From: Hobbs (deadheadblues@gmail.com)
Date: Mon Jul 28 2008 - 21:32:34 ART
Thank you Huan. That explains it, sometimes I forget that autorp traffic is
still just multicast packet following the same rules as all multicast
traffic.
So tunnels should be able to solve this right? In my example it did for one
ping and then it stopped...are there any caveats I should know? Should it
work with GRE Tunnels? Perhaps it worked once because there was some
leftover entries when I had the Hub as the MA, then I switched it to
R3...can't seem to recreate it.
Here is my example topology:
[R2]
|
[R1]---FR---[R3]---[R4]
In this case R1 is the hub and also RP-candidate, and R3 is the MA.
I have a join group on R2's loopback 232.0.0.2.
When R4 sends pings to this it fails...
R2 and R3 are PIM neighbors through the tunnel
thanks
On Mon, Jul 28, 2008 at 5:50 PM, Huan Pham
<Huan.Pham@peopletelecom.com.au>wrote:
>
> Hi Hobbs,
>
> It is required because of the way multicast dense mode works. When a
> router receive a mpacket, it floods the multicast packet out to all
> other interface, except the interface it is received on.
>
> Let's look at the example provided at the given link, if MA is placed
> under a spoke router R2, then RP discovery message sent via 224.0.1.40
> can be propogated to R2, then to Hub R1, and further to any router
> behind R1, but not to other spoke. This is because, when R1 receives
> this "RP discovery message", it floods out all other interface(s), but
> not S0.1
>
> If you do sh ip mroute on R1, you will see that the interface S0.1 is
> not in the OIL (outgoing interface list) for the group 224.0.1.40.
>
> Similar candidate RP and mapping agent should be communicatable via
> 224.0.1.39. If they are both spokes, can only be reachable via a hub,
> then the MA cannot not event see the RP presence.
>
> Huan
>
>
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> Hobbs
> Sent: Tuesday, 29 July 2008 8:44 AM
> To: Cisco certification
> Subject: Re: Why do mapping agents need to be connected to all PIM
> routers in NBMA?
>
> btw, I did get it to work with GRE tunnel between the non-hub MA and the
> regular PIM spoke. I made these two PIM neighbors over the GRE tunnel.
>
> Just wondering why it is necessary even when you have the scope
> argument.
>
> On Mon, Jul 28, 2008 at 4:36 PM, Hobbs <deadheadblues@gmail.com> wrote:
>
> > Hello,
> >
> > Referring to this doc:
> > http://www.cisco.com/en/US/docs/ios/solutions_docs/ip_multicast/White_
> > papers/frm_rlay.html
> >
> > It says that:
> >
> > "All candidate RPs must be connected to the MA.
> > All MAs must be connected to all PIM routers."
> >
> > Why is the necessary? Isn't this why we have the scope argument on the
>
> > send-rp-announce and send-rp-discovery commands?
> >
> > Suppose you have a spoke configured as an RP-candidate or MA, are
> > there workarounds to make it work?
> >
> > Thanks,
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
Blogs and organic groups at http://www.ccie.net
This archive was generated by hypermail 2.1.4 : Mon Aug 04 2008 - 06:11:57 ART