Debugging
It can be tricky to debug Dask run, even with the LocalCluster.
As of late 2024, debugging with PyCharm a LocalCluster fails.
Debug Unittests
The easiest way is to run for example
pytest -s test_generic_zarr_handler_netcdf3_netcdf4.py
and having anywhere in the code
import ipdb; ipdb.set_trace()
Debug a cloud optimised pipeline
Modify a dataset processing script to use a None cluster, for example:
def main():
command = [
"generic_cloud_optimised_creation",
"--paths",
"IMOS/SRS/SST/ghrsst/L3S-1dS/dn/",
"--dataset-config",
"satellite_ghrsst_l3s_1day_daynighttime_single_sensor_southernocean.json",
# "--cluster-mode",
# "local",
]
and add anywhere in the code
import ipdb; ipdb.set_trace()
Run as normal