Add PIO namelist control for CICE #2145
Conversation
|
@junwang-noaa This PR will not resolve the issue w/ the ice ICs since that will require some development work on the downscaling function, but it will enable PIO for CICE. |
* fix CICE for bad merge which missed an update w/in a CESMCOUPLED ifdef block
FernandoAndrade-NOAA
added
Baseline Updates
@RatkoVasic-NOAA @ulmononian this is another troubleshooting case related with pdlib side. we will double check with develop and hercules side. will keep you posted. |
Testing against develop and again with spack-stack 1.6.0 on hercules both resulted in segmentation faults. |
Crashing points to the line https://github.com/NOAA-EMC/MOM6/blob/ab7bd14d209592d55490e75dbfaa61cb4a62df97/config_src/infra/FMS2/MOM_io_infra.F90#L905 |
|
/scratch1/NCEPDEV/stmp2/Jong.Kim/FV3_RT/rt_1375704/cpld_debug_pdlib_p8_gnu runs ok on hera with develop branch. @FernandoAndrade-NOAA can you rerun the case with pr? |
|
I successfully ran the pdlib debug gnu test on Hera on the 4th. See Has something changed on the system since then? |
|
Today, I get the same failure reported above, but I'll note it seems to be coming from the WAV PETS (180-257). |
|
I'm testing w/ a higher task count for Waves and it is currently running. I'm getting the sense that R8 is more sensitive to memory use. I saw it in a recent ufs-utils PR that I have been testing. |
|
The following change allows the cpld_debug_pdlib_gnu test to complete on Hera/R8/GNU The times reported in the log were |
It's passing for me now, I'm not sure what may have been the cause of this issue. I'll push hera up shortly. |
|
Ok, all logs have been pushed up, we can continue with the merge process in CICE. |
|
@BrianCurtis-NOAA Would you please create an issue for removing the WCOSS2 setting in default_vars, once we know PIO is correctly working? |
FernandoAndrade-NOAA
requested review from
BrianCurtis-NOAA,
SadeghTabas-NOAA,
jkbk2004 and
zach1221
commit f234a3e Author: Ufuk Turunçoğlu <[email protected]> Date: Tue Apr 30 11:35:25 2024 -0600 Fix for land component model (ufs-community#2191) * UFSWM - fix fully coupled land component configuration * NOAHMP - get fixed information from surface file commit 04bbc15 Author: jiandewang <[email protected]> Date: Thu Apr 25 14:52:00 2024 -0400 update MOM6 to its main repo. 20240401 commit (ufs-community#2241) * UFSWM - * MOM6 - update MOM6 to its main repo. 20240401 commit (NCAR-candidate-20240319) commit b6c576d Author: Daniel Sarmiento <[email protected]> Date: Tue Apr 23 12:24:22 2024 -0400 Merged global namelist (ufs-community#2173) * UFSWM - global_control.nml_IN has been added as the new regression test namelist template for all global regression tests. The namelist now uses pointers (i.e. @[abc]) for variables and default values have been added to the default_vars.sh script. A new section in default_vars.sh has been added (export_tiled) to account for tiled RTs that pulls the correct parameter files using the ATMRES variable. Regression tests have been modified to account for these changes. Tests that were not compatible with the GFSv17_p8 core have been disabled for now. They will be turned on as they are updated from GFSv16 to GFSv17. commit 5d2ca19 Author: WenMeng-NOAA <[email protected]> Date: Fri Apr 19 13:59:12 2024 -0400 Update upp submodule (ufs-community#2213) * UFSWM - Update inline post * FV3 - Update upp submodule for inline post commit 47c0099 Author: Brian Curtis <[email protected]> Date: Wed Apr 17 15:59:48 2024 -0400 Add bash linting to CI. Cleanup .sh scripts a bit. Address .sh bugs. Adds -v Verbose option. (ufs-community#2218) Remove nowarn Intel compiler flag (ufs-community#2225) * UFSWM - Add bash linting to CI: - uses superlinter to check for consistent bash code writing - Cleans up .sh scripts to comply with superlinter - Cleans up .sh scripts to be more consistent, easier to read. - Add's -v verbose option if debugging outputs needed, otherwise simplifies rt.sh run echo's. - Addresses smaller bugs - quota/timeout search logic adjusted. - check for dirs existing (DISKNM, STMP, PTMP) before starting. - adjustments/cleanup to ecflow/rocoto sections - rt.sh will attempt to start ecflow, and only stop ecflow if it started from rt.sh. - fix for issue where run_dir will not delete properly. * FV3: Address compiler warnings * atmos_cubed_sphere: Address compiler warnings. commit 4f32a4b Author: Rick Grubin <[email protected]> Date: Mon Apr 15 07:21:08 2024 -0600 Document ATMW / ATMAERO / HAFS WM configurations (ufs-community#2160) * UFSWM * doc/Userguide * source * conf.py * Configurations.rst * FAQ.rst * InputsOutputs.rst * Introduction.rst commit ac4445d Author: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Date: Mon Apr 15 08:59:42 2024 -0400 Bump idna from 3.6 to 3.7 in /doc/UsersGuide (ufs-community#2234) *doc/UserGuide *requirements.txt - updates inda version from 3.6 to 3.7 commit 281b32f Author: Samuel Trahan (NOAA contractor) <[email protected]> Date: Mon Apr 15 08:38:01 2024 -0400 bug fixes: kchunk3d ignored, hailwat uninitialized in dycore, tile_num wrong for nests (ufs-community#2201) * UFSWM - None. * FV3 - Write component will use kchunk3d. Model init sends the right tile number to CCPP. * atmos_cubed_sphere - Initialize the hailwat variable. Pass global_tile index to model. commit 8a5f711 Author: Denise Worthen <[email protected]> Date: Thu Apr 11 13:32:26 2024 -0400 Add PIO namelist control for CICE (ufs-community#2145) Update to CICE-Consortium/CICE aca8357. Adds implementation of namelist PIO options for CICE commit 45c8b2a Author: JONG KIM <[email protected]> Date: Thu Apr 4 19:49:13 2024 -0400 Hotfix/cubed sphere hash fix: HAILCAST diagnostic code (units issue) (ufs-community#2223) cubed_sphere hash update: f060e85 for a bug- fix in the HAILCAST diagnostic code (units issue) commit 26e6db6 Author: Denise Worthen <[email protected]> Date: Wed Apr 3 19:57:08 2024 -0400 Enable cpl_scalars export from ATM and NoahMP for use by CMEPS (ufs-community#2175) * CMEPS - allow additional dimension in cpl_scalars for CSG and regional ATM domains for use in mediator history files * CMEPS - fix mapping mask for lnd->atm * FV3 - add export of cpl_scalars * NOAHMP - add export of cpl_scalars commit 1411b90 Author: Dusan Jovic <[email protected]> Date: Mon Apr 1 18:04:44 2024 -0400 Update module_write_netcdf to avoid hangs in RRFS runs (ufs-community#2193) * UFSWM - Update module_write_netcdf to avoid hangs in RRFS runs * FV3 - Update module_write_netcdf to avoid hangs in RRFS runs commit 87c27b9 Author: Matthew Masarik <[email protected]> Date: Fri Mar 29 15:23:42 2024 -0400 WW3 feature: Langmuir turbulence parameterization (ufs-community#2195) * WW3 - Langmuir turbulence parameterization commit c54e986 Author: Samuel Trahan (NOAA contractor) <[email protected]> Date: Wed Mar 27 16:11:03 2024 -0400 regression test system bug fixes, eliminate MOM6 warnings (ufs-community#2197), add xr_cnvcld flag to FV3 (ufs-community#2185) (ufs-community#2202) * UFSWM - atparse.bash: correctly handle input that doesn't end with an end-of-line character. Fix some bugs in Rocoto support and clean up rt.sh. * FV3 - namelist flag xr_cnvcld to control if suspended grid-mean convective cloud condensate should be included in cloud fraction and optical depth calculation in radiation in the GFS suite * ccpp - physics-level changes to implement new namelist variable * MOM6 - update MOM6 code to eliminate all compiler warnings
Commit Queue Requirements:
Description:
Adds namelist configuration of PIO options for CICE and switches from using the netcdf (serial)
CICE_IOAPI to using PIO. This means all netCDF goes through a different API in CICE (ie, underinfrastructure, the code inio_pio2is built and used instead of the code inio_netcdf). Also adds netcdf and PIO error checking.The available IO types are:
Independent PIO settings are used for restart and history files. New export variables are added to
default_varsto allow all available settings to be used. Settings which are not relevant to a given PIO IOtype are ignored. If non-valid settings are given, appropriate values will be substituted.The available settings are
(history,restart)_format:
cdf1,2,5,hdf5(netcdf4),pnetcdf1,2,5. For cdf and pnetcdf, the 1,2,5 refer to netcdf3-classic, netcdf3-64-bit-offset and netcdf3-64bit-data.(history,restart)_iotasks:
(history,restart)_stride:
(history,restart)_rearranger:
(history,restart)_root:
(history,restart)_chunksize, (history,restart)_deflate:
Testing
The switch to PIO was not expected to change baselines compared to a baseline generated using the netCDF API. However, because of issues with the IC files, change were found in some, but not all tests using CICE.
All tests using CICE should therefore get new baselines to ensure they are now consistent w/ using the PIO interface.
A test was run to verify that a baseline generated using PIO did reproduce itself. All tests using CICE passed.
Commit Message:
Priority:
Git Tracking
UFSWM:
Sub component Pull Requests:
UFSWM Blocking Dependencies:
Changes
Regression Test Changes (Please commit test_changes.list):
Input data Changes:
Library Changes/Upgrades:
Testing Log: