[c-nsp] Multicast Issue

[Jay Ford]

On Thu, 27 Apr 2006, Tim Stevenson wrote:
> The similarity here has to do with the concept of destination indexes
> in 6500, but these two problems have 2 separate causes at a higher level.
>
> The CSCds22021 has to do with flood traffic where the tx source port
> is also a member of the floodset (the flood index for the particular
> packet) for received packets. Source filtering prevents the flood
> packet from going out the original port, but doesn't prevent it from
> ending up on the SPAN dest port, which has the same indexes
> programmed as the SPAN source.
>
> On the other hand, the PIM problem has to do with the destination
> index that PIM packets are sent to when snooping is enabled, which is
> basically a flood to all ports. Agree this is indeed suboptimal.

Thanks for the ACK on suboptimal.  ;^)

> In any case, PIM is immaterial to this problem since there is only
> one router in the picture, so I would not discount the usefulness of SPAN.

Agreed.  I just wanted to point out the possible presence of some very
puzzling traffic to help others avoid the type of confusion we went through.

> Also, even for the PIM problem, you could always configure the SPAN
> dest port as a 1Q trunk & then you'd at least be able to figure out
> which PIM packets came from the vlan you are interested in.

I think the tag usage is not an option in our case because the port being
monitored is routed (not a switchport), so there is either no tag or some
internal tag which is not actually visible/usable.  That's as much as I
remember; you probably know more about it.

Jay