[mpich-discuss] mpdboot hanging

[Pavan Balaji]

Jason,

On 02/15/2010 09:27 AM, Jason Palmer wrote:
> Yes that fixes it, thanks. Is there reason to prefer hydra over mpd aside
> from fewer bugs? Say in terms of process maintenance / efficiency in
> cleaning up all processes associated with a run (I'm using sge on linux) or
> speed? My preference would be to use the manager that does the best job of
> cleaning up broken runs, or provides the best facility for automatically
> killing all of a run's processes on all its nodes. Does hydra improve on the
> way Mpich1 did things?

There are a few reasons we recommend Hydra:

1. Hydra has better performance and a cleaner/more-easy-to-use interface
as compared to MPD. But if you are comfortable with MPD in this respect,
you can ignore Hydra for now (though, in the 1.3.x series, Hydra is the
default process manager, replacing MPD).

2. With respect to how reliable it is in cleaning up stray processes and
aborting in corner cases, it has been quite reliable in our tests. But,
in general, this hasn't been as widely used as MPD. So, till it is, we
can't really say.

3. But most importantly, Hydra is where most of our development/support
resources are. So, if you find a bug, it's more likely to get fixed
early in Hydra.

 -- Pavan

-- 
Pavan Balaji
http://www.mcs.anl.gov/~balaji