[ewg] RE: [ofa-general] Re: [GIT PULL] OFED 1.2: CM scaling fixes

[Michael S. Tsirkin]

> Quoting Sean Hefty <mshefty at ichips.intel.com>:
> Subject: Re: [ewg] RE: [ofa-general] Re: [GIT PULL] OFED 1.2: CM scaling fixes
> 
> >>I think it is very good that you changed the define to a module 
> >>parameter, but shouldn't the default value be smaller (21). (15*32 sec = 
> >>8 Min) seems like quite a long time.
> >>
> >>Ishai
> >
> >
> >Hmm, indeed. Sean, wouldn't 1 minute be sufficient?
> 
> This really depends on the app.  For scale-up testing, we need to include 
> the time that it takes an app to come down and fetch the connection event, 
> process it, and respond.  When new connections form along with data 
> transfers, the app may be slow to handle this.
> 
> I have no objections to setting the default lower though.  But this is a 
> max timeout.  If the timeout in the path record/CM messages is smaller, 
> that would be used instead.

OK, so can you change the default to lower value in your branch?

-- 
MST