| 1 | = Installation on SMCE head node (external packages, binaries only) |
| 2 | - Added the following to `~/.bashrc`, |
| 3 | {{{ |
| 4 | # Spack Modules |
| 5 | source /shared/spack/share/spack/setup-env.sh |
| 6 | module load intelmpi |
| 7 | #module load python-3.8.12-gcc-9.4.0-uyaszn4 |
| 8 | module load subversion-1.14.0-gcc-11.1.0-evgneyf |
| 9 | |
| 10 | ################### |
| 11 | ### Environment ### |
| 12 | ################### |
| 13 | export CC=/opt/intel/mpi/2021.4.0/bin/mpicc |
| 14 | export GCC=/opt/intel/mpi/2021.4.0/bin/mpicxx |
| 15 | export FC=/opt/intel/mpi/2021.4.0/bin/mpifc |
| 16 | |
| 17 | export MPI_HOME=/opt/intel/mpi/2021.4.0 |
| 18 | |
| 19 | ###################### |
| 20 | ### Path Variables ### |
| 21 | ###################### |
| 22 | |
| 23 | # ISSM |
| 24 | export ISSM_DIR="/efs/issm/binaries/repos/trunk-jpl-working" |
| 25 | export ISSM_EXT_DIR="/efs/issm/binaries/ext" |
| 26 | }}} |
| 27 | |
| 28 | - Checked out a copy of the ISSM development repository with, |
| 29 | {{{ |
| 30 | mkdir -p /efs/issm/binaries/repos |
| 31 | svn co https://issm.ess.uci.edu/svn/issm/issm/trunk-jpl $ISSM_DIR |
| 32 | }}} |
| 33 | |
| 34 | - Copied `$ISSM_DIR/jenkins/ross-debian_linux-full` to `$ISSM_DIR/jenkins/eis-smce-binaries` |
| 35 | - Modified `$ISSM_DIR/jenkins/eis-smce-binaries` with, |
| 36 | {{{ |
| 37 | #--------------------# |
| 38 | # ISSM Configuration # |
| 39 | #--------------------# |
| 40 | |
| 41 | ISSM_CONFIG='\ |
| 42 | --prefix="${ISSM_DIR}" \ |
| 43 | --disable-static \ |
| 44 | --with-wrappers=no \ |
| 45 | --enable-development \ |
| 46 | --enable-debugging \ |
| 47 | --with-numthreads=48 \ |
| 48 | --with-fortran-lib="-L/usr/lib/gcc/x86_64-linux-gnu/9 -lgfortran" \ |
| 49 | --with-mpi-include="/opt/intel/mpi/2021.4.0/include" \ |
| 50 | --with-mpi-libflags="-L/opt/intel/mpi/2021.4.0/lib -lmpi -lmpicxx -lmpifort" \ |
| 51 | --with-blas-lapack-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 52 | --with-metis-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 53 | --with-parmetis-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 54 | --with-scalapack-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 55 | --with-mumps-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 56 | --with-hdf5-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 57 | --with-petsc-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 58 | --with-gsl-dir="${ISSM_EXT_DIR}/gsl/install" \ |
| 59 | --with-boost-dir="${ISSM_EXT_DIR}/boost/install" \ |
| 60 | --with-dakota-dir="${ISSM_EXT_DIR}/dakota/install" \ |
| 61 | --with-proj-dir="${ISSM_EXT_DIR}/proj/install" \ |
| 62 | --with-triangle-dir="${ISSM_EXT_DIR}/triangle/install" \ |
| 63 | --with-chaco-dir="${ISSM_EXT_DIR}/chaco/install" \ |
| 64 | --with-m1qn3-dir="${ISSM_EXT_DIR}/m1qn3/install" \ |
| 65 | --with-semic-dir="${ISSM_EXT_DIR}/semic/install" |
| 66 | ' |
| 67 | |
| 68 | #-------------------# |
| 69 | # External Packages # |
| 70 | #-------------------# |
| 71 | |
| 72 | EXTERNALPACKAGES=" |
| 73 | autotools install-linux.sh |
| 74 | cmake install.sh |
| 75 | petsc install-3.12-linux.sh |
| 76 | gsl install.sh |
| 77 | boost install-1.7-linux.sh |
| 78 | dakota install-6.2-linux.sh |
| 79 | curl install-7-linux.sh |
| 80 | netcdf install-4.7-parallel.sh |
| 81 | sqlite install.sh |
| 82 | proj install-6.sh |
| 83 | gdal install-3.sh |
| 84 | gshhg install.sh |
| 85 | gmt install-6-linux.sh |
