IBGP Route advertising

[prajithtr_2]

haii,

   I am a CCNP student .While I was trying to make an ibgp network by interconnecting 3 routers seriely(R1------R2-----R3) with basic configuration.Then R1 and R2 & R2and R3 became neigbours its states shows as established.The problem is that no routes were installed in ip routing table and also bgp routing table.what could be the reason for it.

The basic configuration commands I used was

neigbourf AS peergroup

neigbour AS remote-as <ASN>

neoghbour <ip address of neigbour> peergroup

thanks

[Dan-Ciprian Cicioiu]

Hello ,

A route received via iBGP is not advertised to other iBGP peers.

You could configure the router in the middle as route-reflector.

Dan

[saimbt]

Hello,

Either use route reflector or have a full mesh connection something like this

R1 ----R2

|          |

|__R3_ |

Sai.

[Dan-Ciprian Cicioiu]

Hello Sai ,

The iBGP neighbors dont have to be directly connected to be adjancent.

So yes , he could build a full mesh adjancencies between all three routers , without connecting them direcly to each other.

Dan

[saimbt]

Thanks Dan,

You are perfectly right, as BGP works on TCP the neighbour need not be directly connected. I actually meant full mesh but messed it up with the diagram

Sai.

[Mahesh Gohil]

Hi,

To be more specific.

ebgp is having as-path lookup mechanism as a loop avoidance mechanism (split-horizon, means routes received from

same as number will not be installed in bgp unless there is feature configured like allow-as-in or as-override )

so if you have ebgp peering like

    -----r2

r1            r4

    ---- r3-

the routes originated by r1 to upper path (r1>r2>r4) will not be installed back to r1 if it is received by lower path (r4>r3>r1)

but what about ibgp...all the routers will have same AS number and as number is not carried in ibgp updates. so to have some mechanism

of avoiding looping the rule is routes received by ibgp peer will not be advertised to another ibgp peer.

regards

Mahesh