Description
Today I saw a job in which a reduce stage failed and showed up a lot of times in the failed stages. I think the reason is that the DAGScheduler stage complete (with failure) event multiple times in the case of FetchFailed.
Attachments
Issue Links
- duplicates
-
SPARK-2947 DAGScheduler resubmit the stage into an infinite loop
- Resolved
- is duplicated by
-
SPARK-2947 DAGScheduler resubmit the stage into an infinite loop
- Resolved
- links to