| 86 | gmsh install-4-linux.sh |
| 87 | triangle install-linux.sh |
| 88 | chaco install.sh |
| 89 | m1qn3 install.sh |
| 90 | semic install.sh |
| 91 | " |
| 92 | }}} |
| 93 | - Committed `$ISSM_DIR/jenkins/eis-smce-head-node-binaries` to repo |
| 94 | - Installed each of the external packages in `$ISSM_DIR/jenkins/eis-daskhub-python-modules`, after changing each of the installation paths with, |
| 95 | {{{ |
| 96 | PREFIX=${ISSM_EXT_DIR}/<pkg>/install |
| 97 | }}} |
| 98 | and making the following modifications, |
| 99 | - Modified `$ISSM_DIR/externalpackages/petsc/install-3.12-linux.sh` by, |
| 100 | - removing, |
| 101 | {{{ |
| 102 | --download-mpich=1 |
| 103 | }}} |
| 104 | - adding, |
| 105 | {{{ |
| 106 | --with-mpi-dir=/opt/intel/mpi/2021.4.0 |
| 107 | --known-mpi-shared-libraries=1 |
| 108 | }}} |
| 109 | - `$ISSM_DIR/externalpackages/dakota/install-6.2-linux.sh` requires Python 2 to build, so either (1) cannot load Spack module for Python 3, or, (2) have to figure out how to override this in the install script (perhaps with `alias python=/usr/bin/python`) |
| 110 | - Modified `$ISSM_DIR/externalpackages/gdal/install-3.sh` with, |
| 111 | {{{ |
| 112 | --with-curl="${CURL_ROOT}/bin/curl-config" |
| 113 | }}} |
| 114 | |
| 115 | = Installation on MATLAB node (MATLAB modules) |
| 116 | TO BE ADDED AFTER FURTHER INSTALLATION AND DEBUGGING |
| 117 | |
| 118 | = Installation on Daskhub (Python modules) |
| 119 | == NOTE: Binaries are also compiled as we do not have a good way of disabling them. We could ignore them, remove them, or let users know that they're available for smaller test runs. |
| 120 | - Added the following to ~/.bash_profile, |
| 121 | {{{ |
| 122 | # File system modifications that are not currently part of image |
| 123 | # |
| 124 | |
| 125 | # Create a link to gl.h so that the CMake configuration can find it. |
| 126 | # Note that we need to add it here because it does not persist. |
| 127 | if [ ! -f /srv/conda/envs/notebook/include/GL/gl.h ]; then |
| 128 | ln -s /srv/conda/envs/notebook/include/gstreamer-1.0/gst/gl/gl.h /srv/conda/envs/notebook/include/GL/gl.h |
| 129 | fi |
| 130 | |
| 131 | # Set permissions on private key |
| 132 | chmod 400 ~/.ssh/eis-nasa-smce-head-node |
| 133 | |
| 134 | |
| 135 | ## Environment |
| 136 | # |
| 137 | export OPENSSL_ROOT_DIR="/srv/conda/envs/notebook" |
| 138 | |
| 139 | |
| 140 | ## Path Variables |
| 141 | # |
| 142 | |
| 143 | # ISSM |
| 144 | export ISSM_DIR="/efs/issm/python-modules/repos/trunk-jpl-working" |
| 145 | export ISSM_EXT_DIR="/efs/issm/python-modules/ext" |
| 146 | |
| 147 | |
| 148 | ## Aliases |
| 149 | # |
| 150 | alias issm-python-dev-env="source "\${ISSM_DIR}"/etc/environment.sh; export PYTHONPATH="\${ISSM_DIR}"/src/m/dev; export PYTHONSTARTUP="\${PYTHONPATH}"/devpath.py; export PYTHONUNBUFFERED=1; cd "\${ISSM_DIR}"/test/NightlyRun; LD_PRELOAD=/srv/conda/envs/notebook/lib/libpython3.9.so.1.0:"\${ISSM_EXT_DIR}"/petsc/install/lib/libmpifort.so.12 python" # Then run ./runme.py <options> |
| 151 | }}} |
| 152 | |
| 153 | - Checked out a copy of the ISSM development repository with, |
| 154 | {{{ |
| 155 | mkdir -p /efs/issm/python-modules/repos |
| 156 | svn co https://issm.ess.uci.edu/svn/issm/issm/trunk-jpl $ISSM_DIR |
| 157 | }}} |
| 158 | |
| 159 | - Copied `$ISSM_DIR/jenkins/ross-debian_linux-full` to `$ISSM_DIR/jenkins/eis-daskhub-python-modules` |
| 160 | - Modified `$ISSM_DIR/jenkins/eis-daskhub-python-modules` with, |
