Cisco 2600 scenario

C

Captain Kirk

Limp Gawd
Joined
Dec 22, 2002
Messages
288
The scenario:
Here's an odd little scenario with a question I have at the end. Mind you, this is in a lab type environment. I have five Cisco 2600 series routers in a rack, four have one fast ethernet, the fith has two. All routers have two serial interfaces. Now picture them in a stack. Serial 0/0 on router one (top-most router) is plugged into serial 1/0 on router three. Router two has no serial cables connected. The rest of them have their 0/0 "mapped" to the 1/0 on the router below them.

These routers have /30's in 10.0.0.0/24 connecting them. So router one's serial 0/0 is 10.0.0.1 and router three's serial 0/1 is 10.0.0.2. The routers also have their fast ethernets connected to a Catalyst switch in the default vlan. f0/0 on each router is assigned a single IP in 172.17.1.0/24 (172.17.1.1-5). Router two is our "special" case, he has no serial connections, just a fast ethernet (172.17.1.2).

All routers are running RIP, and can see each other in cdp neighbors as well as by ping.

The question:
So now, I'm taking router five (bottom-most router) and plugging his f0/0 into a network. Now this isn't my network, so don't kill the messenger ;). Router five's f0/0 IP is now 10.60.108.X, with a gateway of 10.60.110.1 and a netmask of 255.255.252.0. Router five can ping the outside world via 10.60.110.1.

Now here's the question. I've given all the upper routers router five as my gateway and the next hop in the default route. When attempting to access the outside world from the upper routers I see they get to router five, but no further.

Why does this happen :confused:? I'm thinking router five somehow sees itself as the last hop and the traffic dies there.

Any help in answering this is greatly appreciated. Please feel free to ask me any questions about this setup. I won't be able to paste any router output until another week or so (BTW).

Thanks :)!
 
thats very confusing to me without drawing it out, but i'd say that you need to redistrbute your default route.
 
I believe you are running into a limitation of RIP v1 in that it does not support discontiguous networks.

Here is a link that describes RIP

Here is an excerpt:
Probably the largest problem with RIP is that it makes classful assumptions about the routes it is passing. This means that all subnets must use the same mask and must be contiguous. This is wasteful of addresses and does not neatly match with modern networks.
Click to expand...

You are probably going to need to switch to RIP v2 to make this work as is. Other options are changing your networks to be contiguous or changing routing protocols and redistributing routes (such as OSPF or EIGRP... but not IGRP because it doesn't support discontiguous networks either)
 
oakfan52 said:
thats very confusing to me without drawing it out, but i'd say that you need to redistrbute your default route.
Click to expand...
Here you go:
cisco.png

The left side of the router "object" represents serial 0/0 while the right side represents serial 1/0. The pink serial link is in pink simply to make it easier to see.


Bean Dip said:
You are probably going to need to switch to RIP v2 to make this work as is. Other options are changing your networks to be contiguous or changing routing protocols and redistributing routes (such as OSPF or EIGRP... but not IGRP because it doesn't support discontiguous networks either)
Click to expand...
Stupid question, but, if it's getting to the internet router (R5), shouldn't R5 just send the packet along its gateway of last resort (10.60.110.1)?
 
OK, so you have your routers doing RIP, all have addresses in the 10.xxx range.

You then hook a "external" router up to your R5, it has a 10.xxx address.

OK, first thing, R5 will not tell the other routers about this route, as its in the same subnet as it is (RIP is classfull, so 10.xxx is 10/8 or a class A network) as far as RIP is concerned.

It would be like, R5 would tell R2, hey, I see 10.xxx over here. R2 would then go, yeah, so, so do i......and then nothing hapens.

Use a different routing protocol (if i can recommend, OSPF, its worth learning) that supports classes or VLSM routing (they are the same thing, Variable Length Subnet Mask and classless)

Cheers
 
OK the picture helps and I believe Kristo is right.

R5 sees all 10.xxx.xxx.xxx networks as being on the same network. Therefore when a packet hits him that is actually destined for an outside network he drops it instead of forwarding it. Basically to verbalize it:

R5 - "I received a packet destined for a network that I don't know. But since 10.60.xxx.xxx is on the same network I don't know why you sent it to me. I will send an ICMP redirect back to the router that sent this to me and drop it because there is a better route than sending it through me. Next time just send it straight to the 10.60.xxx.xxx device."

On the other hand if a packet originates on R5 then he will send it out because it is in his routing table.
 
Thanks for the help guys :). I thought thinks seemed fishy when "show ip route" had 10.0.0.0/8 connected via RIP.
 
You must log in or register to reply here.
Back
Top