[MITgcm-support] Problem with *nc output

[Galen McKinley]

Dear support list -

I'm running the MITgcm in the North Atlantic with Dutkiewicz's ecosystem and
carbon system, run in parallel on 16 processors. Its running fine, but I am
having strange problems with the output at daily resolution. I've run tests
3 times and the same thing occurs each time.

Many of the output files are fine, but some of them are simply not filled in
with data. Its just zeros. This starts at random times in the run, changing
from tile to tile and time within the run. The run isn't corrupted by it.
Matlab will still open the files, but NCO operators won't accept them.

I am thinking this is an issue with the RAM on the system -- I have 8GB on
each node. Any thoughts of how to understand what's going on better or how
to fix it? 

Thanks,
Galen

--------------------------------------------------------------------------
Galen A. McKinley, Assistant Professor
Atmospheric and Oceanic Sciences                         608 262 4817
University of Wisconsin - Madison        www.aos.wisc.edu/~galen/
--------------------------------------------------------------------------