| 161 | {{{ |
| 162 | #--------------------# |
| 163 | # ISSM Configuration # |
| 164 | #--------------------# |
| 165 | |
| 166 | ISSM_CONFIG='\ |
| 167 | --prefix="${ISSM_DIR}" \ |
| 168 | --disable-static \ |
| 169 | --enable-development \ |
| 170 | --enable-debugging \ |
| 171 | --with-numthreads=4 \ |
| 172 | --with-python-dir="/srv/conda/envs/notebook" \ |
| 173 | --with-python-version="3.9" \ |
| 174 | --with-python-numpy-dir="/srv/conda/envs/notebook/lib/python3.9/site-packages/numpy/core/include/numpy" \ |
| 175 | --with-fortran-lib="-L/usr/lib/x86_64-linux-gnu -lgfortran" \ |
| 176 | --with-mpi-include="${ISSM_EXT_DIR}/petsc/install/include" \ |
| 177 | --with-mpi-libflags="-L${ISSM_EXT_DIR}/petsc/install/lib -lmpi -lmpicxx -lmpifort" \ |
| 178 | --with-blas-lapack-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 179 | --with-metis-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 180 | --with-parmetis-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 181 | --with-scalapack-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 182 | --with-mumps-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 183 | --with-hdf5-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 184 | --with-petsc-dir="${ISSM_EXT_DIR}/petsc/install" \ |
| 185 | --with-gsl-dir="${ISSM_EXT_DIR}/gsl/install" \ |
| 186 | --with-boost-dir="${ISSM_EXT_DIR}/boost/install" \ |
| 187 | --with-dakota-dir="${ISSM_EXT_DIR}/dakota/install" \ |
| 188 | --with-proj-dir="${ISSM_EXT_DIR}/proj/install" \ |
| 189 | --with-triangle-dir="${ISSM_EXT_DIR}/triangle/install" \ |
| 190 | --with-chaco-dir="${ISSM_EXT_DIR}/chaco/install" \ |
| 191 | --with-m1qn3-dir="${ISSM_EXT_DIR}/m1qn3/install" \ |
| 192 | --with-semic-dir="${ISSM_EXT_DIR}/semic/install" \ |
| 193 | ' |
| 194 | |
| 195 | #-------------------# |
| 196 | # External Packages # |
| 197 | #-------------------# |
| 198 | |
| 199 | EXTERNALPACKAGES=" |
| 200 | autotools install-linux.sh |
| 201 | cmake install.sh |
| 202 | petsc install-3.16-linux.sh |
| 203 | gsl install.sh |
| 204 | boost install-1.7-linux.sh |
| 205 | dakota install-6.2-linux.sh |
| 206 | curl install-7-linux.sh |
| 207 | netcdf install-4.7-parallel.sh |
| 208 | sqlite install.sh |
| 209 | proj install-6.sh |
| 210 | gdal install-3-python.sh |
| 211 | gshhg install.sh |
| 212 | gmt install-6-linux.sh |
| 213 | gmsh install-4-linux.sh |
| 214 | triangle install-linux.sh |
| 215 | chaco install.sh |
| 216 | m1qn3 install.sh |
| 217 | semic install.sh |
| 218 | " |
| 219 | }}} |
| 220 | |
| 221 | - Committed `$ISSM_DIR/jenkins/eis-daskhub-python-modules` to repo |
| 222 | - Before installing any of the external packages, needed to disable the Conda environment with `conda deactivate` |
| 223 | - Installed each of the external packages in `$ISSM_DIR/jenkins/eis-daskhub-python-modules`, after changing each of the installation paths with, |
| 224 | {{{ |
| 225 | PREFIX=${ISSM_EXT_DIR}/<pkg>/install |
| 226 | }}} |
| 227 | and making the following modifications |
| 228 | - Needed to remove, |
| 229 | {{{ |
| 230 | --FFLAGS="-fallow-argument-mismatch" |
| 231 | }}} |
| 232 | from `$ISSM_DIR/externalpackages/petsc/install-3.16-petsc.sh` |
| 233 | - Needed to modify `$ISSM_DIR/externalpackages/gdal/install-3-python.sh` with, |
| 234 | {{{ |
| 235 | export CPATH="${CPATH}:/srv/conda/envs/notebook/include" |
| 236 | export LDFLAGS="-L${NETCDF_ROOT}/lib -lnetcdf" |
