Oh, I think I understand now. You can use this syntax:sandbox://abcd/my_partitioned_file.csv
Using the example screenshot above, a half of the records would be saved in X:/sandboxes/shared/01/abcd/my_partitioned_file.csv on node01 and the rest would be in /opt/CloverETL/sandboxes/shared/02/abcd/my_partitioned_file.csv on node02.
Is this the result you expect?
By the way, I think the shared file system is not necessary in this situation. Both nodes can work on their own piece of data, they do not have to share the data. If you want to use a shared workspace, you can use shared sandbox instead of partitioned one.