I've reverted the default config for now - not sure when I'll get another chance to tackle it, but hopefully soon. Looks like there is a problem in the TCP oob where we lose messages every now and then.
On Jun 11, 2012, at 11:44 AM, Jeff Squyres wrote: > On Jun 11, 2012, at 1:42 PM, Ralph Castain wrote: > >>>> will have to look more at the loop_spawn issue later. >>> The original one I reported, I assume? I see similar stacks on segfaults >>> with a variety of tests. So, I think it's not specific to loop_spawn. >> >> It's a race condition in the oob, so I'm looking at it. > > FWIW, I'm seeing some hangs on the trunk this morning, too -- even when I > specify the tcp oob. Mebbe 1 run in 10. I haven't had time to diagnose -- > I'm working on something else... > > -- > Jeff Squyres > jsquy...@cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > > > _______________________________________________ > devel mailing list > de...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/devel