| 237 | export LIBRARY_PATH="${LIBRARY_PATH}:/srv/conda/envs/notebook/lib" |
| 238 | export LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/srv/conda/envs/notebook/lib" |
| 239 | }}} |
| 240 | and, |
| 241 | {{{ |
| 242 | --with-sqlite3="${SQLITE_ROOT}" \ |
| 243 | --with-libjson-c="/srv/conda/envs/notebook" |
| 244 | }}} |
| 245 | |
| 246 | - Needed to modify `$ISSM_DIR/externalpackages/gmt/install-6-linux.sh` with, |
| 247 | {{{ |
| 248 | export CC=mpicc |
| 249 | export CPATH="${CPATH}:/srv/conda/envs/notebook/include" |
| 250 | export LIBRARY_PATH="${LIBRARY_PATH}:/srv/conda/envs/notebook/lib" |
| 251 | export LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/srv/conda/envs/notebook/lib" |
| 252 | }}} |
| 253 | |
| 254 | - With Conda environment still disabled, ran, |
| 255 | {{{ |
| 256 | alias envsubst="/srv/conda/envs/notebook/bin/envsubst" |
| 257 | jenkins/jenkins.sh jenkins/eis-daskhub-python-modules |
| 258 | }}} |
| 259 | |
| 260 | - Ran, |
| 261 | {{{ |
| 262 | mkdir ~/.ssh |
| 263 | ssh-keygen -f ~/.ssh/eis-nasa-smce |
| 264 | }}} |
| 265 | then gave contents of public key to SMCE team to add to SMCE head node |
| 266 | - Ran, |
| 267 | {{{ |
| 268 | chmod 400 ~/.ssh/eis-nasa-smce |
| 269 | }}} |
| 270 | |
| 271 | - Created `$ISSM_DIR/src/m/classes/clusters/eis_nasa_smce.py` with contents, |
| 272 | {{{ |
| 273 | import os |
| 274 | import shutil |
| 275 | import subprocess |
| 276 | |
| 277 | try: |
| 278 | from eis_nasa_smce_settings import eis_nasa_smce_settings |
| 279 | except ImportError: |
| 280 | print('You need eis_nasa_smce_settings.py to proceed, check presence and sys.path') |
| 281 | from fielddisplay import fielddisplay |
| 282 | from helpers import * |
| 283 | from IssmConfig import IssmConfig |
| 284 | from issmssh import issmssh |
| 285 | from MatlabFuncs import * |
| 286 | from pairoptions import pairoptions |
| 287 | |
| 288 | class eis_nasa_smce(object): |
| 289 | """EIS_NASA_SMCE cluster class definition |
| 290 | |
| 291 | Usage: |
| 292 | cluster = eis_nasa_smce() |
| 293 | cluster = eis_nasa_smce('np', 3) |
| 294 | cluster = eis_nasa_scme('np', 3, 'login', 'username') |
| 295 | """ |
| 296 | |
| 297 | def __init__(self, *args): # {{{ |
| 298 | self.name = '52.10.233.96' |
| 299 | self.login = 'jdquinn1' |
| 300 | self.idfile = '~/.ssh/eis-nasa-smce' |
| 301 | self.modules = ['intelmpi'] |
| 302 | self.numnodes = 4 |
| 303 | self.cpuspernode = 1 |
| 304 | self.port = 0 |
| 305 | self.time = 12 * 60 * 60 |
| 306 | self.processor = 'skylake' |
| 307 | self.partition = 'sealevel-c5xl-spot' |
| 308 | self.srcpath = '/efs/issm-new/binaries/repos/trunk-jpl-working' |
| 309 | self.extpkgpath = '/efs/issm-new/binaries/ext' |
| 310 | self.codepath = '/efs/issm-new/binaries/repos/trunk-jpl-working/bin' |
| 311 | self.executionpath = '~/issm-exec' |
| 312 | self.interactive = 0 |
| 313 | self.numstreams = 1 |
| 314 | self.hyperthreading = 0 |
| 315 | self.email = '' |
| 316 | |
| 317 | # Use provided options to change fields |
| 318 | options = pairoptions(*args) |
| 319 | |
| 320 | # Initialize cluster using user settings if provided |
| 321 | try: |
| 322 | self = eis_nasa_smce_settings(self) |
| 323 | except NameError: |
| 324 | print('eis_nasa_smce_settings.py not found, using default settings') |
| 325 | |
| 326 | # OK get other fields |
| 327 | self = options.AssignObjectFields(self) |
| 328 | # }}} |
| 329 | |
| 330 | def __repr__(self): # {{{ |
