[c-nsp] Multicast issues on 7600s with WS-6748-sfp blades

[John Neiberger]

On Thu, Jul 29, 2010 at 7:28 AM, Asbjorn Hojmark - Lists
<lists at hojmark.org> wrote:
> On Thu, 29 Jul 2010 08:29:18 -0400, you wrote:
>
>> I have seen some pretty low level problems with multicast packets
>> being dropped on the floor by the hardware forwarded on 67xx
>> linecards. The bug seems to appear when changes that effect the asic
>> (turning on "mls qos" for example). Resetting the sup engine doesn't
>> resolve it. Everything looks good at the CEF/mfib level, and all
>> counters show the packet count increasing, but the packets never get
>> forwarded out of the linecard.
>
> Sounds a bit like CSCtc24959, which I saw with a couple of customers.
> It was typically seen on changes to topology (i.e. rerouting).
>
> Flipping the replication mode:
>  mls ip multicast replication-mode ingress
>  no mls ip multicast replication-mode ingress [1]
> causes reprogramming of the hardware, which fixed the problem without
> having to reset the whole blade. Note that it *is* traffic-affecting
> for all active multicast flows.
>
> CSCtc24959 is fixed in SRD4 and will not be integrated in earlier
> releases, or so I was told.
>
> HTH,
> -A
> [1] Or the other way round if you use ingress replication
>

That sounds a lot like what we're seeing, but we're not on the SR
train. We're running 12.2(18)SXF9 at the moment.