[ofa-general] Multicast traffic generates Bad P_Key trap in SM when working in partial member setup

[Yevgeny Kliteynik]

Hal Rosenstock wrote:
> On Thu, 2008-06-12 at 15:58 +0300, Yevgeny Kliteynik wrote:
>> AFAIK, by default, OpenSM should
>> configure pkeys on switch ports that are connected to these HCAs,
>> so that partial member wouldn't get packet from another partial
>> member.
> 
> In one sense, this is a workaround. It still wastes the network
> bandwidth as it has to get all the way to the egress switch before
> dropping.

Right, I wasn't proposing solution, but trying to understand the setup.

-- Yevgeny

> So it's going to come down to how this scenario is decided to be dealt
> with in the spec and then in the implementations.
> 
> -- Hal
> 
>