Return to BSD News archive
Xref: sserve comp.os.386bsd.questions:13371 alt.winsock:10637 Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!agate!howland.reston.ans.net!news.sprintlink.net!uunet!news1.digex.net!digex.net!not-for-mail From: dkrapf@access1.digex.net (Don Krapf) Newsgroups: comp.os.386bsd.questions,alt.winsock Subject: Re: Routing Problem Date: 24 Sep 1994 04:01:03 -0400 Organization: Express Access Online Communications, Greenbelt, MD USA Lines: 69 Message-ID: <360mbv$73q@access1.digex.net> References: <35kaui$ph9@herald.indirect.com> NNTP-Posting-Host: access1.digex.net In article <35kaui$ph9@herald.indirect.com>, Thomas D. Simes <tsimes@bslnet.com> wrote: >Hello all! > >We are having a puzzling routing problem at our site - at least I >think that it is a routing problem. Here is the situation: >We are running a 386bsd 1.1.5.1 box as an Internet node with >modems attached for dial-in service. This bsd box is attached >to the net using a router and CSU/DSU running frame relay to >our provider. Both the router and bsd box can ping each >other and the net without a problem. The problem comes >when I SLIP into the bsd machine using Trumpet winsock 1.0a. >While the SLIP'd machine and the bsd box it is attached to can >ping each other, the SLIP'd machine cannot see our router or >the rest of the net (naturally). So far I have tried the >following to no avail. The problem is that nothing on your local net knows that it must send packets to the SLIP gateway in order to get them to the destination. When the router (for example) wants to send a packet to one of the SLIPped in machines, it looks up the machines IP address in its routing table and gets a match on the network number which tells it that the destination is on the local net. Then it arps for the destination but the arp goes unanswered. You can fix this in any of three ways: 1 Let the SLIP gateway proxy arp for the SLIPped machines. I run MoringStar PPP on my SCO SLIP gateway. It lets me automatically run a script whenever a SLIP/PPP connection is started or stopped. In the start script I do "arp -s ..." to add the remote machine to the arp table and in the stop script I do "arp -d ..." to remove the remote machine from the arp table. 2 Use a seperate subnet for the SLIPped machines. By putting the SLIPped machines in there own subnet you can do subnet routing and let the SLIP gateway advertise a route to it. Remember that the addresses with a subnet part of all ones or all zeros are reserved so that you must have at least two bits of subnet which leaves you with only 6 bits for the host. This give you two useable subnets with up to 62 hosts each. To lose a smaller part of your address space to the reserved subnet numbers you can make the subnet field wider but then you end up with a lot of little subnets. 3 Give each machine on your local net a static route through the SLIP gateway for each remote machine. While this is probably not a practical solution, it is a good way to check that I've correctly diagnosed the problem. SLIP into your SLIP gateway and verify that the remote machine and the gateway can communicate with each other. Then, go to the router and give it a specific route for the remote machine through the SLIP gateway. It will use this preferentially over the network route. Your remote machine should then be able to communicate with the router and with the internet beyond. Don