Fix CI/CD workflow#975
Conversation
|
As of now, the Docs workflow in the GitHub CI is 100% working. There is still an issue with CCPP in the GCC workflow. |
|
Ok, this should be ready to get reviewed. All issues have been sorted out. Main issues were the change to the ESMF library (going from CMakeModules repo to ESMF) and splitting the FV3 library to ufsatmfv3. I ran RTs on Hercules. There were no changes to fv3atm but I was checking to be cautious. Here's the CI/CD passing on my develop fork of fv3atm. |
| include_directories(${ESMF_ROOT}/include) | ||
| link_directories(${TEMP_DIR}) |
There was a problem hiding this comment.
Why are these two (include_directories and link_directories) needed? If we use FindESMF.cmake from esmf's install tree, find_package (at line 65) should properly set esmf import targets.
There was a problem hiding this comment.
There we two separate issues that I was running into when Github was running the CI. The first issue was the following:
2025-06-03T15:11:51.2167598Z [ 96%] Building Fortran object fv3/atmos_cubed_sphere/CMakeFiles/fv3.dir/driver/UFS/fv_nggps_diag.F90.o
2025-06-03T15:11:51.4050342Z /home/runner/work/fv3atm/fv3atm/fv3atm/fv3/atmos_cubed_sphere/driver/UFS/fv_nggps_diag.F90:1165:8:
2025-06-03T15:11:51.4051124Z
2025-06-03T15:11:51.4051240Z 1165 | use esmf
2025-06-03T15:11:51.4052000Z | 1
2025-06-03T15:11:51.4052978Z Fatal Error: Cannot open module file ‘esmf.mod’ for reading at (1): No such file or directory
2025-06-03T15:11:51.4055508Z compilation terminated.
2025-06-03T15:11:51.4078729Z make[2]: *** [fv3/atmos_cubed_sphere/CMakeFiles/fv3.dir/build.make:650: fv3/atmos_cubed_sphere/CMakeFiles/fv3.dir/driver/UFS/fv_nggps_diag.F90.o] Error 1
2025-06-03T15:11:51.4079870Z make[2]: *** Waiting for unfinished jobs....
2025-06-03T15:11:51.6629706Z make[1]: *** [CMakeFiles/Makefile2:1458: fv3/atmos_cubed_sphere/CMakeFiles/fv3.dir/all] Error 2
2025-06-03T15:11:51.6632028Z make: *** [Makefile:146: all] Error 2
2025-06-03T15:11:51.6650138Z ##[error]Process completed with exit code 2.
which was resolved with line 14.
The second issue:
2025-06-03T16:44:49.8904603Z [100%] Built target ufsatm_fv3
2025-06-03T16:44:49.9029247Z [100%] Building Fortran object tests/CMakeFiles/test_post_nems_routines.dir/test_post_nems_routines.F90.o
2025-06-03T16:44:50.0182534Z [100%] Linking Fortran executable test_post_nems_routines
2025-06-03T16:44:50.1388910Z /usr/bin/ld: cannot find -lesmf: No such file or directory
2025-06-03T16:44:50.1550362Z collect2: error: ld returned 1 exit status
2025-06-03T16:44:50.1568824Z make[2]: *** [tests/CMakeFiles/test_post_nems_routines.dir/build.make:123: tests/test_post_nems_routines] Error 1
2025-06-03T16:44:50.1571012Z make[1]: *** [CMakeFiles/Makefile2:1424: tests/CMakeFiles/test_post_nems_routines.dir/all] Error 2
2025-06-03T16:44:50.1573892Z make: *** [Makefile:146: all] Error 2
2025-06-03T16:44:50.1594681Z ##[error]Process completed with exit code 2.
2025-06-03T16:44:50.1679520Z ##[group]Run actions/upload-artifact@v4
was only resolved when adding in Line 15.
I'm not as familiar with CMake and the ESMF libraries, so if there's something I'm missing or somewhere else I could be looking at, let me know.
|
Closing this PR since changes have been merged into #966. |
github-project-automation
Bot
moved this from In Progress
to Done
in Model Infrastructure development FY25Q3
3 participants
Description
The GitHub CI/CD workflow was broken after a recent PR. This PR will update the workflows to allow the check to be run again.
Two distinct issues were found so far:
Issue(s) addressed
Testing
These changes were tested using the GitHub CI/CD workflow.
The changes were also tested on Hercules in UFS to make sure that there weren't any unexpected changes.
Dependencies
None