[
https://issues.apache.org/jira/browse/SPARK-27663?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16836392#comment-16836392
]
Sean Owen commented on SPARK-27663:
-----------------------------------
Also, you'd need to reproduce on master. 2.1.0 is very old.
> Task accomplished incompletely but marked as success
> ----------------------------------------------------
>
> Key: SPARK-27663
> URL: https://issues.apache.org/jira/browse/SPARK-27663
> Project: Spark
> Issue Type: Bug
> Components: Spark Core, SQL
> Affects Versions: 2.1.0
> Reporter: Yunbo Fan
> Priority: Major
> Attachments: image-2019-05-09-11-10-04-602.png, incomplte-task-0.png,
> incomplte-task-1.png, incomplte-task-2.png, reran-0.png, reran-1.png
>
>
> It happens when running sql queries using spark sql.
> The task was accomplished incompletely but marked as success since there were
> not any exceptions and failed or killed tasks.
> When I checked the query result， it missed about 4000 records.
> The history web ui shows that the task input size is 23.5 MB, but the log in
> the executor shows the split size is 326992763, about 300 MB.
> And this task was finished in 1 second, but others’ duration was about 15
> seconds.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
