Currently the dask distributed clients are instructured to save the results of the tasks to disk instead of sending them back and this results in the model runs being saved in the individual worker pods.
[ ] change the logic of model runner to return the results back to master node and therefore the webapp and save the results after in a single file (to get closer to a db write)
[ ] once the database is set up save the results directly to DB
Currently the dask distributed clients are instructured to save the results of the tasks to disk instead of sending them back and this results in the model runs being saved in the individual worker pods.