Routing
Reply
Contributor
krenjt
Posts: 78
Registered: ‎11-12-2010
0

exchanging ipv6 routing information over bgp

Hi guys,

 

I have a question about two ebgp router exchanging ipv6 routing information.

 

There's a ebgp session between R1 and R2 over ipv4 address. The address of R1 is 172.27.0.1and R2 is 172.27.0.2
R1 in AS 100 and R2 in AS 200, they both support family inet unicast and family inet6 unicast.

 

 

Here is traceoption from R1:

Apr 12 20:52:20.353054 BGP RECV 172.27.0.2+179 -> 172.27.0.1+62980
Apr 12 20:52:20.353080 BGP RECV message type 2 (Update) length 62
Apr 12 20:52:20.353096 BGP RECV Update PDU length 62
Apr 12 20:52:20.353114 BGP RECV flags 0x40 code Origin(1): IGP
Apr 12 20:52:20.353135 BGP RECV flags 0x40 code ASPath(2) length 6: 200
Apr 12 20:52:20.353171 BGP RECV flags 0x90 code MP_reach(14): AFI/SAFI 2/1
Apr 12 20:52:20.353199 BGP RECV nhop ::ffff:172.27.0.2 len 16
Apr 12 20:52:20.355271 BGP RECV ::/0
Apr 12 20:52:20.355400 bgp_nexthop_sanity: peer 172.27.0.2 (External AS 200) next hop ::ffff:172.27.0.2 unexpectedly remote, ignoring routes in this update

 

R2 advertised a ipv6 route ::/0 to R1 and the next-hop of this route is a ipv4-mapped ipv6 address.
but i didn't configured a ipv6 address on R1 and R2.

 

How to solve this problem: configured a ipv4-mapped ipv6 address on R1, is that right?

 

 

ThankS!

Visitor
shijot
Posts: 1
Registered: ‎02-26-2012
0

Re: exchanging ipv6 routing information over bgp

Juniper Employee
jstar
Posts: 22
Registered: ‎04-28-2008
0

Re: exchanging ipv6 routing information over bgp

Hi,

 

Configuring IPV6 address at interface should solve this.

As mentioned above if its a 6PE setup (from  AFI/SAFI in above traceoption , it dont look though) then you need "ipv6-tunneling" knob.

Contributor
krenjt
Posts: 78
Registered: ‎11-12-2010
0

Re: exchanging ipv6 routing information over bgp

Hi, thanks for you reply.

 

Which type of ipv4 address should i configured on the interface: ipv4-mapped ipv6 address or ipv4-compatible ipv6 address. I configured a ipv4-mapped ipv6 address on both r1 and r2, r1 received the route and accept it, but if configuring a ipv4-compatible ipv6 address, it didn't work well.  

 

 

Juniper Employee
jstar
Posts: 22
Registered: ‎04-28-2008
0

Re: exchanging ipv6 routing information over bgp

strange...Not sure why :smileysad:

May be experts can comment

 

Recognized Expert
mosesnehru
Posts: 214
Registered: ‎09-15-2011
0

Re: exchanging ipv6 routing information over bgp

I have the same problem. When I tested in the lab on MX with 10.4R7, it gives the same result.

( works for "V4 Mapped V6 addres" and doesn't work for "V4 Compatible V6 address".)

 

I checked the RFC 4291 (IPv6 Addressing Architecture)

http://tools.ietf.org/html/rfc4291

 

This mentions ( Section 2.5.5.1 )

The "IPv4-Compatible IPv6 address" is now deprecated because the
   current IPv6 transition mechanisms no longer use these addresses.
   New or updated implementations are not required to support this
   address type.

 

 

Perhaps, the new Junos implementation doesn't support "v4 compatible v6" for this purpose.

 

So, I think it is better to use "V4 Mapped V6 address"  ( Eg ::ffff:10.10.10.1/126 )

Contributor
placek
Posts: 34
Registered: ‎04-19-2010
0

Re: exchanging ipv6 routing information over bgp

Hi,

This is the known problem. If You are exchanging IPv6 routes using IPv4 BGP session next-hop-for IPv6 prefixex is IPv6 (IPv4 mapped to precise).  Normally it makes this route unusable because next-hop is not reachable.  There is not clear solution for it. You can of course try to fix it by setting those addreses and static routing but this is not common and dislike practise. The only way to do it nice is to peer using IPv6 addreses. So You have two ebgp sessions. One for IPv4 (peering using IPv4) and another for IPv6 family (peering using IPv6).

 

Cheers,

 

Artur

Recognized Expert
mosesnehru
Posts: 214
Registered: ‎09-15-2011
0

Re: exchanging ipv6 routing information over bgp


placek wrote:

Hi,

This is the known problem. If You are exchanging IPv6 routes using IPv4 BGP session next-hop-for IPv6 prefixex is IPv6 (IPv4 mapped to precise).  Normally it makes this route unusable because next-hop is not reachable.  There is not clear solution for it. You can of course try to fix it by setting those addreses and static routing but this is not common and dislike practise. The only way to do it nice is to peer using IPv6 addreses. So You have two ebgp sessions. One for IPv4 (peering using IPv4) and another for IPv6 family (peering using IPv6).

 

Cheers,

 

Artur


Hi Thanks for the reply.

 

Ofcourse , Seperate BGP sessions for family inet and inet6 will work, My intention was to test the "IPv6 BGP Routes over IPv4 Transport" solution described in the Juniper documentation. ( see below)

 

http://www.juniper.net/techpubs/en_US/junos11.1/topics/usage-guidelines/routing-configuring-ipv6-bgp...

 

 

As I mentioned before, this solution works perfectly with "v4 mapped v6" address configured on the peering interface. ( Not with the v4 comaptible v6 address as described in the documentation)

 

Juniper Employee
mohsinul
Posts: 15
Registered: ‎03-23-2012
0

Re: exchanging ipv6 routing information over bgp

Starting from JUNOS 10.4 you need to Enable Processing of IPv4-mapped IPv6 Addresses with

“set system allow-v4mapped-packets”

- Mohsinul Malik
JNCIE-SP | CCIE-SP | JNCSP-SP | JTNOC | JNCIP (SP, ENT, SEC) | JNCIS-E | CCNA | CCNP | MCSE | EXCH SVR - 2003 | MCTS 2008 | RHCE (EL5) | ITIL V3
Trusted Contributor
NetDonkey
Posts: 110
Registered: ‎10-05-2010
0

Re: exchanging ipv6 routing information over bgp

Also have a look at the Day One advanced ipv6 guide which has an example of what you are looking for.

 

Nick

--
Network Architect
Juniper Ambassador
JNCIE-ENT #232
Twitter: @NetDonkey
http://taildrop.net
Copyright© 1999-2013 Juniper Networks, Inc. All rights reserved.