port RDASApp to Gaea

README.md

@@ -6,6 +6,7 @@ Repository for JEDI-based data assimilation for the Rapid Refresh Forecast Syste
 -   NOAA RDHPCS Jet
 -   NOAA RDHPCS Orion
 -   NOAA RDHPCS Hercules
+-   NOAA RDHPCS Gaea (C5 and C6)
 
 ## [How to clone and build RDASAPP?](docs/build_and_test.md)
 ## [How to run RRFS CTest?](docs/build_and_test.md#3-rrfs-ctest)

build.sh

@@ -77,10 +77,10 @@ while getopts "p:c:m:j:hvfsx" opt; do
 done
 
 case ${BUILD_TARGET} in
-  hera | orion | hercules | jet)
+  hera | orion | hercules | jet | gaea )
     echo "Building RDASApp on $BUILD_TARGET"
     echo "  Build initiated `date`"
-    source $dir_root/ush/module-setup.sh
+    [[ "${BUILD_TARGET}" != *gaea* ]] && source $dir_root/ush/module-setup.sh
     module use $dir_root/modulefiles
     module load RDAS/$BUILD_TARGET.$COMPILER
     CMAKE_OPTS+=" -DMPIEXEC_EXECUTABLE=$MPIEXEC_EXEC -DMPIEXEC_NUMPROC_FLAG=$MPIEXEC_NPROC -DBUILD_GSIBEC=ON -DMACHINE_ID=$MACHINE_ID"
@@ -95,7 +95,7 @@ esac
 # Set default number of build jobs based on machine
 if [[ $BUILD_TARGET == 'orion' ]]; then # lower due to memory limit on login nodes
   BUILD_JOBS=${BUILD_JOBS:-4}
-else # hera, hercules, jet
+else # hera, hercules, jet, gaea
   BUILD_JOBS=${BUILD_JOBS:-6}
 fi
 

docs/build_and_test.md

@@ -1,5 +1,5 @@
 ## 1.Clone RDASApp
-If running on Orion/Hercules, you will need to run `module load git-lfs` before cloning. 
+If running on Orion/Hercules/Gaea, you will need to run `module load git-lfs` before cloning. 
 ```
 git clone --recurse-submodules https://github.com/NOAA-EMC/RDASApp.git
 ```
@@ -14,4 +14,7 @@ Run `./build.sh -h` to learn more about command line options supported by build.
 ```
     ush/run_rrfs_tests.sh $account
 ```
-Where `$account` is your valid slurm resource account (e.g., `fv3-cam`, `da-cpu`, `wrfruc`, `rtrr`, `nrtrr`, etc.). 
+Where `$account` is your valid slurm resource account (e.g., `fv3-cam`, `da-cpu`, `wrfruc`, `rtrr`, `nrtrr`, etc.).   
+- To run ctest mannualy without using the above bash script, follow these two steps first:  
+`source ush/load_rdas.sh`    
+`export SLURM_ACCOUNT=$account`

modulefiles/EVA/gaeaC5.lua

@@ -0,0 +1,19 @@
+help([[
+Load environment for running EVA.
+]])
+
+local pkgName    = myModuleName()
+local pkgVersion = myModuleVersion()
+local pkgNameVer = myModuleFullName()
+
+conflict(pkgName)
+
+prepend_path("MODULEPATH", '/gpfs/f5/gsl-glo/world-shared/gge/miniconda3/modulefiles')
+
+load("miniconda3/4.6.14")
+load("eva/1.0.0")
+
+whatis("Name: ".. pkgName)
+whatis("Version: ".. pkgVersion)
+whatis("Category: EVA")
+whatis("Description: Load all libraries needed for EVA")

modulefiles/EVA/gaeaC6.lua

@@ -0,0 +1,19 @@
+help([[
+Load environment for running EVA.
+]])
+
+local pkgName    = myModuleName()
+local pkgVersion = myModuleVersion()
+local pkgNameVer = myModuleFullName()
+
+conflict(pkgName)
+
+prepend_path("MODULEPATH", '/gpfs/f6/bil-fire10-oar/world-shared/gge/miniconda3/modulefiles')
+
+load("miniconda3/4.6.14")
+load("eva/1.0.0")
+
+whatis("Name: ".. pkgName)
+whatis("Version: ".. pkgVersion)
+whatis("Category: EVA")
+whatis("Description: Load all libraries needed for EVA")

