jkrasting
commented
9 years ago Not good news! I agree - waiting for a FRE/MOAB fix could take too long. That's the preferred solution, so I don't think we should give up on that option even if we find a band-aid ourselves.
Dr. John Krasting Physical Scientist (NOAA Federal) NOAA/Geophysical Fluid Dynamics Laboratory Biogeochemistry, Ecosystems, and Climate Group Princeton University Forrestal Campus 201 Forrestal Road Princeton, NJ 08540
P. (609) 452-5359 F. (609) 987-5063
On Tue, Dec 9, 2014 at 6:50 PM, Niki Zadeh notifications@github.com wrote:
Due to a limitation of moab it is safer to have at most one refineDiag script in the xml.
The limitation is as follows: FRE adds the pathnames of all refineDiag scripts to some environment variable (on top of some other variables). That ENV then get prepended to all the commands that appear in the gfdl platform section. The whole thing is then dispatched by moab (on gaea) to run on PAN (at gfdl) as the pp.starter job. BUT moab has a hard set limit for the length of that ENV and if it is longer it is just cut (most likely in the middle of a command string). Hence the commands that we put in the section of gfdl platform could (and did) become useless and lead to PP errors.
We have run into this limitation in our CM4 runs.
Since the refineDiag scripts pathnames are part of this ENV it is advisable to have at most one refineDiag to save something like 130 chars and avoid these situation as much as possible. A band-aid, I know, but should we rather wait for moab/FRE to fix the issue? That would take a few months at least.
— Reply to this email directly or view it on GitHub https://github.com/CommerceGov/NOAA-GFDL-MOM6-examples/issues/12.
adcroft
jwdGFDL
Due to a limitation of moab it is safer to have at most one refineDiag script in the xml.
The limitation is as follows: FRE adds the pathnames of all refineDiag scripts to some environment variable (on top of some other variables). That ENV then get prepended to all the commands that appear in the gfdl platform section. The whole thing is then dispatched by moab (on gaea) to run on PAN (at gfdl) as the pp.starter job.
BUT moab has a hard set limit for the length of that ENV and if it is longer it is just cut (most likely in the middle of a command string). Hence the commands that we put in the section of gfdl platform could (and did) become useless and lead to PP errors.
We have run into this limitation in our CM4 runs.
Since the refineDiag scripts pathnames are part of this ENV it is advisable to have at most one refineDiag to save something like 130 chars and avoid these situation as much as possible. A band-aid, I know, but should we rather wait for moab/FRE to fix the issue? That would take a few months at least.