| 331 | # Display the object |
| 332 | s = 'class eis_nasa_smce object\n' |
| 333 | s += ' name: {}\n'.format(self.name) |
| 334 | s += ' login: {}\n'.format(self.login) |
| 335 | s += ' idfile: {}\n'.format(self.idfile) |
| 336 | s += ' modules: {}\n'.format(strjoin(self.modules, ', ')) |
| 337 | s += ' numnodes: {}\n'.format(self.numnodes) |
| 338 | s += ' cpuspernode: {}\n'.format(self.cpuspernode) |
| 339 | s += ' np: {}\n'.format(self.nprocs()) |
| 340 | s += ' port: {}\n'.format(self.port) |
| 341 | s += ' time: {}\n'.format(self.time) |
| 342 | s += ' processor: {}\n'.format(self.processor) |
| 343 | s += ' partition: {}\n'.format(self.partition) |
| 344 | s += ' srcpath: {}\n'.format(self.srcpath) |
| 345 | s += ' extpkgpath: {}\n'.format(self.extpkgpath) |
| 346 | s += ' codepath: {}\n'.format(self.codepath) |
| 347 | s += ' executionpath: {}\n'.format(self.executionpath) |
| 348 | s += ' interactive: {}\n'.format(self.interactive) |
| 349 | s += ' numstreams: {}\n'.format(self.numstreams) |
| 350 | s += ' hyperthreading: {}\n'.format(self.hyperthreading) |
| 351 | return s |
| 352 | # }}} |
| 353 | |
| 354 | def nprocs(self): # {{{ |
| 355 | return self.numnodes * self.cpuspernode |
| 356 | # }}} |
| 357 | |
| 358 | def checkconsistency(self, md, solution, analyses): # {{{ |
| 359 | # Now, check cluster.cpuspernode according to processor type |
| 360 | if self.processor == 'skylake': |
| 361 | if self.cpuspernode > 14 or self.cpuspernode < 1: |
| 362 | md = md.checkmessage('cpuspernode should be between 1 and 14 for \'skyw\' processors in hyperthreading mode') |
| 363 | else: |
| 364 | md = md.checkmessage('unknown processor type, should be \'skylake\'') |
| 365 | |
| 366 | # Miscellaneous |
| 367 | if not self.login: |
| 368 | md = md.checkmessage('login empty') |
| 369 | if self.port: |
| 370 | md = md.checkmessage('port must be set to 0 as we do not have an SSH tunnel') |
| 371 | if not self.codepath: |
| 372 | md = md.checkmessage('codepath empty') |
| 373 | if not self.executionpath: |
| 374 | md = md.checkmessage('executionpath empty') |
| 375 | |
| 376 | return self |
| 377 | # }}} |
| 378 | |
| 379 | def BuildQueueScript(self, dirname, modelname, solution, io_gather, isvalgrind, isgprof, isdakota, isoceancoupling): # {{{ |
| 380 | if isgprof: |
| 381 | print('gprof not supported by cluster, ignoring...') |
| 382 | |
| 383 | issmexec = 'issm.exe' |
| 384 | mpiexec = 'mpiexec' # Set to alternative mpiexec if desired |
| 385 | |
| 386 | if isdakota: |
| 387 | version = IssmConfig('_DAKOTA_VERSION_')[0:2] |
| 388 | version = float(str(version[0])) |
| 389 | if version >= 6: |
| 390 | issmexec = 'issm_dakota.exe' |
| 391 | if isoceancoupling: |
| 392 | issmexec = 'issm_ocean.exe' |
| 393 | |
| 394 | # Write queuing script |
| 395 | fid = open(modelname + '.queue', 'w') |
| 396 | |
| 397 | fid.write('#!/bin/bash\n') |
| 398 | fid.write('#SBATCH --partition={} \n'.format(self.partition)) |
| 399 | fid.write('#SBATCH -J {} \n'.format(modelname)) |
| 400 | fid.write('#SBATCH -o {}.outlog \n'.format(modelname)) |
| 401 | fid.write('#SBATCH -e {}.errlog \n'.format(modelname)) |
| 402 | fid.write('#SBATCH --nodes={} \n'.format(self.numnodes)) |
| 403 | fid.write('#SBATCH --ntasks-per-node={} \n'.format(self.cpuspernode)) |
| 404 | fid.write('#SBATCH --cpus-per-task={} \n'.format(self.numstreams)) |
| 405 | fid.write('#SBATCH -t {:02d}:{:02d}:00 \n'.format(int(floor(self.time / 3600)), int(floor(self.time % 3600) / 60))) |
