You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "ASF GitHub Bot (JIRA)" <ji...@apache.org> on 2017/08/01 07:36:00 UTC

[jira] [Commented] (FLINK-7298) Records can be cleared when all data in state is invalid

    [ https://issues.apache.org/jira/browse/FLINK-7298?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16108517#comment-16108517 ] 

ASF GitHub Bot commented on FLINK-7298:
---------------------------------------

Github user fhueske commented on the issue:

    https://github.com/apache/flink/pull/4421
  
    Hi @hequn8128, I think the change is good.
    
    I don't think we need to remove `!validTimestamp` though. The clean-up is a best effort cleanup and will register another clean-up timer if there are records left in the state. So, we might not remove all expired records in one run but also do not have to iterate over all timestamps. This is an optimization for RocksDB state which provides the keys in sorted order, so in this case all expired records are removed without going over all keys of the MapState.


> Records can be cleared when all data in state is invalid
> --------------------------------------------------------
>
>                 Key: FLINK-7298
>                 URL: https://issues.apache.org/jira/browse/FLINK-7298
>             Project: Flink
>          Issue Type: Improvement
>            Reporter: Hequn Cheng
>            Assignee: Hequn Cheng
>
> In {{ProcTimeWindowInnerJoin}}.{{expireOutTimeRow}}, we need not to remove records from state when there is no valid records. Instead, we can clear them all at once.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)