doyle ospf chap 4 case 2...

to rid or not to rid...

it is evident that the router id, if changed on an active ospf process, will not alter the current rid in the process... the process needs to be shut down for this to be changed... there is no argument against the validity of either using the highest (if configured) loopback or a manual rid to protect the ospf process... also, once established as the rid, a loopback in shutdown will still remain the rid so long as the process isn't restarted...

we know it is important that the ospf process not be saddled with a rid dependant on the state of an interface that may be subject to shutdown (interface failure)

we also understand that a manual rid is simply an ospf process identifier and has nothing to do with the state of routing for the process...

for the example, loopbacks are added, and only the network statement for router r is changed as more specific...



r

interface Loopback0

ip address 192.168.50.1 255.255.255.255

outer ospf 10

network 192.168.30.0 0.0.0.255 area 1

c

interface Loopback0

ip address 192.168.50.2 255.255.255.255

g

interface Loopback0

ip address 192.168.50.3 255.255.255.255

m

ip address 192.168.50.4 255.255.255.255

note how doyle chooses a uniform scheme...

the simple presence of the loopback does not automatically change the rid...

r#sh ip ospf

 Routing Process "ospf 10" with ID 192.168.30.9

nor is it changed at the neighbor...

c#sh ip ospf neigh

Neighbor ID     Pri   State           Dead Time   Address         Interface

192.168.20.2      0   FULL/  -        00:00:33    192.168.20.2    Serial1/2

192.168.30.9      0   FULL/  -        00:00:36    192.168.30.9    Serial1/1

the process for router r is closed, and in a few moments, the id expires on the neighbor...

c#sh ip ospf neigh

Neighbor ID     Pri   State           Dead Time   Address         Interface

192.168.20.2      0   FULL/  -        00:00:34    192.168.20.2    Serial1/2

lacking a manual rid, once the process is brought back the highest loopback is reflected in the topology...

r#sh ip ospf

 Routing Process "ospf 10" with ID 192.168.50.1

c#sh ip ospf neigh

Neighbor ID     Pri   State           Dead Time   Address         Interface

192.168.20.2      0   FULL/  -        00:00:37    192.168.20.2    Serial1/2

192.168.50.1      0   FULL/  -        00:00:33    192.168.30.9    Serial1/1

the importance of this cannot be understated... a common practice is to use public ip addresses for loopbacks such as 1.1.1.1 for router1, 2.2.2.2 for router2, etc.  the threat here is internet leakage because the ip's are public, however...

manual id's are perhaps the best choice; i prefer loopbacks because they perform double duty... when the ospf process comes back, at least one interface needs to be active... the manual id is of course, not an interface...