RT1-----RT2-------RT3

RT1 is RR client of RT2, we run label BGP among RT1, RT2 and RT3; MPLS LDP is OK;
RT1 network route 192.168.1.18/32 with MPLS-label 4123, we hope RT2 propagate it to RT3 by its address 192.168.8.2 and swap MPLS-label, however, RT2 just propagate it without any changing(192.168.1.18/32 nexthop 192.168.8.1).

router bgp 65000
bgp router-id 192.168.8.2
address-family ipv4 unicast
!
neighbor 192.168.8.3
remote-as 65000
update-source Loopback0
address-family ipv4 labeled-unicast
next-hop-self
!

!
neighbor 192.168.8.1
remote-as 65000
update-source Loopback0
address-family ipv4 labeled-unicast
route-reflector-client
!
!

_________________
http://blog.alwaysthenetwork.com

We use IoS XR 3.9, route-map has been replaced by Route-policy?
By nei route-policy IN, we can change next-hop-self, but BGP can not distribute a new bgp label for the route.
By nei route-policy out, We can not change next-hop-self.....

How can Cisco SRS support Seamless MPLS solution without chaning next-hop-self and swap MPLS label?

How can Cisco CRS-1 support Seamless MPLS solution by next-hop-self??

I hope ABR to reflect label IBGP route , modify next-hop-self and distribute new label for the route.
Who can help me? Thanks first....

Any CRS configuration for changing nexthop and mpls label in IBGP?

Can't you change the values via a route-map outbound to the reflector clients?

_________________
www.mellowd.co.uk/ccie/

ABR recieves label route from RR client, I hope to change its nexthop to ABR, then ABR should change its label.
I tried to use route-policy(route-map in IOS) to realize it, however i failed;
If i use route-policy to Import direction, ABR can change next-hop-self, but it didn't change label;
If i use route-policy to export direction, ABR can not change next-hop-self, however, in debug bgp policy run the policy had been executed sucessfully.

Any answer available?

What does your route-map say about the next-hop? Are you using next-hop address or next-hop peer-address?

I must admit I don't know of a solution as I've not done what you are trying to do

_________________
www.mellowd.co.uk/ccie/

Don't be an ass, if you need an answer instantly then a forum isn't the place to get it.

Don't post a follow-up <3hrs after your last post.

You might get some more help as well if you actually describe why you are wanting to do what you are trying to do, it could be that there is a better method.

_________________
---
David
CCIE R&S #38338, CCIP, CCNP

http://networkbroadcast.co.uk - My Blog
http://twitter.com/davidrothera

I has many Metros interconnected by IP core;
Taditionally, we ran ISIS level-1 in Metro, and ISIS level-2 in IP core;
We used MPLS in Metros and IP core seperatelly, every area in an isolated MPLS domain;

Now, we want to integrated all MPLS domains into a whole MPLS domain. We have to use BGP arrounce route from one Metro to Core and another Metro, and use Label BGP distribute MPLs label over LDP label.

Because all our boxes in a same AS. we want to make ABRs as BGP route reflector, PE in Metro as their clients. when ABR reflect route from its clients to another ABR, it should modify its nexthop self which is reachable from another metro, and distribute a new BGP label(transit BGP label node).

that is requiremetn, can you help me ? Thanks..

Ok, to summarise then you have two separate networks essentially and you are wanting to ensure that the label that is sent to the CE's is that of the ABR between the two MPLS networks and not the address of a far end PE in the other network?

_________________
---
David
CCIE R&S #38338, CCIP, CCNP

http://networkbroadcast.co.uk - My Blog
http://twitter.com/davidrothera

yes.
I got the answer. Cisco add a new command "ibgp policy out enforce-modifications" in order to meet the Seamless MPLS solution.
After i added "ibgp policy out enforce-modifications", next-hop-self took effective for RR.

Thank you very much.