[ 
https://issues.apache.org/jira/browse/SPARK-29302?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17067972#comment-17067972
 ] 

koert kuipers edited comment on SPARK-29302 at 3/26/20, 7:23 PM:
-----------------------------------------------------------------

i believe we are seeing this issue. it shows up in particular when pre-emption 
is turned on and we are using dynamic partition overwrite. pre-emption kills 
tasks, they get restarted, and then they fail again because the output 
directory already exists (so task throws FileAlreadyExistsException). as a 
result entire job fails.

so i dont think this is just a speculative execution issue. this is a general 
issue with dynamic partition overwrite not being able to recover from task 
failure.


was (Author: koert):
i believe we are seeing this issue. it shows up in particular when pre-emption 
is turned on and we are using dynamic partition overwrite. pre-emption kills 
tasks, they get restarted, and then they fail again because the output 
directory already exists (so task throws FileAlreadyExistsException). as a 
result entire job fails.

> dynamic partition overwrite with speculation enabled
> ----------------------------------------------------
>
>                 Key: SPARK-29302
>                 URL: https://issues.apache.org/jira/browse/SPARK-29302
>             Project: Spark
>          Issue Type: Bug
>          Components: SQL
>    Affects Versions: 2.4.4
>            Reporter: feiwang
>            Priority: Major
>         Attachments: screenshot-1.png, screenshot-2.png
>
>
> Now, for a dynamic partition overwrite operation,  the filename of a task 
> output is determinable.
> So, if speculation is enabled,  would a task conflict with  its relative 
> speculation task?
> Would the two tasks concurrent write a same file?



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org

Reply via email to