turboFei edited a comment on issue #25863: [WIP][SPARK-29037][CORE][SQL] For static partition overwrite, spark may give duplicate result. URL: https://github.com/apache/spark/pull/25863#issuecomment-534064726 When there are two write can execute currently. such as A:`Insert overwrite table ta partition(a=1,b=2) select ...` and B:`Insert overwrite table ta partition(a=1,b=3) select ...` . The file tree is: ``` .spark-staging-2/a=1/b=2 /b=3 ``` When A is complete, we should delete `b=2` firstly. Then we need list files under `a=1`, if there is not another partition path, we should delete `a=1`. In this case, B is not completed, we can not delete `a=1`. When we delete output path, we also rely on file listing in the commit protocol. But when we detecting conflict, we do not rely on file listing.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
