Multicast RPF problem with multiple IGP path

From: Jung-I Lin (easyman.lin@gmail.com)
Date: Sun Oct 29 2006 - 05:28:14 ART


Dear GS folks,

I have a question about multicast rpf problem.
The topology as following

       /------------------------R4---------------------\
      / \
R6 R5
      \ /
       \------------------------R1---------------------/
The IGP is eigrp on all routers.
R5 has two equal cost path to R6's loop0 which is bsr+rp-candidate.
R1,R4 can accept the bsr and choose R6's loop0 to be the RP.
But R5 will not accept this BSR messages, and complaints about not from RPF
interface.
on R5
#sh ip route
D EX 150.1.6.0 [170/2300416] via 155.1.45.4, 00:19:27, Serial1/1
                  [170/2300416] via 155.1.0.1, 00:19:27, Serial1/0

Rack1R5(config)#do sh ip rpf 150.1.6.6
RPF information for ? (150.1.6.6)
  RPF interface: Serial1/1
  RPF neighbor: ? (155.1.45.4)
  RPF route/mask: 150.1.6.0/24
  RPF type: unicast (eigrp 10)
  RPF recursion count: 0
  Doing distance-preferred lookups across tables

*Mar 1 02:18:13.539: PIM-BSR(0): 150.1.6.6 bootstrap forwarded on Serial1/0
*Mar 1 02:18:13.567: PIM-BSR(0): bootstrap on non-RPF path Serial1/0

All I can do to have R5 accept this BSR messages is to turn on "multicast
multipath"
Then BSR messages is accepted by R5 and R6's loop0 is choose to be the RP
for R5.
Rack1R5(config)#ip multicast multipath
Rack1R5(config)#
Rack1R5(config)#do sh ip rpf 150.1.6.6
RPF information for ? (150.1.6.6)
  RPF interface: Serial1/0
  RPF neighbor: ? (155.1.0.1)
  RPF route/mask: 150.1.6.0/24
  RPF type: unicast (eigrp 10)
  RPF recursion count: 0
  Doing distance-preferred lookups across tables
  Multicast Multipath enabled
Rack1R5(config)#
Rack1R5(config)#
*Mar 1 02:20:13.563: PIM-BSR(0): 150.1.6.6 bootstrap forwarded on Serial1/1
*Mar 1 02:20:13.567: PIM-BSR(0): bootstrap on non-RPF path Serial1/1
Rack1R5(config)#do sh ip pim rp map
PIM Group-to-RP Mappings

Group(s) 224.0.0.0/4
  RP 150.1.6.6 (?), v2
    Info source: 150.1.6.6 (?), via bootstrap, priority 0, holdtime 150
         Uptime: 00:02:11, expires: 00:02:14

My question is wouldn't the RPF check take all IGP paths into account?
Is there any other ways except "multicast multipath" could be use to solve
this RPF failure?

-- 
Thanks
Best Regards,

Jung-I Lin



This archive was generated by hypermail 2.1.4 : Wed Nov 01 2006 - 07:29:07 ART