The log entries show that R2's BGP session to neighbor 10.0.0.1 (R1) was torn down because the number of received VPNv4 prefixes (1560) exceeded the configured maximum-prefix limit (1500). This is a standard BGP maximum-prefix protection mechanism, and once triggered it drops the peering (and any associated VPNv4/MDT topologies) until the condition is manually cleared and, if the true prefix count is legitimately higher, the limit is raised. Because R2 functions as the ABR/route-reflector point between OSPF 100 and the rest of the Unified MPLS core, losing that VPNv4 session prevents propagation of L3VPN routes between OSPF 100 and OSPF 200/300, while OSPF 100's own internal Unified MPLS LSPs/services remain unaffected since they do not depend on that BGP session. The fix is to raise the maximum-prefix limit on R2 for that neighbor to at least 1560 (or higher) so the session can stay up and carry the additional prefixes introduced by the new data center.
Community Discussion