[ https://issues.apache.org/jira/browse/SPARK-44970?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17850475#comment-17850475 ]
Steve Loughran commented on SPARK-44970: ---------------------------------------- correct. file is only saved on close(). The incomplete multipart uploads are still there and you do get billed for them -which is why its critical to have a lifecycle rule to clean this stuff up. In theory you may be able to rebuild it on a failure, in practise you'd have a hard time working out the order and you'll still be short the last 32-64 MB of data > Spark History File Uploads Can Fail on S3 > ----------------------------------------- > > Key: SPARK-44970 > URL: https://issues.apache.org/jira/browse/SPARK-44970 > Project: Spark > Issue Type: Bug > Components: Spark Core > Affects Versions: 4.0.0 > Reporter: Holden Karau > Assignee: Holden Karau > Priority: Major > > Sometimes if the driver OOMs the history log will not upload finish. -- This message was sent by Atlassian Jira (v8.20.10#820010) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org