[Mochi-devel] Server as client

[Latham, Robert J.]

On Thu, 2019-12-05 at 17:08 +0000, Jerome Soumagne wrote:
> Hi Philip
> 
> Yes you can send RPCs to yourself, we do have optimizations for that
> case and it will bypass the network layer. Just make sure you have
> compiled mercury with MERCURY_USE_SELF_FORWARD turned ON in cmake.

note that if you built Mercury via spack, this option is already set
(unless you explicitly asked to turn it off)

==rob

> Thanks
> 
> Jerome
> 
> On 12/5/19, 11:03 AM, "mochi-devel on behalf of Philip Davis" <
> mochi-devel-bounces at lists.mcs.anl.gov on behalf of 
> philip.e.davis at rutgers.edu> wrote:
> 
>     Hello,
>     
>     Apologies if these have been asked before and I missed it:
>     
>     1. Is there any obstacle to a process being both an origin and a
> target for Mercury RPC calls?
>     
>     2. If not, is it legal for the same process to be both the orgin
> and target of the *same* RPC call? The use case I’m picturing is
> implmenting some sort of structured, distributed object store that
> uses keyval or similar as a backing store. It might be convenient to
> colocate these two services in the same process group, and distribute
> the backing store throughout the process group. As such, it seems
> feasible that the object store would need to access the backing
> storage administered by its own process, so I’m wondering if that
> case needs to be handled outside Mercury, or if it can be implemented
> as if the backing store was not process-local.
>     
>     3. If the answer to the previous is yes, then are there any
> optimizations in Mercury to recognize and handle this case (i.e. the
> case where the origin and target are the same process)?
>     
>     Thanks,
>     Philip
>     _______________________________________________
>     mochi-devel mailing list
>     mochi-devel at lists.mcs.anl.gov
>     https://lists.mcs.anl.gov/mailman/listinfo/mochi-devel
>     https://www.mcs.anl.gov/research/projects/mochi
>     
> 
> _______________________________________________
> mochi-devel mailing list
> mochi-devel at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/mochi-devel
> https://www.mcs.anl.gov/research/projects/mochi