fixed UDUNITS error

[weiyuan-jiang]

Without this PR, GEOSldas run got an error
pe=00002 FAIL at line=00047 ApplicationSupport.F90 <status=102>
pe=00004 FAIL at line=00114 MAPL_Cap.F90

It will not affect the final results

[mathomp4]

For more context, the issue was that g5_modules was setting and then module load nco was prepending to UDUNITS2_XML_PATH. So in the end the codes saw:

UDUNITS2_XML_PATH=/usr/local/other/nco/5.1.7/Deps/share/udunits/udunits2.xml:/discover/swdev/gmao_SIteam/Baselibs/ESMA-Baselibs-7.33.0/x86_64-pc-linux-gnu/ifort_2021.13.0-intelmpi_2021.13.0/Linux/share/udunits/udunits2.xml

and udunits went nuts. It expects a single path.

So what @weiyuan-jiang is doing here is just "resetting" the UDUNITS2_XML_PATH to the one from Baselibs.

[weiyuan-jiang]

For more context, the issue was that g5_modules was setting and then module load nco was prepending to UDUNITS2_XML_PATH. So in the end the codes saw:

UDUNITS2_XML_PATH=/usr/local/other/nco/5.1.7/Deps/share/udunits/udunits2.xml:/discover/swdev/gmao_SIteam/Baselibs/ESMA-Baselibs-7.33.0/x86_64-pc-linux-gnu/ifort_2021.13.0-intelmpi_2021.13.0/Linux/share/udunits/udunits2.xml

and udunits went nuts. It expects a single path.

So what @weiyuan-jiang is doing here is just "resetting" the UDUNITS2_XML_PATH to the one from Baselibs.

@mathomp4 , After sourcing g5 modules, I see /usr/local/other/nco/5.1.7/Deps/bin is in the PATH, but can we also add /usr/local/other/nco/5.1.7/bin to the PATH ? If we have that path, we don't need to loan nco anymore

[mathomp4]

@weiyuan-jiang I mean...maybe? I don't know if the nco modulefile adds more things that just to PATH?

Note, nco is in baselibs so you could try that. You'd just need to add the Baselibs bin path to path.

Your change was the simplest though with less chance of side effects.

[weiyuan-jiang]

@weiyuan-jiang I mean...maybe? I don't know if the nco modulefile adds more things that just to PATH?

Note, nco is in baselibs so you could try that. You'd just need to add the Baselibs bin path to path.

Your change was the simplest though with less chance of side effects.

Great. I can get the command from baselib. Now I remove the loading of nco

[gmao-rreichle]

I can get the command from baselib. Now I remove the loading of nco'

Awesome! This is much cleaner. Is it ready for testing by @biljanaorescanin?

[biljanaorescanin]

All tests passed.

Just fyi in log file some glitching happened not sure why? Here is that segment from log:

===========================================================
#-----------------------
# Input tests : gnuconus
# Input log file : /discover/nobackup/borescan/SystemTests/logs/LDAS/2025-06-05/gnuconus
# Input tag : testing/borescan_pr101
# Input account : g0620
#-----------------------

Inactive Modules:

1. mpi/impi/2021.13

Checkout and build
------------------
Cloning [repo: [email protected]:GEOS-ESM/GEOSldas.git, tag: ['testing/borescan_pr101']]...done.
Running mepo in /discover/nobackup/borescan/SystemTests/builds/LDAS_GNUCONUS/CURRENT/GEOSldas ...done.
Converting checkout /discover/nobackup/borescan/SystemTests/builds/LDAS_GNUCONUS/CURRENT/GEOSldas to use g5_modules from /home/mathomp4/GitG5Modules/SLES15/7.17.0/g5_modules.gcc14.openmpi4.1.6 ...Traceback (most recent call last):
File "/home/mathomp4/SI_Team/scripts/systest/Model_Chkout_Git.py", line 83, in
main()
File "/home/mathomp4/SI_Team/scripts/systest/Model_Chkout_Git.py", line 38, in main
utils.git_change_g5modules(SRCDIR, MODTYP, G5MODF)
File "/gpfsm/dhome/mathomp4/SI_Team/scripts/systest/utils.py", line 2489, in git_change_g5modules
shutil.copy(G5FILE,ORIG_G5_FILE)
File "/usr/local/other/GEOSpyD/23.5.2-0_py3.11/2023-11-02/lib/python3.11/shutil.py", line 419, in copy
copyfile(src, dst, follow_symlinks=follow_symlinks)
File "/usr/local/other/GEOSpyD/23.5.2-0_py3.11/2023-11-02/lib/python3.11/shutil.py", line 256, in copyfile
with open(src, 'rb') as fsrc:
^^^^^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: '/home/mathomp4/GitG5Modules/SLES15/7.17.0/g5_modules.gcc14.openmpi4.1.6'
Build dir: /discover/nobackup/borescan/SystemTests/builds/LDAS_GNUCONUS/CURRENT/GEOSldas
Build log: /discover/nobackup/borescan/SystemTests/logs/LDAS/2025-06-05/gnuconus/bld.gnuconus.log_15:46:57
Building gnuconus model using cmake...done.

Build time for gnuconus model (in minutes): 13

MODEL run
---------
Work dir: /discover/nobackup/borescan/SystemTests/runs/LDAS_GNUCONUS/model/CURRENT
Run log: /discover/nobackup/borescan/SystemTests/logs/LDAS/2025-06-05/gnuconus/run.gnuconus.log_15:46:57
Running model gnuconus...done.
Comparison log: /discover/nobackup/borescan/SystemTests/logs/LDAS/2025-06-05/gnuconus/cmp_model.gnuconus.log_15:46:57
Comparing model BASELINE and CURRENT runs...PASSED.

[mathomp4]

That error might be on me. I'm not sure why it's using such an old baselibs for GCC.

I've just updated my scripts to use the latest Baselibs. Should be happier now