Full Scale Lab 4 6.1 and 6.2

I am running into an issue that I don't think that the SG takes into consideration. 


  • R2 uses the TE tunnel to get to XR1's loopback 10.0.0.19/32  (it uses autoroute destination which is the same thing as doing a static /32 route to the destination out the tunnel). 
  • R2 uses the IPv4 unicast BGP learned routes to get to AS2000 PEs loopbacks (R1, R3, R4). 
  • The next hop to get to some of those addresses is XR1's loopback
  • All of the PE's have to form PIM adj over the MTI to all other PEs in order to create a full mesh.
The problem that I am seeing is that R2 needs to have proper RPF entries to get to all other PEs. RPF entries get populated no problem for all PEs within AS1000 since R2 has IGP routes to them. XR1 gets a static mroute, so that is taken care of. But what I am seeing is that the other PE's who's routes recurse to XR1s loopback FAIL to get RPF entries. 


Example:



R2#show ip route 10.0.0.3


Routing entry for 10.0.0.3/32

  Known via "bgp 1000", distance 200, metric 0

  Tag 2000, type internal

  Last update from 10.0.0.19 00:00:12 ago

  Routing Descriptor Blocks:

  * 10.0.0.19, from 10.0.0.19, 00:00:12 ago

      Route metric is 0, traffic share count is 1

      AS Hops 1

      Route tag 2000

      MPLS label: 19002



R2's best route to get to R3's loopback is by using XR1 as the next hop. However, XR1 uses a TE tunnel. 



R2#show ip cef 10.0.0.19

10.0.0.19/32

  attached to Tunnel0



So since to get to XR1 the tunnel is used, the entry to get to R3 also uses the tunnel:



R2#show ip cef 10.0.0.3 

10.0.0.3/32

  nexthop 10.0.0.19 Tunnel0 label 19002





R2#show ip rpf 10.0.0.19

RPF information for ? (10.0.0.19)

  RPF interface: Ethernet0/0.220

  RPF neighbor: ? (10.0.220.20)

  RPF route/mask: 10.0.0.19/32

  RPF type: multicast (static)

  Doing distance-preferred lookups across tables

  RPF topology: ipv4 multicast base





R2#sh ip rpf 10.0.0.3

 failed, no route exists




This happens for all other Inter-AS destinations that use XR1 as the next hop. If I want to establish an adj over the MTI, I am having to add static mroutes for each PE's loopback that recurses to XR1. 


This may be just a version thing though. I am using 15.2(4). Can anyone confirm if they are having the same issue on 12.2SRE? 


Thanks,

Comments

Sign In or Register to comment.