[openib-general] osm: More simulation faiures on trunk

[Hal Rosenstock]

Hi Eitan,

On Wed, 2006-12-06 at 16:56, Eitan Zahavi wrote:
> Hi Hal,
> 
> Hal Rosenstock wrote:
> > Hi Eitan,
> >
> > On Wed, 2006-12-06 at 13:02, Eitan Zahavi wrote:
> >   
> >> Hi Hal,
> >>
> >> Looks like the osm.fdbs file is now created with "UNREACHABLE" mark when 
> >> opensm
> >> is invoked with updn routing engine.
> >>     
> >
> > Are you referring to certain LIDs being UNREACHABLE like this:
> > LID    : Port : Hops : Optimal
> > 0x0001 : UNREACHABLE
> > 0x0002 : UNREACHABLE
> > 0x0003 : 000  : 00   : yes
> > 0x0004 : 001  : 02   : yes
> > 0x0005 : 003  : 02   : yes
> > 0x0006 : 001  : 01   : yes
> > 0x0007 : UNREACHABLE
> > 0x0008 : UNREACHABLE
> > 0x0009 : UNREACHABLE
> > 0x000A : 001  : 02   : yes
> > 0x000B : 003  : 02   : yes
> >
> > So should UNREACHABLE LIDs just not be put into the file ? Or is it
> > something else ?
> >
> >   
> The UNREACHABLE is fine.
> The problem is that ALL LFTs are full of UNREACHABLE. Actually there are 
> no reachable nodes ...

Weird. It works on my topology (with UPDN).

> >> I will be working on finding what changed between OFED 1.1 and the trunk.
> >>     
> >
> > It was likely introduced by the changes to the routing engines committed
> > yesterday and sent on the last in late Novemeber. git-bisect can help
> > isolate exactly which change.
> >   
> Thanks I will follow that trail
> >   
> >> This is another cause for the failure of all osmMulticastRoutingTest and 
> >> osmStability tests runs.
> >>     
> >
> >   
> >> Another one would be the change of the osm.mcfdbs which is parsed by 
> >> IBDM too.
> >>     
> >
> > Are you asking about the other patch again ?
> >   
> Do you have an estimate for when my patch will be merged ?

I already answered this in an earlier email.

-- Hal

> > -- Hal
> >
> >   
> >> Eitan
> >>
> >>     
> >
> >
> > _______________________________________________
> > openib-general mailing list
> > openib-general at openib.org
> > http://openib.org/mailman/listinfo/openib-general
> >
> > To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
> >   
>