Skip to content

DAVAï environment for testing experiment creation

License

Notifications You must be signed in to change notification settings

rolfhm/DAVAI-env

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DAVAI environment & interface

This project contains:

  • the DAVAI command-line tools, to create DAVAI experiments and run them,
  • the configuration files to handle general preferences and machine-dependent installation,
  • the documentation

Note that the core of DAVAI (tests templates, jobs sequences, config files and launching wrappers) are hosted separately, in the DAVAI-tests repository.

Installation

  • on belenos@MF

  • on Atos/bologna@ECMWF

  • To setup your own install:

    • Clone this repository, e.g. in ~/repositories/:
      git clone https://github.com/ACCORD-NWP/DAVAI-env.git
    • Set paths:
      • DAVAI_ENV=~/repositories/DAVAI-env
      • export PATH=$DAVAI_ENV/bin:$PATH
      • export PYTHONPATH=$DAVAI_ENV/src:$PYTHONPATH
    • Assert dependencies (see below) are satisfied
  • If you want to inspect possible customizations:

    • davai-config show
    • davai-config preset_user

Quick start

A Tutorial is available in the User Guide (cf. below).

For a quick start:

  1. Prepare an experiment based on version <v> of the tests, to validate an IAL Git reference <r>:
    davai-new_xp <r> -v <v> [-h]
    (you may need to specify the path to your IAL repository by argument, cf. options with -h)
    To know what version of the tests to use, cf. below.
  2. Go to the prompted directory of the experiment (ending with .../dv-<nnnn>-<platform>@<user>/davai/nrv/)
  3. Run the tests: davai-run_xp and monitor (standard output for the build, then job scheduler).
  4. If you need to re-build & re-run tests:
  • davai-build [-h]
  • davai-run_tests [-h]

More details to be found in the User Guide.

Tests versions and reference experiments

=> https://github.com/ACCORD-NWP/DAVAI-tests/wiki/Versions-of-tests

Documentation

The User Guide is available for main releases on: https://github.com/ACCORD-NWP/DAVAI-env/releases

Otherwise it is also available under doc/ directory. Part of the documentation needs to be compiled from .tex sources, using pdflatex. To do so, from within the repository root, type:

  • make doc

and the generated PDF document will be found under doc/pdf/Davai_User_Guide.pdf.

Dependencies

DAVAI is mainly written in Python3. Make sure you have Python3.6 at least. It also uses Git, make sure you have a "recent enough" version of it, or some commands may not work properly. DAVAI works over a number of NWP packages, tools, software, that need to be installed on the machine with their own procedures. These include:

  • Vortex: scripting system used for the definition of tasks (resources, executables launch, ...) and the running of the jobs. It embeds a number of necesary-as-well sub-packages.
  • ecbundle: a utility from ECMWF to gather codes from several repositories, in the required version for each, based on a YAML descriptive file (called bundle); used in IAL-build
  • IAL-build: wrappers around git and gmkpack (and eventually other building tools) to build IAL executables from Git
  • IAL-expertise: tools to analyse automatically the outputs of NWP configurations -- norms, Jo-tables, fields in FA/GRIB files, ...
  • EPyGrAM: a Python library for handling output data from the IAL models; it is used here within IAL-expertise.

These packages may already be pre-installed on MF's HPC or other platforms where DAVAI is already ported. Cf. the "Local complements" in "Installation" section, in this case.

To install them on a new machine, cf. the projects' installation instructions.

Lexicon

  • DAVAI = "Device Aiming at the VAlidation of IAL"
  • IAL = IFS-Arpege-LAM

About

DAVAï environment for testing experiment creation

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 93.5%
  • Tcl 3.7%
  • Shell 1.9%
  • Makefile 0.9%