Use hpc-stack for Actions CI

[MinsukJi-NOAA]

Description

Use hpc-stack (with upp support) in docker container used by CI.
Minor change in rt_utils.sh to fail regression test if comp_ncfile.py fails to load required modules.

Issue(s) addressed

Resolves #240 #277 #296

Testing

GitHub Actions tested for working properly: https://github.com/MinsukJi-NOAA/ufs-weather-model/actions/runs/379981603

[climbfuji]

@junwang-noaa @DusanJovic-NOAA @MinsukJi-NOAA We originally planned to include this in my PRs (#297 & Co). I though this would only be touching ci files, but given the change in rt_utils.sh I am reluctant to pull this into my PR. I already completed the regression testing on hera.intel, hera.gnu, orion.intel, and I am close to completing gaea.intel, too. Thoughts?

[junwang-noaa]

@climbfuji If possible, I do hope to get this fix in as soon as possible. Can we help on rerunning RT on hera/orion?

[MinsukJi-NOAA]

@climbfuji , Better solution is in the pipe to address #296. I will revert back the rt_utils.sh change for now.

[climbfuji]

@junwang-noaa wants to include the change. I'll pull your PR in and rerun on hera.intel, hera.gnu, orion.intel, gaea.intel.

[climbfuji]

Please convert your PR into "ready for review", I'll mention it in my PR. Thanks!

[junwang-noaa]

Minsuk, I found a bug in compile.sh:

mv ${BUILD_DIR}/ufs_model ${PATHTR}/tests/${BUILD_NAME}.exe
cp ${PATHTR}/modulefiles/${MACHINE_ID}/fv3 ${PATHTR}/tests/modules.${BUILD_NAME}

should be:
mv ${BUILD_DIR}/ufs_model ${PATHTR}/tests/${BUILD_NAME}.exe
if [[ "${MAKE_OPT}" == "DEBUG=Y" ]]; then
cp ${PATHTR}/modulefiles/${MACHINE_ID}/fv3_debug ${PATHTR}/tests/modules.${BUILD_NAME}
else
cp ${PATHTR}/modulefiles/${MACHINE_ID}/fv3 ${PATHTR}/tests/modules.${BUILD_NAME}
fi

Would you mind adding it in your PR? Thanks.

[climbfuji]

Note. This PR was pulled into #297. However, the change in rt_utils.sh to fix a problem with silently failing regression tests if comp_ncfile.py fails to load required modules (commit 2eaea2e) had to be reverted manually.

I suggest closing this PR and issues #240 and #277, but keep issue #296 open.

[MinsukJi-NOAA]

Note. This PR was pulled into #297. However, the change in rt_utils.sh to fix a problem with silently failing regression tests if comp_ncfile.py fails to load required modules (commit 2eaea2e) had to be reverted manually.

I suggest closing this PR and issues #240 and #277, but keep issue #296 open.

I agree.