cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
452
Views
0
Helpful
3
Replies

BGP Sync Problem - Is it?

kfarrington
Level 3
Level 3

Hi Guys,

Can I ask a question in regards to synchronisation. (pls see diagram)

Actually two questions here. How Sync works, and order of operation for BGP and Routing Table (RT)

If I have the following topology, the statement I would like to make is this.

If you are running BGP on a pair of routers (eBGP and iBGP type of config) and it is receiving a prefix via eBGP and iBGP and synchronisation is ON. The router will chose the eBGP learned route for injection into the routing table all the time, the iBGP route will never be considered - even if you are trying to manipulate the decision, via use of local-pref/MED etc etc? This is because of a next-hop mismatch?

Is this actually correct? And if so, is there a configuration tech-note that explains in more detail this behaviour. (I have ask a similar question before and a very kind member of the forum put this idea to me)

Also, if someone could confirm that the order of operation (especially in the sync off scenario) is correct, ie router will actually remove prefix from RT once a new iBGP path has come into R1 and is better than its current eBGP prefix.

Ie,

Sync on

--------------

R1 received 2.0.0.0/8 via eBGP and sets LP to 95

R1 puts prefix into RT (as R2 as yet has not received the prefix via eBGP)

R2 received 2.0.0.0/8 via eBGP and sets LP to 100

R2 puts prefix into RT

R2 now sends iBGP update to R1 with BGP prefix. Now the as the eBGP next hop (which is in the RT of R1) is different to the iBGP next hop received from R2, the route never gets compared because synchronisation has failed. The iBGP prefix with the highest LP never gets used.

Sync off

--------------

R1 received 2.0.0.0/8 via eBGP and sets LP to 95

R1 puts prefix into RT (as R2 as yet has not received the prefix via eBGP)

R2 received 2.0.0.0/8 via eBGP and sets LP to 100

R2 puts prefix into RT

Now R2 has put eBGP route into RT, and sends iBGP update to R1. R1 now needs to remove the route from the RT as a BGP decision process needs to be made and the iBGP route is now the best so the route gets ribbed out and RT now has IGP route to prefix.

Is this correct guys?

Many thx indeed,

Ken

3 Replies 3

Harold Ritter
Cisco Employee
Cisco Employee

Hello Ken,

Q1:

If you are running BGP on a pair of routers (eBGP and iBGP type of config) and it is receiving a prefix via eBGP and iBGP and synchronisation is ON. The router will chose the eBGP learned route for injection into the routing table all the time, the iBGP route will never be considered - even if you are trying to manipulate the decision, via use of local-pref/MED etc etc? This is because of a next-hop mismatch?

A1: If you are not redistributing 2/8 in your IGP then this prefix will be marked as "not synchronized" in BGP and will never be eligible for selection, no matter what its local preference is. This is assuming syncronization is enabled.

If you are redistributing 2/8 in your IGP then this prefix should be marked as "synchronized" in BGP and will be used by the BGP best path selection process. Care must be taken if you have route reflectors in your AS and that OSPF is used as the IGP. If the prefix is received via OSPF, IOS checks that the BGP RID the prefix was received from is the same as the OSPF RID, in which case it won't if the BGP prefix was received via a RR.

Q2: R2 now sends iBGP update to R1 with BGP prefix. Now the as the eBGP next hop (which is in the RT of R1) is different to the iBGP next hop received from R2, the route never gets compared because synchronisation has failed. The iBGP prefix with the highest LP never gets used.

A2: The reason the iBGP prefix is not used is not that the eBGP and iBGP NH don't match but rather that there is no IGP learnt route for the iBGP received prefix (2/8 in this scenario). The iBGP received prefix is therefore marked as "not synchronized" and is never used by the BGP best path selection process.

Hope this helps.

Harold Ritter
Sr Technical Leader
CCIE 4168 (R&S, SP)
harold@cisco.com
México móvil: +52 1 55 8312 4915
Cisco México
Paseo de la Reforma 222
Piso 19
Cuauhtémoc, Juárez
Ciudad de México, 06600
México

Hi Harold, Very nice to speak with you again and thanks for the response.

Both routers are unconditionally redistributing BGP into EIGRP. Here is the output below. This is why I cannot understand the "sync failed".

Version shown as well, so i will search the bug toolkit. Any thoughts now that you have the IGP info

Pls see attached :-

Hello Ken, It is good to hear from you as well.

The reason for the iBGP path not to be synchronized is that the route installed in the RIB is not the IGP route but the eBGP route. Synchronization will only work if the IGP route is installed in the RIB.

Hope this helps,

Harold Ritter
Sr Technical Leader
CCIE 4168 (R&S, SP)
harold@cisco.com
México móvil: +52 1 55 8312 4915
Cisco México
Paseo de la Reforma 222
Piso 19
Cuauhtémoc, Juárez
Ciudad de México, 06600
México
Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Review Cisco Networking products for a $25 gift card