CROW Documentation  0.1
 All Pages
Quick End-to-end

How to setup a stock experiment with Rocoto:

Checkout repository “CROWdemo2” branch

git clone gerrit:global-workflow cd global-workflow git checkout CROWdemo2

Get CROW submodule

git submodule update –init –recursive

Build/link global-workflow

> cd sorc
> sh checkout.sh
> sh build_all.sh
> sh link_fv3gfs.sh emc $target
...where $target is machine (wcoss_cray, theia, jet, or dell)

Set up user file

While at the top of global-workflow clone:

cd workflow

Copy default user file to your user file:

> cp user.yaml.default user.yaml

Open and modify user.yaml, the following are required to be set:

- PROJECT_DIR: Place for experiment directory, make sure you have write access.
- cpu_project: cpu project that you are working with.
- hpss_project: hpss project that you are working with.

Run experiment setup script

> cd CROW
> ./setup_case.sh tutorial_free_forecast test

If you see the following error:

ERROR:crow.model.fv3gfs:More than one platform is available: $PLAT1, $PLAT2
ERROR:crow.model.fv3gfs:Pick one with -p option

That means you could access more than one computing platform from this environment (which is good). You need to make selection of computing platform between $PLAT1 and $PLAT2 by do the following:

In CROW, there's a '-p' command option to the setup_case command:

> ./setup_case.sh -p $PLAT_x tutorial_free_forecast test

Execute this instead of the standard one, while $PLAT_x is the platform that you want to select.

Run workflow generation script

> ./make_rocoto_xml_for.sh $EXPERIMENT_DIRECTORY

...where $EXPERIMENT_DIRECTORY is the location of the folder created in the prior step (which contains your generated config files); this is where your resulting xml file will be found

Navigate to your experiment directory and start your experiment with Rocoto:

module load rocoto rocotorun -d workflow.db -w workflow.xml