| 406 | if (self.email.find('@')>-1): |
| 407 | fid.write('#SBATCH --mail-user={} \n'.format(self.email)) |
| 408 | fid.write('#SBATCH --mail-type=BEGIN,END,FAIL \n\n') |
| 409 | fid.write('source /etc/profile\n') |
| 410 | fid.write('source /shared/spack/share/spack/setup-env.sh\n') |
| 411 | for i in range(len(self.modules)): |
| 412 | fid.write('module load {} &> /dev/null\n'.format(self.modules[i])) |
| 413 | fid.write('export MPI_GROUP_MAX=64\n\n') |
| 414 | fid.write('export MPI_UNBUFFERED_STDIO=true\n\n') |
| 415 | fid.write('export PATH="$PATH:/opt/slurm/bin"\n') |
| 416 | fid.write('export PATH="$PATH:."\n\n') |
| 417 | fid.write('export ISSM_DIR="{}"\n'.format(self.srcpath)) |
| 418 | fid.write('export ISSM_EXT_DIR="{}"\n'.format(self.extpkgpath)) |
| 419 | fid.write('source $ISSM_DIR/etc/environment.sh\n') |
| 420 | fid.write('cd {}/{}/\n\n'.format(self.executionpath, dirname)) |
| 421 | fid.write('{} -n {} {}/{} {} {}/{} {}\n'.format(mpiexec, self.nprocs(), self.codepath, issmexec, solution, self.executionpath, dirname, modelname)) |
| 422 | |
| 423 | if not io_gather: # concatenate the output files |
| 424 | fid.write('cat {}.outbin.* > {}.outbin'.format(modelname, modelname)) |
| 425 | fid.close() |
| 426 | |
| 427 | # In interactive mode, create a run file, and errlog and outlog file |
| 428 | if self.interactive: |
| 429 | fid = open(modelname + '.run', 'w') |
| 430 | if not isvalgrind: |
| 431 | fid.write('{} -np {} {}/{} {} {}/{} {}\n'.format(mpiexec, self.nprocs(), self.codepath, issmexec, solution, self.executionpath, dirname, modelname)) |
| 432 | else: |
| 433 | fid.write('{} -np {} valgrind --leak-check=full {}/{} {} {}/{} {}\n'.format(mpiexec, self.nprocs(), self.codepath, issmexec, solution, self.executionpath, dirname, modelname)) |
| 434 | if not io_gather: # concatenate the output files |
| 435 | fid.write('cat {}.outbin.* > {}.outbin'.format(modelname, modelname)) |
| 436 | fid.close() |
| 437 | fid = open(modelname + '.errlog', 'w') # TODO: Change this to system call (touch <file>)? |
| 438 | fid.close() |
| 439 | fid = open(modelname + '.outlog', 'w') # TODO: Change this to system call (touch <file>)? |
| 440 | fid.close() |
| 441 | # }}} |
| 442 | |
| 443 | def UploadQueueJob(self, modelname, dirname, filelist): # {{{ |
| 444 | # Compress the files into one zip |
| 445 | compressstring = 'tar -zcf {}.tar.gz'.format(dirname) |
| 446 | for file in filelist: |
| 447 | compressstring += ' {}'.format(file) |
| 448 | if self.interactive: |
| 449 | compressstring += ' {}.run {}.errlog {}.outlog'.format(modelname, modelname, modelname) |
| 450 | subprocess.call(compressstring, shell=True) |
| 451 | |
| 452 | print('uploading input file and queueing script') |
| 453 | if self.interactive: |
| 454 | directory = '{}/Interactive{}'.format(self.executionpath, self.interactive) |
| 455 | else: |
| 456 | directory = self.executionpath |
| 457 | |
| 458 | # NOTE: Replacement for issmscpout(self.name, directory, self.login, self.port, ['{}.tar.gz'.format(dirname)]) |
| 459 | copystring = 'cp {}.tar.gz /efs/issm/tmp'.format(dirname) |
| 460 | subprocess.call(copystring, shell=True) |
| 461 | # }}} |
| 462 | |
| 463 | def LaunchQueueJob(self, modelname, dirname, filelist, restart, batch): # {{{ |
| 464 | if self.interactive: |
| 465 | if not isempty(restart): |
| 466 | launchcommand = 'cd {}/Interactive{}'.format(self.executionpath, self.interactive) |
