[Xorp-users] Route redistribution...

[Bruce M Simpson]

Sproat, Jackalynn wrote:
> ...
> |R1|-10.1.0.5--rip---10.1.0.50-|R2|-10.10.0.50--ospf---10.10.0.5-|R3|
>
> R1 learns of the 10.10.0.x subnet via R2 exporting all "connected
> routes" into rip.
> However,  R2 does not by default advertise (export) it's "connected" rip
> subnet to its OSPF neighbors.  I.e., R3 doesn't have a route to 10.1.0.0
> network in routing table. Is this expected?
>   

Yup, 10.1.0.0/24 is not part of any OSPF area in the first set of 
configs, nor is any external route explicitly redistributed for it -- so 
there is no reason why it would appear in any LSAs.

> Therefore, I can either:
> ...
> 2. have R2 export rip into ospf, by adding the following policy:
> ..
> However, this didn't work. Why??? Do I have to set the OSPF "ebit" to
> true or false when exporting RIP entries to OSPF? Should export command
> be inside "area 0.0.0.0{}"?
>   

Area 0 has special meaning, i.e. "backbone".

OSPF backbone routers expect the E-bit capability in HELLO by default.

It looks like the policy you provided should work, however, off the top 
of my head, you wouldn't see 10.1.0.0/24 in RIP advertisements because 
that *is* the LAN that RIP is running on. Advertising a prefix *on* the 
LAN itself would clash with the other peers' "connected" route. If in 
doubt, capture packets.

> ...
> This worked. I.e., R3 learnt of route to 10.1.0.x via redistribution of
> connected routes into OSPF.  However, R1 no longer had route to
> 10.10.0.x eventhough redistribution of connected-to-rip was still
> configured.  Why is this?
>   

Did you remove the policy statement(s) which redist connected into RIP 
on R2?

10.10.0/24 is not part of any RIP config, and RIP needs to be explicitly 
told to advertise prefixes which weren't learned via RIP.

thanks
BMS