[
https://issues.apache.org/jira/browse/FLINK-3404?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15265376#comment-15265376
]
Aljoscha Krettek commented on FLINK-3404:
-----------------------------------------
I think implementing atomic checkpoint-and-stop is harder than it is made to
look here. The problem is the synchronization of all parallel sources in
stopping and checkpointing. I have some ideas on this but I think it is a
bigger think that warrants at least a discussion on the dev list and possibly
(probably) a design doc. (This is independent of the Kafka source.)
> Extend Kafka consumers with interface StoppableFunction
> -------------------------------------------------------
>
> Key: FLINK-3404
> URL: https://issues.apache.org/jira/browse/FLINK-3404
> Project: Flink
> Issue Type: Improvement
> Components: Kafka Connector
> Reporter: Matthias J. Sax
>
> Kafka consumers are not stoppable right now. To make them stoppable, they
> must implement {{StoppableFunction}}. Implementing method {{stop()}} must
> ensure, that the consumer stops pulling new messages from Kafka and issues a
> final checkpoint with the last offset. Afterwards, {{run()}} must return.
> When implementing this, keep in mind, that the gathered checkpoint might
> later be used as a savepoint.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
