: || >: || >: ************************* >: * 128.23.35.250 * >: * * >: * * >: ************************* > >: The current routing table for PC #1 is >: 128.97.97.0 128.97.97.10 lan0 >: default 128.83.35.250 sl0 > >: The current routing table for PC #2 is >: 128.97.97.0 128.97.97.30 lan0 > > >: Any help or information is appreciated. > >: Thanks, >: Kevin > > >Kevin Tapperson (kevintap@mail.utexas.edu) wrote: > >: The ethernet network works fine, and the SLIP connection works fine, however PC >: #2 cannot communicate with computers on the other side of the SLIP connection. >: (ie. packets are not being transfered between the ethernet (lan0) and SLIP (sl0) >: interfaces) > >: Is the problem with the routing, if so, how should this be routed. > >You have to publish an ARP entry onto the lan for the SLIP machine with the >hardware adress of the lan0 interface from PC#1. >You get the adress with netstat -n there. And do: >arp -s pub > You have violated rule uno of using SLIP -- you forgot to use a separate network or subnetwork for your SLIP network. It works exactly as it should work. You must not mix topology types within the same subnet. TCP/IP was not designed for such foolish setups. Alternatively, you can merely run dynamic routing on all the machines in the network, that is, detach routed -g -s on EVERY MACHINE ON THE NETWORK including making sure that RIP routing is active on all the routers in the network. Then, make sure that the GATEWAYS file on the node that connects to the badly formed SLIP node is correct, and that the GATEWAYS file on the SLIP node with the defective address is correct. And that the GATEWAYS file on the default router for your network accurately reflects the remote routers that access the slip nodes. Then all will work as you expect. Isn't Dynamic Routing wonderful? This is precicesely why it was invented. Isn't it cool that computers on a network can talk to each other and tell them how to route packets from one place to another? You can also make the defective SLIP node arpable by proxy, however then you will have to contend with the difficulties created by having inaccurate hop counts to the outside world. It may truncate the extents of the world a little, but that should not be too much of a problem, depending on how far you are from the rest of the internet. Your proper course of action is to allocate a non-arpable network number to use as you slip network. Then run routed on both sides of all slip connections. +--------------------------------------------------------------------+ | Keith Medcalf | Internet: kmedcalf@io.org | | Software Engineer | SnailMail: 275 Broadview Ave #152 | | Database and Client/Server | Toronto ON CANADA M4M 3H5 | | Application Systems | Telephone: (416) 465-9578 | +--------------------------------------------------------------------+