Reasons for BGP to withdraw routes?

[limtohsoon]

Hi Sir,

I'm practising a BGP lab and have encountered a problem beyond my comprehension. Hope you guys can help me.

I have five AS. Each AS has only one router and advertises one prefix. The routers are physically connected in a ring with eBGP sessions between each other.

On one of the routers, I did a "clear ip bgp *" and followed by continuous "show ip bgp" to observe its BGP table. During BGP was converging, I observed the following table:

RTD#sh ip bgp

BGP table version is 8, local router ID is 192.208.10.174

Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,

r RIB-failure, S Stale

Origin codes: i - IGP, e - EGP, ? - incomplete

Network Next Hop Metric LocPrf Weight Path

*> 128.213.0.0 192.208.10.6 0 100 200 i

* 192.208.10.1 0 500 400 200 i

*> 192.208.10.0 0.0.0.0 0 32768 i

* 195.211.10.0 192.208.10.6 0 100 200 400 500 i

*> 192.208.10.1 0 0 500 i

* 200.200.10.0 192.208.10.6 0 100 200 400 i

*> 192.208.10.1 0 500 400 i

*> 203.250.13.0 192.208.10.6 0 100 i

* 192.208.10.1 0 500 400 200 100 i

The router learned two paths (from its two eBGP peers) for each prefix. The best path was selected based on shortest AS-path in this case. However, after BGP fully converged, I observed a different table:

RTD#sh ip bgp

BGP table version is 8, local router ID is 192.208.10.174

Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,

r RIB-failure, S Stale

Origin codes: i - IGP, e - EGP, ? - incomplete

Network Next Hop Metric LocPrf Weight Path

*> 128.213.0.0 192.208.10.6 0 100 200 i

* 192.208.10.1 0 500 400 200 i

*> 192.208.10.0 0.0.0.0 0 32768 i

*> 195.211.10.0 192.208.10.1 0 0 500 i

* 200.200.10.0 192.208.10.6 0 100 200 400 i

*> 192.208.10.1 0 500 400 i

*> 203.250.13.0 192.208.10.6 0 100 i

Some paths have been withdrawn. From debug output, it indeed shows that it received UPDATE message from its peer that a route was withdrawn. I'd like to know under what condition does BGP withdraw routes.

Thank you.

B.Rgds,

Lim TS

[limtohsoon]

I notice the BGP tables are a bit difficult to read in the post. Attached is text file of the BGP table.

[thisisshanky]

Do you have any filters of any sort (inbound) on this router.

[limtohsoon]

Hi,

There are no filters (inbound or outbound) of any sort on all the routers. It's just basic BGP configurations; I haven't started configuring any routing policies.

The thing is, certain paths appeared momentarily in the BGP table and disappeared after BGP converged. It happened to all the routers.

Thank you.

B.Rgds,

Lim TS

[Harold Ritter]

This is normal behavior. BGP peers only advertise the best path to all of its peer except the one the best path is received from.

If you look at the output you provided, the following two paths were withdrawn:

195.211.10.0 192.208.10.6 0 100 200 400 500 i

203.250.13.0 192.208.10.1 0 500 400 200 100 i

At some point, the router in AS100 realized that its best path for 195.211.10.0/24 was via AS 300. Since it had already advertised a path for this network to the router in AS300, it sends a withdraw message to this same router.

In the same way, the router in AS 500 realized that its best path to 203.250.13.0/24 was via AS300 and also sent a withdraw message to its peer in AS300.

Let me know if I answered your question,

[AlexanderLV]

good explanation. 

[paul driver]

Hello
Based the bgp update the rtr receives, Routes are calculated to be installed/updated or even withdrawn, Then its down to its selection process that decides what routes will be/not entered in the bgp rib and then eventually the route table or what will be/not selected to be advertised to it bgp peers.

What you are seeing there is the decision/ selection process as it converges