IPv6 NLRI over an IPv4-Only BGP Session
AS 65000's Nokia SR OS PE peers with a customer CE over an IPv4-only link 192.0.2.0/30. The BGP session is Established and IPv4 prefixes are exchanged. The CE advertises IPv6 prefix 2001:db8:100::/48 and its BGP configuration includes family ipv4 ipv6. The PE neighbor configuration contains only family ipv4; no IPv6 route appears in show router bgp routes ipv6. The PE has an IPv6 address 2001:db8:ffff::1/64 on a different interface, but no global IPv6 address on the CE link. The CE's IPv6 next-hop for the BGP announcement is 2001:db8:100::1, reachable only through the CE. The operator wants to keep the single IPv4-based BGP session and add IPv6 without redesigning the link. A previous attempt enabled family ipv6 only on the PE and reset the session; IPv4 traffic blipped and IPv6 still did not work.
1. Why is the session Established but no IPv6 routes are exchanged?
Read the worked answer
BGP uses multiprotocol capability negotiation. The session is Established because both sides agreed on IPv4 unicast; the PE did not advertise IPv6 unicast capability, so the CE is not permitted to send IPv6 NLRI. family ipv4 only on the PE means no IPv6 AFI/SAFI appears in the OPEN. Enabling IPv6 on only the PE may reset the session, but after renegotiation the CE still sees no mutual IPv6 capability, so IPv6 still does not work. Both neighbors must include IPv6 unicast in their family configuration.
2. After IPv6 is enabled on both sides, what next-hop problem remains and how should it be fixed?
Read the worked answer
The PE must resolve the IPv6 next-hop 2001:db8:100::1. The IPv4 session can carry IPv6 NLRIs, but BGP still requires a reachable IPv6 next-hop. Because the CE link has no global IPv6 address, the PE has no connected or routed path to that next-hop. Add global IPv6 addresses on the CE-PE link, for example 2001:db8:1::1/64 and 2001:db8:1::2/64, and have the CE advertise its link address as the next-hop, or configure an explicit route to 2001:db8:100::1 via the CE. Then check show router route-table ipv6 for next-hop reachability and show router bgp routes ipv6 for the prefix.
3. What verification steps confirm the fix, and what should be checked before changing families in production?
Read the worked answer
Verify with show router bgp summary for peer state, show router bgp neighbor 192.0.2.2 for negotiated families, show router bgp routes ipv6 for the prefix, and show router route-table ipv6 for next-hop reachability. Confirm the prefix is both present and selected. Because family changes can reset the session, schedule a maintenance window and confirm IPv4 is still negotiated afterward. Do not assume an Established session means every address family is active.