Why doesn't OSPF redistribute secondary address unless "redistribute connected"?

Hello, everyone.

I make a redistribution experiment, about redistribution between OSPF
and EIGRP.

The topology is:

R2 (E0)--------------------(E0) R3 (S0)---------------------
(S0) R1

R2
E0: 192.168.33.2 / 24

R3
E0: 192.168.33.1 / 24
S0: 192.168.0.6 / 30

R1
S0: 192.168.0.5 / 30

OSPF run between R2 and R3, while EIGRP run between R3 and R1.
Redistribution is configured on R3.

Before redistributing, R1 does not know the network 192.168.33.0.
And then I configure basic redistribution commands, everything goes
well, network 192.168.33.0 is successfully redistributed into EIGRP
area, and is known by R1.
After that, I add a secondary ip address onto R3's E0, and add network
172.16.0.0 into OSPF process, that is

R3
E0: 192.168.33.1 / 24 172.16.0.1 / 16 (secondary)
S0: 192.168.0.6 / 30

router ospf 1
......
network 172.16.0.1 0.0.0.0 area 0
......

I thought the network 172.16.0.0 can also be redistributed into EIGRP,
but it does not appear in R1's routing table.
Very strange, if I add command "redistribute connected", 172.16.0.0
can be redistributed into EIGRP, and can be known by R1.

What's the the reason, please?

 

I've just read you post... It seems I got an answer for you...

I'm studying for my BSCI, just finished the "Manipulating Routing
Updates" chapter and this is what I read on my book...


"EIGRP and OSPF use an interface's primary IP address as the source of
their updates. They expect the routers on both sides of a link to
belong to the same subnet"


As you can see, that is something Protocol-related, not a Cisco's IOS
programming problem...

I hope this helps...


HACabrera

 

Split horizon. There is no need to advertise the secondary IP on E0 to other
routers on E0 because by the rules of IP addressing, ALL routers on E0 must
have all primary and secondary subnets identically defined.

Good luck and have fun!