IBGP and EBGP not routing correctly

Unanswered Question
Jun 19th, 2010

Hello, I have setup a lab environment with five routers  Below is a diagram.

R1 - -------------R2

|     as5500     |

R3 ---------------R4


                          R5  AS6500 is  BGP advertising via distrubted connected interface  ip address

All of the AS 5500 BGP Peer Group neighbor relationships have been established using thier loopbacks and are considered fully meshed.

I am using EIGRP  to only advertise the loopback networks for AS5500 (R1 =L1: , R2 =L2:, R3 =L3:, R4 =L4:

I have the proper configurations of update loop back source, Multi hop for routing to occur to R5's loop back, etc. in place.

Router5 (AS6500) is advertising via redistribute route map command connected network

This network can be seen by all AS 5500 routers and each router is marked with valid and best for the The Best and valid routes is R4 router which is directly connected to R5.

The route even appears in the routing table (show IP Route) with B (for BGP) but with R4's loopback address of as the next hop.

The problem I am having is that R1, R2, R3 are unable to ping network

R4 pings to R5, with no problem.  What is preventing R1, R2, and R3 from reaching

Thanks in advanced.


I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
syed.jaleel25 Sat, 06/19/2010 - 18:31


You must provide a static route on R4 to reach 200. network. After which you should be able to ping.



milan.kulik Tue, 06/22/2010 - 00:07

Or Ping with the loopback source address from R1 - R3 (I undrestand loopback addresses are being advertised by BGP to R5?).



ctrejo Tue, 06/22/2010 - 18:09

I want to thank you for all of your responses.  The problem is now solved by adding defaults routes in R4 and R5 to each other. Now I can ping from all iBGP router (R1,R2,R3) to R5's host.

But now I have another problem and I thought the resolution would be similar. R1 is advertising via BGP network command.  R5 can see that network with no problem. But R5 is unable to ping R1's loopback address of I added some static routes but no success.  I don't understand because if R1 can now ping R5 loopback shouldn't the pings work?

Thank you for all of your help.  As you can see I'm a newbie and I will be starting a new job on Monday and there will be BGP involved.  Thank you very much.

Thank you all once again.


PS:I meant to post this message all the way in the bottom of the last post- Sorry.

milan.kulik Tue, 06/22/2010 - 22:27


I guess Ping source address problem again.

Are you able to Ping from R5 using an extended Ping with source address?

If yes, your R1 router (or some router on the path - R2 or R3) doesn't know a route to R5's interface address

Don't forget the rule: When the source address is not specified, a router is using the IP address of his outgoing interface as the source address in the ICMP Echo Request packet sent while pinging. And the target device has to reply to that address when sending an Echo Reply - i.e., the address must be routable along the whole path through the network.



Giuseppe Larosa Tue, 06/22/2010 - 05:18

Hello Chistian,

as noted by Milan and others when you perform a  ping test we need to take care of source address.

A router uses as source address that of the outgoing interface to ping destination, that is that of an inter-router link on the path to R4.

Inside AS 5500 the EIGRP allows all routers to be able to answer to an ICMP packet with a source of any interface of inter-router links.

For R5 this is not true as it is not taking part in EIGRP so the source used really counts and influences the ping result.

Otherwise you would to redistribute connected on R1,R2, R3, R4 routers in order to make R5 able to answer or simply a default route pointing to R4's ip address on shared link

the fact that R5's ip address is seen with BGP next-hop= = R4's IP address can be normal if you have used next-hop-self on R4 on iBGP sessions to R1,R2, R3.

To be noted you don't need ebgp-multihop on internal BGP sessions where TTL is set to 255  instead of 1 as it happens for eBGP sessions.

Hope to help



This Discussion