[mvapich-discuss] Forcing a dump?

[Jonathan Perkins]

Hi Ed,
Did you rebuild the application with the new debugging build of
mvapich2?  The output should give more information in this case.

Also what launcher are you using?  If you're using mpirun_rsh you'll
want to put the envs on the command line instead of just exporting
before hand.

Ex:
mpirun_rsh -n 2 host1 host2 MV2_DEBUG_SHOW_BACKTRACE=1 ./application

On Thu, Jul 26, 2012 at 02:27:24PM +0000, Ed Wahl wrote:
> Howdy.  I've been dragging my feet on debugging an annoying problem with a third party application that misbehaves with mvapich2 when it hits the MV2_ON_DEMAND_THRESHOLD value.
> 
> I've enabled debugging--enable-debuginfo  --enable-g=dbg, and  export MV2_DEBUG_SHOW_BACKTRACE=1    export MV2_DEBUG_CORESIZE=unlimited.  But I still just get :
> 
> [cli_1]: aborting job:
> Fatal error in MPI_Init:
> Other MPI error
> 
> I do have a small backtrace before this, but it does not seem to be that useful. 
> 
> 
> Is there another way to force mvapich2 to give me any more information than this from the fault?
> 
> 
> Ed Wahl
> OSC
> 
> _______________________________________________
> mvapich-discuss mailing list
> mvapich-discuss at cse.ohio-state.edu
> http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
> 

-- 
Jonathan Perkins
http://www.cse.ohio-state.edu/~perkinjo