I'm unavailable and in no condition to participate for a bit. You might want to consider waht you want us to do behind the scenes. In other words, if FT is enabled and things have a problem - or if you simply don't have the nodes you want/need, e.g., using hard vs soft when doing mpiexec - do you want to automatically exploit PMIx_Allocate_resources to attempt to obtain them? Should this only happen if pass an explicit flag? Or never happen since the user didn't call some (possibly new?) MPI function to explicitly invoke that PMIx function?

Ralph


On Nov 7, 2022, at 7:20 AM, Pritchard Jr., Howard via mpiwg-sessions <mpiwg-sessions@lists.mpi-forum.org> wrote:

Hi All,
 
We’ll meet today and pick up where we were two weeks ago discussing FT aspects of MPI_Comm_create_from_group, spawn, hard vs soft mpiexec (explicit or implicit in MPI_Comm_spawn call), etc.
 
Also if there’s time we’ll briefly discuss upcoming SC events, etc.
 
Howard
 
 

—
 
<image001.png>
Howard Pritchard
Research Scientist
HPC-ENV
 
Los Alamos National Laboratory
 
 
 
_______________________________________________
mpiwg-sessions mailing list
mpiwg-sessions@lists.mpi-forum.org
https://lists.mpi-forum.org/mailman/listinfo/mpiwg-sessions