ompi is failing netcdf-fortran tests

[jsquyres]

Per 's @opoplawski post (http://www.open-mpi.org/community/lists/devel/2015/05/17383.php), it looks like OMPIO is failing netcdf-fortran testing.

@edgargabriel Can you have a look?

[edgargabriel]

ok, I can but I am shocked to see that ompio in 1.8.5 is the default?
THat was not the plan, that should be ROMIO. THere are a huge number
of fixes on master that we can not backport to 1.8 series (since it
involves interface changes of the frameworks), so by all means, ROMIO
should be the default in 1.8 series...

THanks
Edgr

On 5/5/2015 10:12 AM, Jeff Squyres wrote:

Per 's @opoplawski https://github.com/opoplawski post
(http://www.open-mpi.org/community/lists/devel/2015/05/17383.php), it
looks like OMPIO is failing netcdf-fortran testing.

@edgargabriel https://github.com/edgargabriel Can you have a look?

—
Reply to this email directly or view it on GitHub
#571.

Edgar Gabriel
Associate Professor
Parallel Software Technologies Lab http://pstl.cs.uh.edu
Department of Computer Science University of Houston
Philip G. Hoffman Hall, Room 524 Houston, TX-77204, USA

Tel: +1 (713) 743-3857 Fax: +1 (713) 743-3335

[rhc54]

I fixed it

[jsquyres]

Slightly less terse than @rhc54: it was an accident that OMPIO was the default on v1.8. It was fixed in open-mpi/ompi-release@02bdb63. :-)

[edgargabriel]

thanks! I will still have a look at it, since there is always a chance that people might use it accidentally, e.g. if they configure without profiling interface, ROMIO can not be used and it will default to omio.If it is something that can be fixed without the changes in master, I am happy to provide a patch.

[jsquyres]

@edgargabriel Question: do the netcdf-Fortran tests pass on master with OMPIO?

[edgargabriel]

honestly I have not tried them, that's another reason to look into that

[rhc54]

@edgargabriel sorry for that terse reply - I was just buried in meetings yesterday, didn't have time to say a lot, but also didn't want you chasing something that I'd already fixed.

[edgargabriel]

no worries :-) even if it doesn't make it into the 1.8 series, I would still like to understand what is causing the problem

[edgargabriel]

even with the options mentioned on the mailing list, I can not reproduce the problem :-( I will continue to look into it next week, but both, the C and Fortran tests of netcdf pass for me on the 1.8 series as well as master.

[jsquyres]

@edgargabriel Any further progress on this? If not, I guess we should close the issue...?

[github-actions]

It looks like this issue is expecting a response, but hasn't gotten one yet. If there are no responses in the next 2 weeks, we'll assume that the issue has been abandoned and will close it.

[edgargabriel]

We should probably just close this ticket, this is from 2015