From: Koen Zeilstra (koen@koenzeilstra.com)
Date: Thu Dec 21 2006 - 18:25:34 ART
I agree with you that it just works. However there is no failover for
multicast traffic as there is for unicast in this scenario.
RPF check is done on to select the incoming interface not the outgoing. R4
will have R1 or R2's source address in the PIM updates not the source of
the multicast stream.
If all the salmon caught in Canada in one year were laid end to end
across the Sahara Desert, the smell would be absolutely awful.
On Tue, 19 Dec 2006, Jian Gu wrote:
| I totally don't understand the problem you are trying to solve, 1)multicast
| from hostB will have hostB's source IP address, routers won't change IP
| multicast packet's source IP address, 2) RPF check only check outgoing
| interface no specific IP next hop address involved.
|
| Multicast and HSRP are two different entities, they don't work together
| (except that HSRP uses multicast for neighbor discover etc). In this
| topology, multicast will elect a DR in a LAN segement(either R1 or R2), DR
| will be responsible for registering and/or multicast forwarding, (because
| only DR will send PIM joins upstream or register message to RP), the other
| one will take over once it does not hear PIM hellos from its neighbor.
|
| So to answer your question, you don't need a solution, it just works.
|
| On 12/18/06, Koen Zeilstra <koen@koenzeilstra.com> wrote:
| >
| > Hey group,
| >
| > Imagine the following scenario:
| >
| > |--R1--|
| > HOSTA---R4------| |----R3---HOSTB
| > |--R2--|
| >
| >
| > On the hostA subnet HSRP runs with virt. ip 192.168.1.3.
| >
| > R1 f0/0: 192.168.1.1 HSRP prio 120
| > R2 f0/0: 192.168.1.2 HSRP prio 110
| >
| > Host B sends multicast to group 239.1.1.1.
| >
| > R4 does a RPF check on the mcast packets which are sourced from R1 f0/0
| > (192.168.1.1). The RPF check fails because the unicast route to R3 is via
| > the HSRP virt. ip address 192.168.1.3.
| >
| > How is this solved?
| >
| > 1. We can use static mroutes on R4 pointing to R1, this will work however
| > there is no failover for multicast as there is for unicast traffic. The
| > mroute will not dissapear because if R1 fails the ethernet interface on R4
| > will still be up and running. If R4 f0/0 fails there is no connectivity
| > towards R2 anyways...
| >
| > 2. I can imagine a PBR routing scenario where all traffic from R1 and R2
| > towards R4 is sourced from the HSRP virt. ip address? Not nice, since this
| > will cause all traffic to be process switched.
| >
| > Any nice solutions anybody?
| >
| > all help appreciated!
| >
| > thanks,
| >
| > Koen
| >
| > _______________________________________________________________________
| > Subscription information may be found at:
| > http://www.groupstudy.com/list/CCIELab.html
|
| _______________________________________________________________________
| Subscription information may be found at:
| http://www.groupstudy.com/list/CCIELab.html
|
This archive was generated by hypermail 2.1.4 : Tue Jan 02 2007 - 07:50:38 ART