Github user szhem commented on the issue:
https://github.com/apache/spark/pull/19373
> so the fix might be to change to call checkpoint() to checkpoint(eager:
true) - this ensures by the time checkpoint call is returned the checkpointing
is completed.
Even if checkpoint is completed, `PeriodicRDDCheckpointer` removes files of
the checkpointed and materialized RDDs later on, so it may happen that another
RDD depends on the already removed files.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
