One point two-way redistribution

[polo123456]

Hello,

According to the below schema, Eigrp100 is redistributed into OSPF100 and OSPF100 is redistributed inside Eigrp100.

I read that one point 2-way redistribution is safe and that there cannot be routing loops.

Why routing loops cannot occur ?

Cause of Eigrp split-horizon ? But what about OSPF ?

Eigrp routes redistributed into OSPF will be marked as "O E2". How OSPF knows that those routes must not be redistributed (back) into Eigrp ?

Regards.

[Mohamed Sobair]

Hello,

One-Way redistribution as indicated by your example between two routing protocols is always Safe!

Look at your example, R2 is Performing Mutual redistribution between Eigrp & OSPF. There is no need for routing Protocol Structure Or Routing Protocol Routing Loop Prevention Mechanisms in such Example. Why? Because there is no two exit points for the router performs the redistribution to the same destination. And another reason there is no a Backdoor link connecting the Edge routers together in this case (R1 & R3).

In such Example, there wont be any sort of routing Loop because R2 is always going to have One Specific Network Installed in the routing Table and Prefered by the Router Itself by Looking at the IGP Administrative Distance Value (The Lower the Better).

So, the Answer, you have Eigrp Redistributed into OSPF at R2, R2 will always Prefer the Eigrp Direct Route By default because of the AD=90 , R3 will recieve it as OE2 routes , However, It will never be back into Eigrp , Because R2 is always Having it as Internal Eigrp Routes Installed.

Make Sense!! Did you get it yet?

Regards,

Mohamed

[polo123456]

No, I still misunderstand something.

What if we use RIPv2 instead of EIGRP ? OSPF administrative distance will be prefered over RIPv2.

R1 will learn OSPF routes via R2 and won't be able to advertised those routes back to R2 thanks to RIPv2 split horizon.

R3 will learn RIPv2 routes via R2 but which OSPF mechanism will forbid R3 to advertised back those routes to R2 ?

Cause thoses routes (OSPF administrative distance) will be prefered over RIPv2 routes and it will create a loop.

Regards,

[milan.kulik · Accepted solution]

Hi,

I believe it's the "Advertising router" field in an LSA type 5 header which would prevent such a routing loop.

See https://tools.ietf.org/rfc/rfc2328.txt

point 16.4. Calculating AS external routes:

....

"(2) If the LSA was originated by the calculating router itself, examine the next LSA."

I.e., ignore it.

HTH,

Milan

[polo123456]

Very nice, thank you Milan.

Hope you don't read the full RFC to find it out

Regards.