Github user cloud-fan commented on the issue:
https://github.com/apache/spark/pull/20414
> Not quite - coalesce will not combine partitions across executors (aka
shuffle) so you could still end up having many many files.
I'm not sure if I follow here. For `coalesce(1)` Spark just launches a
single task to handle all the data partitions. If the final action is saving
file, we still have only one file at the end. Compared to `repartition(1)`, I
think the only difference is the cost of task retry.
I think we can special case `repartition(1)`, if there is only one reducer,
we don't need to add the local sort.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