modulefiles/RDAS/gaea.intel.lua

@@ -0,0 +1,82 @@
+help([[
+Load environment for running the RDAS application with Intel compilers and MPI.
+]])
+
+local pkgName    = myModuleName()
+local pkgVersion = myModuleVersion()
+local pkgNameVer = myModuleFullName()
+
+prepend_path("MODULEPATH", '/ncrc/proj/epic/spack-stack/c6/spack-stack-1.6.0/envs/unified-env/install/modulefiles/Core')
+
+-- below two lines get us access to the spack-stack modules
+load("stack-intel/2023.2.0")
+load("stack-cray-mpich/8.1.29")
+-- JCSDA has 'jedi-fv3-env/unified-dev', but we should load these manually as needed
+load("cmake/3.23.1")
+load("gettext/0.20.2")
+--load("libunistring/1.1")
+--load("libidn2/2.3.4")
+load("pcre2/10.42")
+load("curl/8.4.0")
+load("zlib/1.2.13")
+load("git/2.42.0")
+load("pkg-config/0.29.2")
+load("hdf5/1.14.0")
+load("parallel-netcdf/1.12.2")
+load("netcdf-c/4.9.2")
+load("nccmp/1.9.0.1")
+load("netcdf-fortran/4.6.1")
+load("nco/5.0.6")
+load("parallelio/2.5.10")
+load("wget/1.20.3")
+load("boost/1.83.0")
+load("bufr/12.0.1")
+load("git-lfs/2.11.0")
+load("ecbuild/3.7.2")
+load("openjpeg/2.3.1")
+load("eccodes/2.32.0")
+load("eigen/3.4.0")
+load("openblas/0.3.24")
+load("eckit/1.24.5")
+load("fftw/3.3.10")
+load("fckit/0.11.0")
+load("fiat/1.2.0")
+--load("ectrans/1.2.0")
+load("atlas/0.35.1")
+load("sp/2.5.0")
+load("gsl-lite/0.37.0")
+load("libjpeg/2.1.0")
+load("krb5/1.20.1")
+load("libtirpc/1.3.3")
+load("hdf/4.2.15")
+load("jedi-cmake/1.4.0")
+load("libpng/1.6.37")
+load("udunits/2.2.28")
+load("ncview/2.1.9")
+load("netcdf-cxx4/4.3.1")
+load("json/3.10.5")
+--load("crtm/v2.4_jedi")
+load("prod_util/2.1.1")
+load("fms/2023.04")
+
+load("py-jinja2/3.0.3")
+load("py-netcdf4/1.5.8")
+load("py-pybind11/2.11.0")
+load("py-pycodestyle/2.11.0")
+load("py-pyyaml/6.0")
+load("py-scipy/1.11.3")
+load("py-xarray/2023.7.0")
+
+setenv("CC","cc")
+setenv("FC","ftn")
+setenv("CXX","CC")
+
+local mpiexec = '/usr/bin/srun'
+local mpinproc = '-n'
+setenv('MPIEXEC_EXEC', mpiexec)
+setenv('MPIEXEC_NPROC', mpinproc)
+
+whatis("Name: ".. pkgName)
+whatis("Version: ".. pkgVersion)
+whatis("Category: RDASApp")
+whatis("Description: Load all libraries needed for RDASApp")

rrfs-test/scripts/templates/mpasjedi_expr/run_bump.sh

@@ -1,6 +1,7 @@
 #!/bin/sh
 #SBATCH --account=rtrr
-#SBATCH --qos=batch
+#SBATCH --qos=batch # use the normal queue on Gaea
+###SBATCH -M c6 # for Gaea
 ###SBATCH --partition=bigmem
 ###SBATCH --partition=kjet
 ###SBATCH --reservation=rrfsens
@@ -13,7 +14,9 @@
 
