Multiple mutual redistribution points

pk · Guest

I admit that I'm purposely trying to create broken and/or unnecessarily
complex examples (to try to understand how things work), but this one I
really can't understand.

Topology:

+----+
| R0 +--------LAN0--
+-+--+
|
+---------LAN1------------+
| |
+-+--+ +--+-+
| R1 | | R2 |
+-+--+ +-+--+
| |
r1r3 r2r4
| |
+-+--+ +-+--+
| R3 +----- r3r4 -------+ R4 |
+----+ +----+

IP addresses:

LAN0: 192.168.0.0/24
LAN1: 172.31.3.0/24 (R1 is .1, R2 is .2, R0 is .3)
r1r3: 10.3.1.0/24 (R1 is .1, R3 is .3)
r2r4: 10.3.2.0/24 (R2 is .2, R4 is .4)
r3r4: 10.3.3.0/24 (R3 is .3, R4 is .4)

ospf is running on R0, R1 and R2 for LAN0 and LAN1 (the upper half of the
figure); RIP (v2, no auto-summary) is running on R1, R2, R3, R4 for links
r1r3, r2r4, and r3r4 (the lower half of the figure).

At R1, I do mutual redistribution between RIP and ospf, with these commands:

router rip
redistribute ospf 1 metric 5

router ospf 1
redistribute rip metric 10000 metric-type 1 subnets

The metrics are high to avoid route feedback.
At this point, R1 advertises RIP routes into ospf on LAN1, R2 learns these
redistributed routes and prefers them to the old RIP ones:

R2#sh ip route
[snip]
172.31.0.0/24 is subnetted, 1 subnets
C 172.31.3.0 is directly connected, Ethernet2/0
10.0.0.0/24 is subnetted, 3 subnets
O E1 10.3.1.0 [110/10064] via 172.31.3.1, 00:00:13, Ethernet2/0
O E1 10.3.3.0 [110/10064] via 172.31.3.1, 00:00:13, Ethernet2/0
C 10.3.2.0 is directly connected, FastEthernet0/0
O 192.168.0.0/24 [110/65] via 172.31.3.3, 00:00:13, Ethernet2/0

Of course, R1 routing table does not change:

R1#sh ip route
[snip]
172.31.0.0/24 is subnetted, 1 subnets
C 172.31.3.0 is directly connected, Ethernet2/0
10.0.0.0/24 is subnetted, 3 subnets
C 10.3.1.0 is directly connected, FastEthernet0/0
R 10.3.3.0 [120/1] via 10.3.1.3, 00:00:19, FastEthernet0/0
R 10.3.2.0 [120/2] via 10.3.1.3, 00:00:19, FastEthernet0/0
O 192.168.0.0/24 [110/65] via 172.31.3.3, 00:01:56, Ethernet2/0

No surprises up to this point.
Now, I do mutual redistribution also on R2, with the same commands used on
R1.
My understanding here is that R2 should redistribute ospf routes into RIP,
with a metric of 5. Thus, R1 ignores those routes since it has already
ospf, or connected, or better RIP routes to the destinations advertised
into RIP by R2.
But, in the reverse direction, R2 should NOT redistribute any RIP route into
ospf, since it has no RIP routes in its routing table (they were all
discarded in favor of ospf routes).
Contrary to what I thought, an ospf E1 route appears at R1:

R1#sh ip route
[snip]
172.31.0.0/24 is subnetted, 1 subnets
C 172.31.3.0 is directly connected, Ethernet2/0
10.0.0.0/24 is subnetted, 3 subnets
C 10.3.1.0 is directly connected, FastEthernet0/0
R 10.3.3.0 [120/1] via 10.3.1.3, 00:00:18, FastEthernet0/0
O E1 10.3.2.0 [110/10064] via 172.31.3.2, 00:00:14, Ethernet2/0
O 192.168.0.0/24 [110/65] via 172.31.3.3, 00:00:14, Ethernet2/0

That is, R1 changes its route to 10.3.2.0/24 (the network directly connected
to R2). The same happens if I make routes redistributed into ospf E2 (the
default).
The only explanation I can imagine for this is that somehow the
redistribution makes R2 advertise the 10.3.2.0/24 route into ospf. But how
can it do that, if it has no RIP routes? Does redistributing RIP routes
into ospf imply redistributing connected networks? Or is there some other
explanation?

Thanks

Gabriele Beltrame · Guest

"pk" <pk@pk.pk> ha scritto nel messaggio news:f800ii$ukf$1@aioe.org...

pk · Guest

Gabriele Beltrame wrote: