[
https://issues.apache.org/jira/browse/SPARK-24738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16553903#comment-16553903
]
Apache Spark commented on SPARK-24738:
--------------------------------------
User 'LiShuMing' has created a pull request for this issue:
https://github.com/apache/spark/pull/21856
> [HistoryServer] FsHistoryProvider clean outdated event logs at start
> ---------------------------------------------------------------------
>
> Key: SPARK-24738
> URL: https://issues.apache.org/jira/browse/SPARK-24738
> Project: Spark
> Issue Type: Bug
> Components: Spark Core
> Affects Versions: 2.3.1
> Reporter: ShuMing Li
> Priority: Minor
>
> Now FsHistoryProvider's policy :
> * 1. FsHistoryProvider create checkEvent thread and cleanLog thread, they
> share one thread pool to avoid conflicts;
> * 2. check thread replay all event logs first;
> * 3. clean thread iterators all events to check the outdated events and
> remove about it;
>
> Why check thread at step2 not clean the outdated eventLogs at start, so no
> need replay the outdated logs to save times ?
>
> ```
> // Used by check event thread and clean log thread.
> // Scheduled thread pool size must be one, otherwise it will have concurrent
> issues about fs
> // and applications between check task and clean task
> ```
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
