[mpich-discuss] How to spawn a non-MPI process with MPI_Comm_spawn

[Mccall, Kurt E. (MSFC-EV41)]

Ken,

Issue created.  Can you suggest any other way to launch non-MPI processes?

Thanks,
Kurt

-----Original Message-----
From: Raffenetti, Kenneth J. <raffenet at mcs.anl.gov> 
Sent: Tuesday, January 7, 2020 9:32 AM
To: discuss at mpich.org
Cc: Mccall, Kurt E. (MSFC-EV41) <kurt.e.mccall at nasa.gov>
Subject: [EXTERNAL] Re: [mpich-discuss] How to spawn a non-MPI process with MPI_Comm_spawn

We do not support such a key at this time, but it seems reasonable to add. Could you create an issue on our GitHub? 
https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_pmodels_mpich&d=DwIF-g&c=ApwzowJNAKKw3xye91w7BE1XMRKi2LN9kiMk5Csz9Zk&r=6cP1IfXu3IZOHSDh_vBqciYiIh4uuVgs1MSi5K7l5fQ&m=H8FYbetRLxBRSaC9l0ktwX7ok3zyPRdhUlWXZZq9nsc&s=PFcC_rnOjUzKTea_1RX-vu-Nz1gCgfNP5fqj6BFly7I&e= 

Thanks,
Ken

On 1/6/20 9:50 AM, Mccall, Kurt E. (MSFC-EV41) via discuss wrote:
> I'm attempting to use MPICH to manage a number of non-MPI processes.    
> The latter call system() and fork(), which as I understand it may  
> cause undefined behavior if they were spawned as MPI processes.   How 
> do I use
> MPI_Comm_spawn() to launch these processes?
> 
> For example, OpenMpi has a key for the "info" field named 
> "ompi_non_mpi", which causes a non-MPI process to be spawned.   Is 
> there something similar in MPICH?
> 
> Thanks,
> 
> Kurt
> 
> 
> _______________________________________________
> discuss mailing list     discuss at mpich.org
> To manage subscription options or unsubscribe:
> https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.mpich.org_m
> ailman_listinfo_discuss&d=DwIF-g&c=ApwzowJNAKKw3xye91w7BE1XMRKi2LN9kiM
> k5Csz9Zk&r=6cP1IfXu3IZOHSDh_vBqciYiIh4uuVgs1MSi5K7l5fQ&m=H8FYbetRLxBRS
> aC9l0ktwX7ok3zyPRdhUlWXZZq9nsc&s=cUS7Q28AAksa_KLLMwTztXCU3BiO0lIEKb-bM
> Wpcvfk&e=
>