Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: Migrated to Confluence 5.3

Table of Contents

Run the code via Make It So!

The easiest way to configure models coupled to the LFM is via the script generator affectionately named "Make It So". MakeItSo requires Python >= 2.4.

Usage

Once you've set the corresponding environment variables, learn more about MakeItSo via the --help flag:

Code Block
makeItSo.py --help

There are two ways to run the script generator:

  1. Interactive mode: At the command-line, type

    Code Block
    /path/to/makeItSo.py
    

    and follow the prompts. You will be asked to set a series of parameters. Press <enter> for the default value or ? <enter> for more information on a particular variable. Once you've entered all model parameters & environment setup, MakeItSo will create input files & job scripts necessary to run the code.

  1. Preconfigured run mode: Each time you run the script generator, it writes a makeItSo.config file at:

    Code Block
    Output directory/RUN_IDENTIFIER_makeItSo.config
    

    Once you've run the script generator via interactive mode (see above), open up a *makeItSo.config file with a text editor. You can edit run parameters, save the file and configure a new run using this simple text file. Execute:

    Code Block
    /path/to/makeItSo.py -c /path/to/RUN_IDENTIFIER_makeItSo.config
    

    MakeItSo will create all the input files & job scripts necessary to run the code.

Make It So will print instructions and create a set of files:

  • Input files that set model parameters
  • Shell script to configure the models (RUN_IDENTIFIER-setup.sh)
  • Run scripts to execute the models (RUN_IDENTIFIER-001.*)

Follow the instructions displayed by Make It So to run the code.

FAQ

How do I restart or continue a long run?

Say you ran MakeItSo to generate a 2 hour run of LFM. You should have several files in your run directory, including:

  • LFM-001.sh: 50 minute spinup interval
  • LFM-002.sh: Job script for 1st hour of run
  • LFM-003.sh: Job script for 2nd hour of run
  • SW-SM-DAT: 6 hours of solar wind data from OMNI
  • LFM_makeItSo.config: configuration file for MakeItSo.

You execute the job scripts and your 2 hour run (plus 50 minute spinup) have completed.  Let's say you want to restart the run and continue for an additional 2 hours.  There are two ways to do this:

  1. Extend the run (recommended)
    1. Edit LFM_makeItSo.config and set the stop top tim 2 hours later
    2. Re-run MakeItSo:

      Code Block
      makeItSo.py -c LFM_makeItSo.config
    3. Submit the next job script in the iteration: LFM-004.sh. This will pick up where the previous job script left off and continue executing the model.
  2. Configure a restart run (advanced)
    1. Edit LFM_makeItSo.config and
      1. change start time to 2 hours later
      2. change stop time to 2 hours later
      3. add the the parameter SPINUP: false to Shared model settings
    2. Re-run MakeItSo:

      Code Block
      makeItSo.py -c LFM_makeItSo.config

      This will create the following run scripts:

      • {LFM-001.sh}}: Job script for 3rd hour of run
      • {LFM-002.sh}}: Job script for 4th hour of run
        Note: if you ran MakeItSo in the original run directory, you may have overwritten the pre-existing job scripts!

How do I turn on HDF5 file support for MFLFM on stampede and yellowstone?

  1. Stampede
    1. First, the hdf5 module needs to be loaded before MFLFM can be built.

      Code Block
      languagetext
      module load hdf5
    2. Build MFLFM as normal.

    3. To turn on HDF5 support, makeItSo.py needs to be run in expert mode. (Warning: This may prompt you to enter dozens of model parameters).  Not using expert mode will default back to HDF4.

      Code Block
      languagetext
      makeItSo.py --expert
    4. When the model parameter option "IO_EXTENSIONS( hdf ):" is reached, enter "hdf5".  Entering "hdf", or just hitting return will default back to HDF4.

      Code Block
      languagetext
              IO_EXTENSION ( hdf ): hdf5
    5. HDF5 file support for MFLFM is now turned on.  Run MFLFM as normal, there should be .hdf5 files in the run directory.

       

  2. Yellowstone
    1. On yellowstone, the hdf5 module doesn't need to be loaded.  Follow the Stampede directions starting at 1b.

  3. More details on changes to help add HDF5 support on other machines.
    1. Changes to env/Makefile.${MACHINE}
      1. Stampede changes to env/Makefile.${MACHINE}
        1. Add HDFnompi path to HDF5 libraries.

          Code Block
          languagetext
          HDF5nompi= /opt/apps/intel13/hdf5/1.8.9
        2. There was also issues with finding the szip library, so EXTRA_LIBS needs to set to find szip.

          Code Block
          languagetext
          EXTRA_LIBS = -L/opt/apps/intel13/hdf5/1.8.9/lib -lsz
      2. Yellowstone changes to env/Makefiel.${MACHINE}
        1. Add HDFnompi path to HDF5 libraries.

          Code Block
          languagetext
          HDF5nompi = /glade/apps/opt/hdf5/1.8.11/intel/12.1.5
    2. There may also need to be changes to the machines enviroment.  On yellowstone, there was no changes needed from the default machine enviroment.  On stampede, the hdf5 module needed to be loaded.    On other systems, the hdf5 and szip modules might need to be loaded, if they exist. 
    3. Switching back and forth between HDF4 and HDF5 file support is possible by running makeItSo.py --expert and specifying a different IO_EXTENSION.  The input files will need to be either HDF4 or HDF5, you cannot mix HDF4 input files with HDF5 output files, and vice versa   Running makeItSo.py without --expert will default to HDF4.

Extending MakeItSo

Note
titleNote

Adding a new environment to MakeItSo is an advanced topic. Do not hesitate to ask for help with initial configuration.

We have tested MakeItSo on a variety of systems.  You can customize for your particular machine in two ways:

Create Config File (recommended)

You can permanently add your environment by defining all parameters in a QUEING-ENVIRONMENT.yml file. This file must reside in the MakeItSo/Environments subdirectory. It describes the computer (MPI installation, flags for the number of processors, etc) and any queue information that is required. We recommend that you start with a pre-existing environment & extend it to your own environment. For example, see LSF.yml or SGE.yml.

At runtime, MakeItSo will locate this YAML file & provide an interface to the user to set up the environment.

Modify at Run Time

You can modify a pre-configured environment at run-time. Start makeItSo.py in "expert" mode (Warning: This may prompt you to enter dozens of model parameters)

Code Block
misc/python/makeItSo.py --expert

When prompted for an environment, select the one most similar to yours. If you are working on a Linux desktop computer, select "Standard Linux Bourne shell without a queue". You will be asked to enter a series of parameters related to your compute environment. You must know the settings used to compile the LTR prerequisites.