I just discovered another method that is easier than tunnels. Again, you have Router A connected to Router B:
A------(igrp /28)-----B-----(ospf /24)------C Instead of creating a tunnel between A and B, use frame relay interfaces. Configure either A or B as frame-relay intf-type dce and then configure your IP address on subinterfaces, one for each mask length you need. This is still a kludge, though. If you have a bunch of different masks then you're going to have a bunch of subinterfaces and burn up a few network prefixes. But, it works. I still haven't found what I think is a "good" way to do this yet. I've tried secondary IP addresses but that does not work reliably. Still thinking.... John ---- On Sat, 15 Dec 2001, John Neiberger ([EMAIL PROTECTED]) wrote: > The other issue if you try to use my tunneling method is that you'd need > a different tunnel for each mask length! My method works very well with > one or two masks. Beyond that it would just be a mess! > > Last night I was playing with a variation of the tunnel technique where > I put the tunnel in an entirely different major network. This would > cause the redistributing router to summarize to the major net before > advertising to the IGRP router. It works great except for the major net > that already exists on the IGRP router. For example.... > > A-------(igrp)-------B > > This link is 172.16.1.0/28. You then create a tunnel and assign it > 4.0.0.0/8. On B--which is also running OSPF-- you add network 4.0.0.0 > to IGRP and then redistribute OSPF into IGRP. B will advertise > 172.16.0.0/16 to A via the tunnel interface. Unfortunately, A ignores > it. I found no way to make A use that route. > > Perhaps I'm headed in the right direction but just on the wrong track. > ;-) I'm hoping this idea that doesn't work will spark an idea that does > work. > > Regards, > John > > >>> "Gregg Malcolm" 12/14/01 3:31:45 PM >>> > Chuck, > > Seems appropriate that you are due for some pain from the dentist > after > screwing up my day (and more than likely, my weekend) with this > question. > It is a very good question tho. Have been thinking about it for awhile > and > have it set up on my home lab. Obviously, if the masks were reversed > on the > routing protocols, it with be a trivial matter w/ an OSPF summary. > > How many routers are you using in this scenario? I am currently using > three > with the middle being the re-dist point (have 6 in my lab so I can make > in > larger). I recall the post from John N regarding the use of a tunnel > for a > situation like this. I believe the problem is that in this case it > would > require using a /27 mask in the IGRP domain. If the scenario calls for > only > /28 masks in IGRP, then this would be a violation. > > So, are the rules : > 1. No default-network > 2. No static > 3. No policy routing > 4. Only /28 in IGRP, /27 in OSPF > > Thanks, Gregg > > ""Chuck Larrieu"" wrote in message > [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > It occurs to me that there is another answer to this problem. > > > > So as a Friday Follies question: what is the other answer I came up > with? > > > > Remember, the IGRP domain is /28 the OSPF domain contains routes /27 > and > > shorter. You must assure reachability to all interfaces in the OSPF > domain. > > You are not allowed to use a default network or any static routes to > attain > > this end. > > > > for extra credit - make it funny. I will be needing a good laugh > after the > > dentist is through with me this afternoon. :-O > > > > Chuck > > > > -----Original Message----- > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf > Of > > > > Chuck Larrieu > > Sent: Wednesday, December 12, 2001 7:56 PM > > To: [EMAIL PROTECTED] > > Subject: The old "how to get routes into IGRP" question again - > possible > > [7:29021] > > > > > > (REPOST) > > > > I've been fighting with one of my practice labs the last couple of > days. > The > > problem is one of those OSPF to IGRP redistribution with a twist. The > IGRP > > domain is /28. So how to get those shorter /24 prefixes advertised. > Oh > yeah, > > you can't use the default-network command to create an IGRP default > route. > > > > So let me offer this possibility. > > > > IP local policy route-map > > > > the route map then goes something like this: > > > > route-map igrp-default permit 10 > > set default interface [whatever the interface is] > > > > I also suspect that set ip default next-hop x.x.x.x works also, but > at the > > time I was testing I hadn't thought through all the implications, and > my > > test failed. > > > > In any case, the local policy would have to be implemented on all > routers > in > > the IGRP domain. A bit of planning, then, is required. > > > > I found out something else that was interesting. Local policy packets > seem > > to have a particular way they are constructed. the first time I > looked at > my > > debug ip packet, the source address was one of my loopback > addresses, > which > > I was not advertising under IGRP. So of course my pings failed, > because > the > > distant end did not have a route back. So I deleted the loopback, > tried > > again, and this time the source address was a LAN interface, this too > not > > advertised under IGRP. I am assuming that Cisco has a hierarchy of > > interfaces. Usually a ping is sourced at the interface out which the > packets > > are headed. But for local policy, it was different. > > > > Any case, I am offering these observations for consideration. > > > > Wish I hadn't turned my routers off last night. Or I could gather > some > > screen shots. > > > > Chuck [EMAIL PROTECTED] > > ________________________________________________ Get your own "800" number Voicemail, fax, email, and a lot more http://www.ureach.com/reg/tag Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=29270&t=29270 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]