I should hope bisecting would be a last resort. The simplest interim solution is to set OMPI_MCA_routed=direct in your environment.
I’ll take a look at a more permanent solution in the morning. > On Oct 30, 2016, at 6:33 PM, Pritchard Jr., Howard <howa...@lanl.gov> wrote: > > Hi Folks, > > While trying to solve a different problem, I optimistically tried to use > head-of –master to work on that problem. Now I’ve found a new problem > with master when trying to do a direct launch with SLURM, srun: > > [nid00012:09456] [[27960,0],0] ERROR: Failed to identify the local daemon's > URI > [nid00012:09456] [[27960,0],0] ERROR: This is a fatal condition when the > radix router > [nid00012:09456] [[27960,0],0] ERROR: has been selected - either select the > unity router > [nid00012:09456] [[27960,0],0] ERROR: or ensure that the local daemon info is > provided > [nid00012:09456] [[27960,0],0] ORTE_ERROR_LOG: Fatal in file > base/ess_base_std_app.c at line 194 > - > > Any ideas what’s going on here? With my configure pmix/s1 is selected for > pmix component, > but death follows soon thereafter. I’m confused why with direct launch Open > MPI should > care at all about what the local deamon’s URI is. > > Is there a way to avoid this problem when using direct launch? I would do a > git bisect > but I’ve no time for such activities at the moment. > > Thanks for any suggestions, > > Howard > > -- > Howard Pritchard > HPC-DES > Los Alamos National Laboratory > > _______________________________________________ > devel mailing list > devel@lists.open-mpi.org > https://rfd.newmexicoconsortium.org/mailman/listinfo/devel
_______________________________________________ devel mailing list devel@lists.open-mpi.org https://rfd.newmexicoconsortium.org/mailman/listinfo/devel
