[ofa-general] Re: [PATCH 0/3] clean up "__get_mgrp_by_mgid" and add option --consolodate_ipv6_snm_req

[Hal Rosenstock]

On Tue, 2008-01-15 at 01:31 +0000, Sasha Khapyorsky wrote:
> On 17:08 Mon 14 Jan     , Hal Rosenstock wrote:
> > On Tue, 2008-01-15 at 00:42 +0000, Sasha Khapyorsky wrote:
> > > On 16:09 Mon 14 Jan     , Hal Rosenstock wrote:
> > > > On Tue, 2008-01-15 at 00:11 +0000, Sasha Khapyorsky wrote:
> > > > > On 11:45 Mon 14 Jan     , Ira Weiny wrote:
> > > > > > The following 3 patches are a much cleaner implementation of what I sent to the
> > > > > > list on Friday.
> > > > > > 
> > > > > > The first 2 patches are just code clean up and I feel should be applied.  The
> > > > > > 3rd adds the option --consolodate_ipv6_snm_req which causes all IPv6 Solicited
> > > > > > Node Multicast requests to be grouped into one MCast group per partition.
> > > > > 
> > > > > All three patches are applied. Thanks.
> > > > 
> > > > This seems a little fast for me.
> > > 
> > > It looks like a nice start for me.
> > > 
> > > > I didn't think we were even done with
> > > > the discussion yet.
> > > 
> > > Sure, and we can continue from this point.
> > > 
> > > (Ira stated that this patch series are not a final solution, but already
> > > provides sophisticated functionality right now. The feature itself is
> > > optional and I don't see a big risk here.)
> > 
> > It may work in Ira's configuration but I'm not even sure it follows the
> > IBA spec.
> 
> Why it is not?

I'll dig this out shortly; it's been discussed on the list before.

> > It may also cause issues if misused (e.g. MTU, rate,
> > partitions, etc.).
> 
> Right, but it is optional. Default OpenSM behavior is not changed.

Doesn't stop someone else from misusing it; e.g. IPv6 is broken, or in
the presence of partitioning, certain things don't work right, etc. I
may be wrong but I think depending on how quickly the shortcomings are
addresses or not, this becomes a support issue.

-- Hal

> Sasha
> _______________________________________________
> general mailing list
> general at lists.openfabrics.org
> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
> 
> To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general