Description

BGP Auto-discovered Neighbor peering is established with inet6 link-local (unnumbered) address but does not accept prefixes.

 

Related configuration:

===============

user@Router> show configuration interfaces xe-0/0/1:0

vlan-tagging;

unit 7 {

  vlan-id 7;

  family inet;

  family inet6;

}

 

user@Router> show configuration protocols bgp

group GR_internal {

  family inet {

    unicast {

      extended-nexthop;

    }

  }

  family inet6 {

    unicast;

  }

  peer-as 12345;

  dynamic-neighbor DN_BGP_Unumb {

    peer-auto-discovery {

      family inet6 {

        ipv6-nd;

      }

      interface xe-0/0/1:0.7;

    }

  }

}

 

 

Symptoms

The BGP prefixes advertised from the remote peer was rejected due to the BGP nexthop sanity check failure: it is the inet6 link local address.

 

Feb 9 11:20:19.635636 BGP_UNUSABLE_NEXTHOP: bgp_nexthop_sanity: peer fe80::ee38:7300:770:808f%xe-0/0/1:0.7 (Internal AD IPv6-ND) next hop fe80::ee38:7300:770:808f remote link local address, ignoring routes in this update (instance master)

 

 

Solution

The user can manually configure a valid global inet6 address in the related interface to form the BGP Auto-discovered Neighborhood

 

When this KB article is written on 20/Feb/2024, BGP Auto-discovered Neighbor peering supports link local address for one-hop eBGP neighbor only.

 

 

Modification History

2024-02-20 : Article Created