[
https://issues.apache.org/jira/browse/FLINK-9681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16533618#comment-16533618
]
ASF GitHub Bot commented on FLINK-9681:
---------------------------------------
Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/6255#discussion_r200330683
--- Diff: docs/dev/table/streaming.md ---
@@ -591,16 +589,14 @@ qConfig.withIdleStateRetentionTime(Time.hours(12);
val qConfig: StreamQueryConfig = ???
-// set idle state retention time: min = 12 hour, max = 16 hours
-qConfig.withIdleStateRetentionTime(Time.hours(12), Time.hours(16))
-// set idle state retention time. min = max = 12 hours
-qConfig.withIdleStateRetentionTime(Time.hours(12)
+// set idle state retention time: min = 12 hour, max = 24 hours
+qConfig.withIdleStateRetentionTime(Time.hours(12), Time.hours(24))
{% endhighlight %}
</div>
</div>
-Configuring different minimum and maximum idle state retention times is
more efficient because it reduces the internal book-keeping of a query for when
to remove state.
+Configuring different minimum and maximum idle state retention times is
more efficient because it reduces the internal book-keeping of a query for when
to remove state. Difference between minTime and maxTime shoud be at least 5
minutes.
--- End diff --
The "... more efficient ..." does not apply anymore. Maybe rephrase to
> Cleaning up state requires additional bookkeeping which becomes less
expensive for larger differences of `minTime` and `maxTime`. The difference
between `minTime` and `maxTime` must be at least 5 minutes.
> Make sure minRetentionTime not equal to maxRetentionTime
> --------------------------------------------------------
>
> Key: FLINK-9681
> URL: https://issues.apache.org/jira/browse/FLINK-9681
> Project: Flink
> Issue Type: Improvement
> Components: Table API & SQL
> Reporter: Hequn Cheng
> Assignee: Hequn Cheng
> Priority: Major
> Labels: pull-request-available
>
> Currently, for a group by(or other operators), if minRetentionTime equals to
> maxRetentionTime, the group by operator will register a timer for each record
> coming at different time which cause performance problem. The reasoning for
> having two parameters is that we can avoid to register many timers if we have
> more freedom when to discard state. As min equals to max cause performance
> problem it is better to make sure these two parameters are not same.
> Any suggestions are welcome.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