| 467 | else: |
| 468 | launchcommand = 'cd {}/Interactive{} && tar -zxf {}.tar.gz'.format(self.executionpath, self.interactive, dirname) |
| 469 | else: |
| 470 | if not isempty(restart): |
| 471 | launchcommand = 'cd {} && cd {} && sbatch {}.queue'.format(self.executionpath, dirname, modelname) |
| 472 | else: |
| 473 | launchcommand = 'cd {} && rm -rf {} && mkdir {} && cd {} && cp /efs/issm/tmp/{}.tar.gz . && tar -zxf {}.tar.gz && /opt/slurm/bin/sbatch {}.queue'.format(self.executionpath, dirname, dirname, dirname, dirname, dirname, modelname) |
| 474 | |
| 475 | print('launching solution sequence on remote cluster') |
| 476 | |
| 477 | # NOTE: Replacement for issmssh(self.name, self.login, self.port, launchcommand) |
| 478 | subprocess.call('ssh -l {} -i {} {} "{}"'.format(self.login, self.idfile, self.name, launchcommand), shell=True) |
| 479 | # }}} |
| 480 | |
| 481 | def Download(self, dirname, filelist): # {{{ |
| 482 | # Copy files from cluster to current directory |
| 483 | |
| 484 | # NOTE: Replacement for issmscpin(self.name, self.login, self.port, directory, filelist) |
| 485 | directory = '{}/{}/'.format(self.executionpath, dirname) |
| 486 | fileliststr = '{' + ','.join([str(x) for x in filelist]) + '}' |
| 487 | downloadcommand = 'scp -i {} {}@{}:{} {}/.'.format(self.idfile, self.login, self.name, os.path.join(directory, fileliststr), os.getcwd()) |
| 488 | subprocess.call(downloadcommand, shell=True) |
| 489 | # }}} |
| 490 | }}} |
| 491 | |
| 492 | - Committed `$ISSM_DIR/src/m/classes/clusters/eis_nasa_smce_head_node.py` to repo |
| 493 | |
| 494 | NOTE: |
| 495 | - It is important that the directory indicated by `self.executionpath` exists on SMCE head node before launch |
| 496 | - Either in `$ISSM_DIR/src/m/classes/clusters/eis_nasa_smce_head_node.py` or runme |
| 497 | - `cluster.login` should be changed to your username |
| 498 | - `cluster.executionpath` should be changed to desired execution path on SMCE head node |
| 499 | - `cluster.partition` should be changed to desired AWS ECS instance type (discuss with SMCE team before changing) |
| 500 | |
| 501 | - Created `$ISSM_DIR/test/NightlyRun/test101eisnasasmce.py` with contents, |
| 502 | {{{ |
| 503 | #Test Name: SquareShelfConstrainedStressSSA2d |
| 504 | from model import * |
| 505 | from socket import gethostname |
| 506 | from triangle import triangle |
| 507 | from setmask import setmask |
| 508 | from parameterize import parameterize |
| 509 | from setflowequation import setflowequation |
| 510 | from solve import solve |
| 511 | from massfluxatgate import massfluxatgate |
| 512 | from generic import generic |
| 513 | from eis_nasa_smce import eis_nasa_smce |
| 514 | |
| 515 | md = triangle(model(), '../Exp/Square.exp', 50000) |
| 516 | md = setmask(md, 'all', '') |
| 517 | md = parameterize(md, '../Par/SquareShelfConstrained.py') |
| 518 | md = setflowequation(md, 'SSA', 'all') |
| 519 | md.cluster = generic('name', gethostname(), 'np', 2) |
| 520 | |
| 521 | if True: |
| 522 | cluster = eis_nasa_smce() |
| 523 | cluster.partition = 'sealevel-c5xl-spot' |
| 524 | md.cluster = cluster |
| 525 | |
| 526 | #outputs |
| 527 | md.stressbalance.requested_outputs = ['default', 'DeviatoricStressxx', 'DeviatoricStressyy', 'DeviatoricStressxy', 'MassFlux1', 'MassFlux2', 'MassFlux3', 'MassFlux4', 'MassFlux5', 'MassFlux6'] |
