cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2916
Views
0
Helpful
1
Replies

Can't establish IPv6 iBGP peering between two ASR9000

data center
Level 1
Level 1
 considering that similary configured IPv4 peering works.

I tried to configure peering between loopbacks, neighbor's state keeps on "idle". Tried with linklocal adresses on physical interfaces, neighbor's state keeps on "active".
It's ok with routes, routes to the loopbacks advertised via OSPF, routers can ping each other. I can't find where's the problem?
Can anyone send working example of dual-stack iBGP peering?

Here is the config of one of ASR's

 

interface Loopback254
ipv4 address 10.1.254.4 255.255.255.255
interface Loopback6
ipv6 address 2a00:dac0::1/128

router bgp 197765
 bgp log neighbor changes detail
 bgp bestpath med always
 address-family ipv4 unicast
  network 31.13.128.0/21
 !
 address-family ipv6 unicast
  network 2a00:dac0::/32
 !
#################################################
### This is IPv4 peering, and it work perfect: 

 neighbor-group iBGP
  remote-as 197765
  update-source Loopback254
  address-family ipv4 unicast
   next-hop-self
  !
 !
 neighbor 10.1.254.3
  use neighbor-group iBGP
  description DC-ASR-2
  address-family ipv4 unicast
   route-policy ASR2_IN in
   route-policy iBGP_OUT_DENY_DEFAULT out
  !
  
################################################### 
### This is IPv6 peering, and neighbor stays "idle": 

  neighbor-group iBGP_IPv6
  remote-as 197765
  update-source Loopback6
  address-family ipv6 unicast
   next-hop-self
  !
 neighbor 2a00:dac0::2
  use neighbor-group iBGP_IPv6
  description DC-ASR-2
  address-family ipv6 unicast
   route-policy ASR2_IPv6_IN in
   route-policy iBGP_IPv6_OUT_DENY_DEFAULT out
  !
1 Reply 1

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello ,

BGP is multiprotocol you don't need to setup an IPv6 iBGP session to exchange IPv6 prefixes with the iBGP neighbor : you can simply activate the AF ipv6 unicast under the existing IPv4 based session adding the appropriate commands to use IPv6 address of loopback6 as next-hop instead of update-source loop6.

To be noted also the opposite is true an AF ipv4 unicast can be enabled / activated over an IPv6 iBGP session.

One thing is the transport protocol for the session TCP/IPv4 or TCP/IPv6 another thing is of what arguments / channels we would like to talk ( the different address families).

 

As a general note the use of a single loopback interface with both an IPv4 address and an IPv6 address is to be preferred.

 

Hope to help

Giuseppe