 RDASApp=@RDASApp@
 
-. /apps/lmod/lmod/init/sh
+if [[ -s /apps/lmod/lmod/init/sh ]]; then
+  . /apps/lmod/lmod/init/sh
+fi
 
 module purge
 source ${RDASApp}/ush/detect_machine.sh 

rrfs-test/scripts/templates/mpasjedi_expr/run_jedi.sh

@@ -1,6 +1,7 @@
 #!/bin/sh
 #SBATCH --account=rtrr
-#SBATCH --qos=batch
+#SBATCH --qos=batch # use the normal queue on Gaea
+###SBATCH -M c6 # for Gaea
 ###SBATCH --partition=kjet
 ###SBATCH --reservation=rrfsens
 #SBATCH --ntasks=120
@@ -16,7 +17,9 @@ inputfile=./rrfs_mpasjedi_2024052700_Ens3Dvar.yaml # FOR ENVAR
 #inputfile=./rrfs_mpasjedi_2024052700_letkf.yaml # FOR LETKF
 #inputfile=./rrfs_mpasjedi_2024052700_getkf.yaml # FOR GETKF
 
-. /apps/lmod/lmod/init/sh
+if [[ -s /apps/lmod/lmod/init/sh ]]; then
+  . /apps/lmod/lmod/init/sh
+fi
 
 module purge
 source ${RDASApp}/ush/detect_machine.sh 

ush/detect_machine.sh

@@ -21,10 +21,10 @@ case $(hostname -f) in
   dlogin0[1-9].dogwood.wcoss2.ncep.noaa.gov) MACHINE_ID=wcoss2 ;; ### dogwood01-9
   dlogin10.dogwood.wcoss2.ncep.noaa.gov)     MACHINE_ID=wcoss2 ;; ### dogwood10
 
-  gaea9)               MACHINE_ID=gaea ;; ### gaea9
-  gaea1[0-6])          MACHINE_ID=gaea ;; ### gaea10-16
-  gaea9.ncrc.gov)      MACHINE_ID=gaea ;; ### gaea9
-  gaea1[0-6].ncrc.gov) MACHINE_ID=gaea ;; ### gaea10-16
+  gaea|gaea5[1-8])     MACHINE_ID=gaea ;;
+  gaea6[1-8])          MACHINE_ID=gaea ;;
+  gaea.ncrc.gov|gaea5[1-8].ncrc.gov) MACHINE_ID=gaea ;;
+  gaea6[1-8].ncrc.gov)               MACHINE_ID=gaea ;;
 
   hfe0[1-9]) MACHINE_ID=hera ;; ### hera01-09
   hfe1[0-2]) MACHINE_ID=hera ;; ### hera10-12
@@ -83,8 +83,10 @@ elif [[ -d /work ]]; then
   else
     MACHINE_ID=orion
   fi
-elif [[ -d /gpfs && -d /ncrc ]]; then
-  # We are on GAEA.
+elif [[ -d /gpfs/f5 && -d /ncrc ]]; then
+  # We are on GAEA
+  MACHINE_ID=gaea
+elif [[ -d /gpfs/f6 && -d /ncrc ]]; then
   MACHINE_ID=gaea
 elif [[ -d /data/prod ]]; then
   # We are on SSEC's S4

ush/init.sh

@@ -14,6 +14,15 @@ case ${MACHINE_ID} in
   orion|hercules)
     RDAS_DATA=/work/noaa/zrtrr/RDAS_DATA
     ;;
+  gaea)
+    if [[ -d /gpfs/f5 ]]; then
+      RDAS_DATA=/gpfs/f5/gsl-glo/world-shared/role.rrfsfix/RDAS_DATA
+    elif [[ -d /gpfs/f6 ]]; then
+      RDAS_DATA=/gpfs/f6/bil-fire10-oar/world-shared/role.rrfsfix/RDAS_DATA
+    else
+      echo "unsupported gaea cluster: ${MACHINE_ID}"
+    fi
+    ;;
   *)
     echo "platform not supported: ${MACHINE_ID}"
     ;;

ush/load_eva.sh

@@ -12,5 +12,15 @@ source ${ushdir}/detect_machine.sh
 
 module purge
 module use ${ushdir}/../modulefiles
-module load EVA/${MACHINE_ID}
+if [[ "${MACHINE_ID}" == "gaea" ]]; then
+  if [[ -d /gpfs/f5 ]]; then
+    module load EVA/${MACHINE_ID}C5
+  elif [[ -d /gpfs/f6 ]]; then
+    module load EVA/${MACHINE_ID}C6
+  else
+    echo "not supported gaea cluster: ${MACHINE_ID}"
+  fi
+else
+  module load EVA/${MACHINE_ID}
+fi
 module list