Resume Sweep after stop

[Delaunay]

my-app/
├─ my_app.py
├─ multirun/
│  ├─ 2022-11-09 
│  │  ├─ 12-37-09            <= old experiment we want to resume from
│  │  │  ├─ experiment
│  │  │  ├─ datatbase.pkl   <= old orion database 

To resume the sweep you should set the database path to the absolute path of the old sweep you want to resume

python my_app.py -m hydra.sweeper.storage.database.host=$(pwd)/multirun/2022-11-09/12-37-09/datatbase.pkl

Could we support resuming directly ?

It all depends on when the folders for the new sweep are created, if they are created soon enough we could copy the old database to the new location for the user so he would not have to change the db path

python my_app.py -m hydra.sweeper.resume=$(pwd)/multirun/2022-11-09/12-37-09/datatbase.pkl