| 528 | md.outputdefinition.definitions = [massfluxatgate('name', 'MassFlux1', 'profilename', '../Exp/MassFlux1.exp', 'definitionstring', 'Outputdefinition1'), |
| 529 | massfluxatgate('name', 'MassFlux2', 'profilename', '../Exp/MassFlux2.exp', 'definitionstring', 'Outputdefinition2'), |
| 530 | massfluxatgate('name', 'MassFlux3', 'profilename', '../Exp/MassFlux3.exp', 'definitionstring', 'Outputdefinition3'), |
| 531 | massfluxatgate('name', 'MassFlux4', 'profilename', '../Exp/MassFlux4.exp', 'definitionstring', 'Outputdefinition4'), |
| 532 | massfluxatgate('name', 'MassFlux5', 'profilename', '../Exp/MassFlux5.exp', 'definitionstring', 'Outputdefinition5'), |
| 533 | massfluxatgate('name', 'MassFlux6', 'profilename', '../Exp/MassFlux6.exp', 'definitionstring', 'Outputdefinition6')] |
| 534 | |
| 535 | md = solve(md, 'Stressbalance') |
| 536 | |
| 537 | #Fields and tolerances to track changes |
| 538 | field_names = ['Vx', 'Vy', 'Vel', 'Pressure', |
| 539 | 'DeviatoricStressxx', 'DeviatoricStressyy', 'DeviatoricStressxy', |
| 540 | 'MassFlux1', 'MassFlux2', 'MassFlux3', 'MassFlux4', 'MassFlux5', 'MassFlux6'] |
| 541 | field_tolerances = [3e-13, 1e-13, 1e-13, 1e-13, |
| 542 | 2e-13, 1e-13, 2e-13, |
| 543 | 1e-13, 1e-13, 1e-13, |
| 544 | 1e-13, 1e-13, 1e-13] |
| 545 | field_values = [md.results.StressbalanceSolution.Vx, |
| 546 | md.results.StressbalanceSolution.Vy, |
| 547 | md.results.StressbalanceSolution.Vel, |
| 548 | md.results.StressbalanceSolution.Pressure, |
| 549 | md.results.StressbalanceSolution.DeviatoricStressxx, |
| 550 | md.results.StressbalanceSolution.DeviatoricStressyy, |
| 551 | md.results.StressbalanceSolution.DeviatoricStressxy, |
| 552 | md.results.StressbalanceSolution.MassFlux1, |
| 553 | md.results.StressbalanceSolution.MassFlux2, |
| 554 | md.results.StressbalanceSolution.MassFlux3, |
| 555 | md.results.StressbalanceSolution.MassFlux4, |
| 556 | md.results.StressbalanceSolution.MassFlux5, |
| 557 | md.results.StressbalanceSolution.MassFlux6] |
| 558 | }}} |
| 559 | |
| 560 | - Committed `$ISSM_DIR/test/NightlyRun/test101eisnasasmce.py` to repo |
| 561 | |
| 562 | = TODO |
| 563 | - When okayed by Denis, remove, |
| 564 | - `/efs/issm` |
| 565 | - `/issm/ext` |
| 566 | - `/issm/repos` |
| 567 | - `/shared/issm` |
| 568 | |
| 569 | then, move `/efs/issm-new` to `/efs/issm` |
| 570 | - Correct link to `libpython` on Daskhub (should not have to `LD_PRELOAD libpython3.9.so.1.0`) |
| 571 | - Correct link to `libproj` on Daskhub (see `ldd $ISSM_DIR/lib/IssmConfig_python.so`; there is a link to `libproj.so.25`) |
| 572 | - After SMCE team has created `issm` group, modify `$ISSM_DIR/src/m/classes/clusters/eis_nasa_smce.py::LaunchQueueJob` so that tarball is moved rather than copied (look at `/efs/issm/issm/trunk-jpl-denis/src/m/classes/clusters/generic.py` for examples of `cp` as opposed to `scp`) |
| 573 | |
| 574 | == For SMCE team |
| 575 | - Run, |
| 576 | {{{ |
| 577 | ln -s /srv/conda/envs/notebook/include/gstreamer-1.0/gst/gl/gl.h /srv/conda/envs/notebook/include/GL/gl.h |
| 578 | }}} |
| 579 | on Daskhub and bake resulting symbolic link onto base image |
| 580 | - Create group `issm` on SMCE head node |
| 581 | - Change group ownership of `/efs/issm/tmp` to `issm` group |
| 582 | - Review `$ISSM_DIR/src/m/classes/clusters/eis_nasa_smce.py` on Daskhub, in particular |
| 583 | - `self.processor` |
| 584 | - `self.interactive` |
| 585 | - `self.numstreams` |
| 586 | - `self.hyperthreading` |
| 587 | - Can we support for `self.email`? |