[
https://issues.apache.org/jira/browse/FLINK-17816?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17120130#comment-17120130
]
Jiayi Liao commented on FLINK-17816:
------------------------------------
[~sewen] Yes it's much simpler and I've created a PR for it.
BTW I see that you comment in
[#TestProcessingTimeService.java#L117|https://github.com/apache/flink/blob/955a6832b9fc3f68c5accef3a22a0b4d45140d68/flink-streaming-java/src/main/java/org/apache/flink/streaming/runtime/tasks/TestProcessingTimeService.java#L117]
{code:java}
// for all testing purposed, there is no difference between the fixed rate and
fixed delay
{code}
Are you sure we should replace fixed delay with fixed rate in unit testing?
because it may affect the testing result like
#StreamSourceOperatorLatencyMetricsTest#testLatencyMarkEmissionEnabledOverrideViaExecutionConfig.
(The unit testing should fail after we use the \{{scheduleAtFixedDelay}})
> Change Latency Marker to work with "scheduleAtFixedDelay" instead of
> "scheduleAtFixedRate"
> ------------------------------------------------------------------------------------------
>
> Key: FLINK-17816
> URL: https://issues.apache.org/jira/browse/FLINK-17816
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Task
> Reporter: Stephan Ewen
> Priority: Major
> Labels: pull-request-available
>
> Latency Markers and other periodic timers are scheduled with
> {{scheduleAtFixedRate}}. That means every X time the callable is called. If
> it blocks (backpressure) is can be called immediately again.
> I would suggest to switch this to {{scheduleAtFixedDelay}} to avoid calling
> for a lot of latency marker injections when there is no way to actually
> execute the injection call.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
