Hello All. I'm using OMPI 4.0.1. I run MPI_Spawn() as a singleton. I need to run, in the same instance of my app, different spawn configurations. In this case I run first using my hostfile (i.e. setting a MPI_Info object with MPI_Info_create() function, and then setting the attributes with MPI_Info_set) that allocates the process ok; then I run spawn again but with MPI_INFO_NULL on spawn function because I want to allocate process just in local, but the allocation ignores this and It puts the process likewise the previous MPI_info setting. Its like the spawn function was it "cached". Any idea? Thanks in advance and regards.
