[openib-general] [ANNOUCEv2] OpenIB OpenSM 1.1.0: trunk now supports 1.8.0 features

Hal Rosenstock halr at voltaire.com
Tue Sep 13 18:31:11 PDT 2005


On Tue, 2005-09-13 at 20:12, Troy Benjegerdes wrote:
> On Tue, Sep 13, 2005 at 12:19:30PM -0400, Hal Rosenstock wrote:
> > On Tue, 2005-09-13 at 12:15, Troy Benjegerdes wrote:
> > > We just had a node crash on our network, and it caused our OpenSM to
> > > stop working.. we were running version openib-1.0.0..
> > 
> > Can you define stop working (more details) ? Are there any logs ?
> > 
> > > I suppose this means I should start beating up on 1.1.0 now, right?
> > 
> > Yes but the same issue might still exist. Can you reproduce it on the
> > OpenSM you are running on now and then move up and see if it still
> > exists ?
> 
> Stop working as in IPoIB arp seems to stop.

I suspect that the multicast tree for the broadcast group somehow gets
broken. 

Were you running any other ULPs other than IPoIB ?

> I've got a log now of the latest opensm-1.1.0 attached.
> 
> The time (was) off on that machine, FYI. 
> 
> At the log entry 'Sep 13 12:06:55', I plugged in the node that is hung/crashed
> .. which caused a bunch of opensm errors.. 

Thanks. That helps orient me. What was the opensm crash ? The log just
ends abruptly.

> I have since unplugged that
> node, and can put it back in tommorow if you want more debug info.

Great. More later on the log itself...

-- Hal




More information about the general mailing list