big problem, OSPF?FR?RIP?Redistribute?Split-horizon?

[samsam_wang]

R1 R2 R3 R4 connect over FR

R1 serial 0.1 to R2 serial 0(Point-to-Point), run rip ver1 between them.

R1 serial 0.2 to R3(s0),R4(s0) (multipoint), run ospf between them.

I can ping from R1 to R2, R3, R4

After run OSPF and RIP, everything is ok. redistribution is ok.

it is about 5 mins later or more, R1 can not ping R2, vice verse. R2 can see the route from R3 and R4 but can not ping. R1 can not see the route from R2.

some experts told me that is the problem of split-horizon, i try to enable it under serial 0 of R1, but it still doesn't work.

after i reboot the routers, everything is ok. then 5 mins or 10 mins later, the problem is still there.

anybody can help? thanks a lot.

[gleithner]

Could you post your configs? At first glance, I'm confused how R1 can't ping R2 when R1 should treat the route to the ser 0.1 subnet as a directly connected network(and R2 should do the same on it's end), regardless of any other routing protocols.

[samsam_wang]

I am strongly confident that this problem is not a routing loop, but I don't know why. I met this situation many times, every time i try to solve it, but i can not. I guess maybe it is the problem of split-horizon, because the phsical serial 0 of R1 disable split-horizon by default and RIP enable it by default. maybe, but I am not sure the answer. I try to enable under s0 of R1, and it still doesn't work.

thanks

[gleithner]

It's probably not split-horizon for a couple of reasons. First off, logical subinterfaces aren't affected by split horizon since they are treated as seperate int's. Secondly, OSPF isn't affected by split horizon. Finally, split horizon isn't going to affect two routing protocols running on different interfaces anyways..

still, it's puzzling why the route for the subnet of serial 0.1 would change from directly connected to anything else, unless the link is going down.

[ruwhite]

It sounds like from your description that you are running some routing on subinterfaces of a given interfaces, and running other routing on the main interface. I wouldn't do this--I would run them all on subinterfaces, or all on the main interface. In this case, you'd need to do all on subinterfaces.

Russ.W

[samsam_wang]

R1

s0.1 140.68.24.2/24

s0.2 140.68.200.2/29

R2

s0 140.68.24.4/24

R3

s0 140.68.200.5/29

R4

s0 140.68.200.6/29

R1 s0.1 vs R2 s0 run RIP V1 between them

R1 s0.2 vs R3 s0 and R4 s0, run OSPF(point to multipoint) between them, R1 is the hub.

R1,R2,R3,R4 have a lots of 140.68.x.x networks connected directly. the four routers connected with a FR switch.

redistribution between OSPF and RIP on R1

after that, everything is ok, i can ping anywhere to anywhere.

on R1, I shutdown the serial 0(not subinterface) then "no shutdown"

then R1 s0.1 and R2 s0 can not ping each other after 5 mins or more

[ziutek]

Try using this command to isolate from where routes are getting replaced:

debug ip routing

I've had some problems with RIP routes being replaced by OSPF routes because of AD. You might want to try putting the command distance 105 under RIP on R1. This would prevent any OSPF route from replacing the original RIP route.

Good luck,

Joe

[ipotts]

Please post the following from R1 before and after the failure.

sh run

sh ip rou | i 0.1

sh ip rou | i 0.2

sh ip rou | i 0.3

Please indicate which of the above routes the ping fails to?

[deilert]

can you post the configs from all routers .