[
https://issues.apache.org/jira/browse/NIFI-11305?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Matt Burgess reopened NIFI-11305:
---------------------------------
Reopening as there is an order-of-events bug where the binlog client waits on
the event listener thread which no longer has anything pulling events off the
queue. The event listener should be unregistered before the binlog client
disconnects.
> CaptureChangeMySQL does not stop if the queue is not empty
> ----------------------------------------------------------
>
> Key: NIFI-11305
> URL: https://issues.apache.org/jira/browse/NIFI-11305
> Project: Apache NiFi
> Issue Type: Bug
> Components: Extensions
> Reporter: Matt Burgess
> Assignee: Matt Burgess
> Priority: Major
> Fix For: 2.0.0, 1.21.0
>
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> There is a logic bug in the handling of stopping the CaptureChangeMySQL
> processor. This causes it to not stop while there are events in the queue. If
> the processor isn't running fast enough to drain the queue, stopping the
> processor will have no effect.
> The logic was being handled in OnStopped, but that won't get called until the
> onTrigger has finished. Instead the loop should be checking to see if the
> processor is still scheduled using isScheduled(), and if not should break out
> of the loop and finish the onTrigger processing, thereby allowing the
> OnStopped logic to be executed.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
