[
https://issues.apache.org/jira/browse/KAFKA-5866?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17036622#comment-17036622
]
Konstantine Karantasis commented on KAFKA-5866:
-----------------------------------------------
Returning here to add another counter argument. After
https://cwiki.apache.org/confluence/display/KAFKA/KIP-558%3A+Track+the+set+of+actively+used+topics+by+connectors+in+Kafka+Connect
the Connect workers will start keeping track of the set of topics that a
connector is actively using. But when such connector gets deleted, this
information should be deleted as well.
Allowing connectors to produce records to topics after they've been stopped
would complicate things in many cases (as for example with KIP-558) and seems
to complicate the programming model for Kafka Connect and its connectors.
> Let source/sink task to finish their job before exit
> ----------------------------------------------------
>
> Key: KAFKA-5866
> URL: https://issues.apache.org/jira/browse/KAFKA-5866
> Project: Kafka
> Issue Type: Bug
> Components: KafkaConnect
> Affects Versions: 0.10.2.0
> Reporter: Oleg Kuznetsov
> Priority: Major
>
> My case is about reading files. When task stops to rebalance or for other
> reason, I want let it to read file till the end at least.
> I found that flag
> {code:java}
> WorkerTask#stopping
> {code}
> is set to true and only then
> {code:java}
> SourceTask.stop()
> {code}
> is called. This stopping flag prevents WorkerSourceTask from further
> ingestion (exit from
> {code:java}
> while ( !isStopped()))
> {code}.
> Is it possible to let task to decide to work some more time and possibly
> produce more records from the moment of stop() was called on rebalance?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
