MPLS ×××网络中的隐形问题

  如下图的MPLS ×××实验拓扑中，R2-R3-R4处于同一AS，IGP使用OSPF，CE设备R1和R5和骨干网运行BGP协议。所有设备的loopback0地址配置24位的掩码。

   按照MPLS ×××组网，完成所有配置，最后发现CE设备R1和R5可以相互学习到对方的路由，但是互ping不通，traceroute仅能到达互联的PE；PE设备R2和R4之间加入vrf的loopback接口，相互也不能ping通。整个网络，路由平面正常，但是数据转发平面不正常。

   死活找不到原因，确认所有配置都完整，还是无法找到问题原因。无意中发现一条log信息：*Mar  1 00:27:00.187: %BGP-4-×××V4NH_MASK: Nexthop 2.2.2.2 may not be reachable from neigbor 4

   检索“%BGP-4-×××V4NH_MASK”找到以下类似cisco官方资料：

Error Message  

% BGP-4-×××V4NH_MASK: Nexthop [IP_address] may not be reachable from neigbor   [IP_address] - not /32 mask

Explanation    A Virtual Private Network Version 4 (×××v4) route is being sent to the specified internal BGP (IBGP) neighbor, and the next-hop address is a loopback that does not have the IP host mask /32. OSPF is being used on this loopback, and the OSPF network type of this interface is LOOPBACK. OSPF advertises this address as a host route with the IP host mask /32, regardless of what mask is configured. Because this configuration conflicts with Tag Distribution Protocol (TDP), which uses configured masks, the TDP neighbors may not receive a tag for this route. This problem can break connectivity between sites that belong to the same ×××.

Recommended Action    Configure the loopback that is used as the next hop with the IP host mask /32.

   从以上资料才得知，loopback0配置了24位掩码是问题根源，以前习惯使用32位掩码的时候从来也没有遇到这种问题。解决方法有两个：

1、修改loopback0的掩码为32位

2、在loopback0接口下增加一条命令，还原其网络类型ip ospf net point-to-point