I wanted to share a very interesting tidbit i figured out the hard way.
Again, its about NSSA areas in OSPF. When redistributing a route into an NSSA area, how is the forwarding address chosen?
Some say the lowest IP address of any loopback interfaces thats advertised into the NSSA area. Others say the highest. Which one is correct?
The answer is suprisingly: Neither!
The question then becomes, how DOES it choose it?
Well, first of, its correct that the loopback interface has to be advertised into the NSSA area to begin with. Secondly, OSPF chooses the interface that was LAST configured. Weird? yes. Fact? yes.
Take a look at this topology:
Now take a look at R3’s output of interfaces in the OSPF routing protocol:
From this we can see that the first interface configured is Fa0/0, then Lo0, Lo100 and finally Lo50. To end it all off, we have redistributed lo1000 (redistribute connected subnets):
Now, lets check out R2’s view of the NSSA external route:
As you can see the forwarding address is now set to lo50’s address. Its not the lowest, nor the highest IP address. It IS however, the lastly added interface. Lets take it a step further and create a lo49 interface, with 49.49.49.49/8 and advertise this loopback into the OSPF domain:
On R3:
And then on R2:
Lets shut down the lo49 interface, and my stipulation is that lo50 will be the forwarding address:
After sh on lo49 on R3, R2 shows:
So go figure 🙂