[c-nsp] Multicast state behavior differences between 7600 and ASR9k

[John Neiberger]

On Mon, Feb 28, 2011 at 9:22 AM, Phil Mayers <p.mayers at imperial.ac.uk> wrote:
> On 28/02/11 15:35, John Neiberger wrote:
>
>> Besides, it's my opinion that if there is a multicast sender on a
>> port, then multicast state exists and we should be able to use a
>> simple multicast show command to view the state.
>
> If you're using SSM, what state exists when nothing is joined? There are no
> PIM asserts being sent to the PIM RP i.e. no (*,g).
>
> Clearly the 7600 has a forwarding architecture that necessitates a null-OIF
> MRIB entry (guess: to prevent CPU punts); presumably the ASR9k doesn't have
> the same architecture, and just drops packets without an MRIB entry.
>
> That doesn't mean it wouldn't be useful of course; but presumably the ASR9k
> would need to be leaking packets to the CPU to build this state in the first
> place, which has its own set of problems.

That's an excellent point. This is SSM, but we have so many sources,
it's very useful to see entries in the mroute table in IOS, even when
there are no active joins. We need to be able to verify which groups a
source is sending to. So, in IOS XR, it appears that our options are
to either run NetFlow and look at the cache, or we can manually join
groups during troubleshooting. I'd much rather be able to just look at
the mrib in the same we can can look at the mroute table on the 7600.
Maybe they'll add "show mrib route null-oil" for us.  :-)