BGP synchronization - want to understand it, not disable it 1

[norteldude78]

Hello,

First of all, I can resolve issue by turning OFF synchronization
However, I am trying to understand how it works...

Topology:

R2------------R4
|
|
R3------FRSW
|
|
R1------------R5

AS 65400: R4 (bgp ID: 14.14.14.14)
AS 65000: R1, R2 and R3
AS 65500: R5

IBGP between R1 and R2
IBGP between R1 and R3
EBGP between R1 and R5
EBGP between R2 and R4
(no IBGP between R2 and R3, but there is IP connectivity)

OSPF between R2 and R3 and R1 over the Frame network
network type is point-to-multipoint non-broadcast.
OSPF neighbors are all FULL state
Synchronization is ON in AS 65000.

The issue:

R4 has interface loopback0: 4.4.4.0/24.
This is advertised via EBGP to R2.
I have redistrubted BGP into OSPF on R2.
R1 and R3 have the route in their route table via OSPF
Yet R1 still does not advertise this network to R5
R1 marks it as "not synchronized"
From my understanding once R1 and R3 learn this network via OSPF, it should now be "synchronized" and will be advertised to R5 via EBGP.

R2:

R2#show ip bgp 4.4.4.0
BGP routing table entry for 4.4.4.0/24, version 1258
Paths: (1 available, best #1, table Default-IP-Routing-Table)
Advertised to non peer-group peers:
172.12.123.1
65400
172.12.24.4 from 172.12.24.4 (14.14.14.14)
Origin IGP, metric 0, localpref 100, valid, external, best

R1:

R1#show ip route 4.4.4.0
Routing entry for 4.4.4.0/24
Known via "ospf 1", distance 110, metric 20
Tag 65400, type extern 2, forward metric 64
Last update from 172.12.123.2 on Serial1/0, 00:00:21 ago
Routing Descriptor Blocks:
* 172.12.123.2, from 2.2.2.2, 00:00:21 ago, via Serial1/0
Route metric is 20, traffic share count is 1

R1#show ip route

4.0.0.0/24 is subnetted, 1 subnets
O E2 4.4.4.0 [110/20] via 172.12.123.2, 00:00:44, Serial1/0

R1#show ip bgp 4.4.4.0
BGP routing table entry for 4.4.4.0/24, version 2417
Paths: (1 available, no best path)
Flag: 0x208
Not advertised to any peer
65400, (Received from a RR-client)
172.12.123.2 (metric 64) from 172.12.123.2 (2.2.2.2)
Origin IGP, metric 0, localpref 100, valid, internal, not synchronized

Anybody have a clue what I have done wrong?
Am I misunderstanding this synchronization thing?

Thanks,

-B

 

[burtsbees]

What routing protocols are on R2? Can you post a sh run?
Also, I know this is likely a lab setup, but you should always use a /32 mask for loopback...

Burt

 

[burtsbees]

I meant R1---what routing protocols are on R1 and post a sh run from R1.

Burt

 

[Jsteve]

It's hard to tell what the problem is with out seeing more of the confi and some show commands, however there is one thing you must be sure of with sync turn on with OSPF/BGP, the Router IDs for both the OSPF and the BGP process must be the same on each router.

 

[dtrtel]

The loopback interface IP address should be assigned on each connected router in the same subnet, and not necessarily /32, the subnet must be assigned the correct mask based on the intended size of the "router management subnet" in loose terms, if you will. This way all dependent routers in this address range can be configured to permit whatever applications and functions are required beteeen routers including security, the iBGP mesh, and especially for consistent Router IDs as Jsteve points out.

Also include "update-source loopback X" in neighbor config statements.

--jeff

http://linkedin.com/in/JeffLynch

 

[plshlpme]

does R1 know about the subnet that connects R2 to R4?
i thought that if you were to use next hop self on r2s bgp config that you would correct this issue... and it would show syncronized?

 

[norteldude78]

Thanks for the replies guys. I didnt want to post all configs because it would be quite large but I will do this on the next post if things don't pan out tonight...

This is a lab, and loopback subnet can really be anything, I have some CCIE practice labs with loopbacks all over the place...

burt, R1 is runing OSPF and BGP, It has learned the R4 loopback via OSPF. I will do a show run in next post.

jsteve, I will check on the router IDs and make sure they match!!

drtel, bgp peers are formed with the interface IP's so I dont need the update-source in this scenario. peers are UP.

plshlpme, I will have to check on this, but remember when I turn on "synchronization" the problem is fixed...so the next hop must be correct. Also, i think what you are describing is "next hop unreachable"...but I will check on this.

Thanks for the input guys, when i get home I will test some more...

-B

 

[norteldude78]

Well...take a look at this:

On R1:

R1#show ip bgp 4.4.4.0
BGP routing table entry for 4.4.4.0/24, version 37
Paths: (1 available, best #1, table Default-IP-Routing-Table)
Advertised to non peer-group peers:
172.12.15.5 172.12.123.3
65400, (Received from a RR-client)
172.12.123.2 (metric 64) from 172.12.123.2 (2.2.2.2)
Origin IGP, metric 0, localpref 100, valid, internal, synchronized, best

On R5:

R5#show ip route 4.4.4.0
Routing entry for 4.4.4.0/24
Known via "bgp 65500", distance 20, metric 0
Tag 65000, type external
Last update from 172.12.15.1 00:04:02 ago
Routing Descriptor Blocks:
* 172.12.15.1, from 172.12.15.1, 00:04:02 ago
Route metric is 0, traffic share count is 1
AS Hops 2


I changed all the bgp router ids to match ospf ids. It didn't take right away and so I thought that didn't work. As I was getting ready to post my configs and show commands, I noticed it said "synchronized".

so thanks to everyone who replied and jsteve gets a star

-B

 

[jneiberger]

BGP synchronization in the lab is evil. You definitely do not want it turned on in the real world, but you do need to understand how it works. As you can see, it can really complicate things.

 

[Jsteve]

Glad it work for you. I went through the same thing last year when I got my CCIE, but did alot of debugging to figure it out.