Do your RR really support Add-Path?
From what I understand, bgp add-path install allows for installing backup paths if it is/are already received.
e.g. two different originating PE routers with 2 different RDs
see the output
PE3
router bgp 65432
bgp log-neighbor-changes
no bgp default ipv4-unicast
neighbor RR peer-group
neighbor RR remote-as 65432
neighbor RR update-source Loopback0
neighbor 172.27.1.1 peer-group RR
neighbor 172.27.2.2 peer-group RR
!
address-family ipv4
exit-address-family
!
!!!! I have 2RRs
address-family vpnv4
neighbor RR send-community both
neighbor 172.27.1.1 activate
neighbor 172.27.2.2 activate
exit-address-family
!
!!! connection to CE3
address-family ipv4 vrf A
neighbor 10.10.33.2 remote-as 65506
neighbor 10.10.33.2 activate
exit-address-family
7.7.1.0/24 is comming from CE1 & 2
7.7.3.3/32 is comming from CE3
PE3#sh bgp vpnv4 unicast all
BGP table version is 32, local router ID is 172.27.33.33
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
r RIB-failure, S Stale, m multipath, b backup-path, f RT-Filter,
x best-external, a additional-path, c RIB-compressed,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found
Network Next Hop Metric LocPrf Weight Path
Route Distinguisher: 1:1
* i 7.7.1.0/24 172.27.11.11 0 100 0 65506 i
*>i 172.27.11.11 0 100 0 65506 i
Route Distinguisher: 2:2
* i 7.7.1.0/24 172.27.22.22 0 100 0 65506 i
*>i 172.27.22.22 0 100 0 65506 i
Route Distinguisher: 3:3 (default for vrf A)
*>i 7.7.1.0/24 172.27.11.11 0 100 0 65506 i
* i 172.27.22.22 0 100 0 65506 i
*> 7.7.3.3/32 10.10.33.2 0 0 65506 i
You can see, that you are - thanks to different RDs - receiving 2 different prefixes in vpnv4.
If you use bgp additional-path install in afi ipv4 vrf A, you will see something like this.
address-family ipv4 vrf A
bgp additional-paths install
neighbor 10.10.33.2 remote-as 65506
neighbor 10.10.33.2 activate
exit-address-family
PE3#sh bgp vpnv4 unicast all
BGP table version is 34, local router ID is 172.27.33.33
Network Next Hop Metric LocPrf Weight Path
Route Distinguisher: 1:1
* i 7.7.1.0/24 172.27.11.11 0 100 0 65506 i
*>i 172.27.11.11 0 100 0 65506 i
Route Distinguisher: 2:2
* i 7.7.1.0/24 172.27.22.22 0 100 0 65506 i
*>i 172.27.22.22 0 100 0 65506 i
Route Distinguisher: 3:3 (default for vrf A)
*>i 7.7.1.0/24 172.27.11.11 0 100 0 65506 i
*bi 172.27.22.22 0 100 0 65506 i
*> 7.7.3.3/32 10.10.33.2 0 0 65506 i
If you want to advertise more routes for the same prefix from a bgp speaker, you need add-path capability. Now I am testing it, and in IOS versions I have it is not supported.
look for CAPABILITY code: 69 in debug bgp all (and clear the session)
If you dont see it, there is no support for sending same prefix multiple times in an update differentiated with the "add-path identifier"
If a BGP speaker receives multiple paths for the same prefix within same session is simply uses the newest one advertisement. (without add path).
Ad: bgp advertise-best-external - it is exactly as Bharat Negi writes. just confirmed it in my lab.
If you use bgp advertise-best-external (not in global bgp, but in address-family ipv4 vrf xyz)
, PE2 WILL advertise the prefix learned from CE2 and PE3 WILL have 2 pfxs available from both PE1 and PE2.
But still, I use different RDs.
might be helpful
http://blog.ine.com/2010/11/22/understanding-bgp-convergence/
please correct me, if I am wrong, still learning as well
