On Feb 11, 2008, at 4:33 PM, David Gunter wrote:
> We have a number of patches and files to be added to ROMIO to make it
> work with recent releases of the Panasas file system. We have reached
> a point where the stock ROMIO included in Open-MPI no longer works for
> what we need. I know that the version of ROMIO forged into the bowels
> of OMPI is a beast to try and patch or mend so that is something we
> won't attempt.
ROMIO upkeep has been a hot topic over the past week. Two patches
went in to the trunk today that are steps in the right direction
towards being able to keep up with official ROMIO releases easier.
The patches could probably be taken to the v1.2 branch without too
much trouble. I will volunteer to do this if it would be helpful to
Another option is to have a LANL-specific ROMIO component with a
custom-patched ROMIO. This was discussed yesterday; Ralph Castain was
going to ponder the possibilities.
We also talked about continually updating the ROMIO in Open MPI (to
officially-released versions). U. Houston has volunteered to do this
for the future, but this is pending a student's visa issues being
cleared up. That could be tomorrow or a few months from now; there's
unfortunately no way to tell.
This is where the OMPI/ROMIO conversation is right now -- I suspect
that it will continue to unfold more in short order...
> Thus we have two choices here at LANL. Either we drop support and no
> longer provide OMPI to our user community and switch to MVAPICH2 for
> our only MPI on systems, or we can try and build OMPI against an
> externally maintained ROMIO.
Per what Brian mentioned, I think that since Brian left LANL, the
"build against an external ROMIO" functionality never made it over to
the v1.2 branch. There's no technical reason it can't be ported over;
someone just has to take the work Brian did and massage the bits into
a v1.2-compatible patch. I'm *guessing* that it's not too much work,
but I unfortunately do not have the cycles to volunteer to do this.