GitHub user srdo opened a pull request:

    https://github.com/apache/storm/pull/2907

    STORM-2990, STORM-3279: Fix issue where Kafka Trident spout could ign…

    …ore EARLIEST and LATEST, and make EARLIEST and LATEST only take effect 
on topology deploy
    
    See https://issues.apache.org/jira/browse/STORM-2990 and 
https://issues.apache.org/jira/browse/STORM-3279.
    
    3279 is fixed by always returning a metadata from the emitter, even if no 
tuples were emitted. If no tuples are emitted, the metadata will now indicate 
that `consumer.position() - 1` was emitted.
    
    2990 is fixed by inspecting the lastBatchMeta topology id, and deciding 
whether to start over based on that. 
    
    I expanded the emitter tests to cover these cases as well, and tried to 
make them a bit more readable by splitting them into multiple classes. The 
Kafka client dependency has been bumped to 0.11.0.0, since the MockConsumer 
wasn't behaving properly in 0.10.1.0.
    
    The only other change is to replace the KafkaUnit internals so we no longer 
depend on Kafka's internal test tooling.

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/srdo/storm STORM-2990

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/storm/pull/2907.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #2907
    
----
commit 9247e4a59a2f8829238179066a7299fa1cafdbaa
Author: Stig Rohde Døssing <srdo@...>
Date:   2018-11-14T15:17:03Z

    STORM-2990, STORM-3279: Fix issue where Kafka Trident spout could ignore 
EARLIEST and LATEST, and make EARLIEST and LATEST only take effect on topology 
deploy

----


---

Reply via email to