[GitHub] flink issue #3378: [FLINK-5849] [kafka] Move FlinkKafkaConsumer start offset...

[tzulitai <gi...@git.apache.org>]

Github user tzulitai commented on the issue:

    https://github.com/apache/flink/pull/3378
  
    Rebased on `master`.
    
    Note about changes to partition assignment logic in deleted lines 538 - 553 and added lines 563 -565 of `FlinkKafkaConsumerBase`:
    The change is irrelevant to this issue, but something I stumbled across when touching that part of the code. Problems:
    
    1. The `KafkaConsumerPartitionAssignmentTest` was testing a no-longer used `assignPartitions` method, so the tests actually never covered the actual behaviour.
    
    2. Previously, the partition assignment was changed from using the "modulo on KafkaTopicPartition hashes" approach to "pre-sorting the partition list and round-robin assigning". This change should actually breaks the tests in `KafkaConsumerPartitionAssignmentTest`, but didn't because as mentioned above, the tests were testing an unused method. The current approach will also be problematic for dynamically growing subscribed partition lists, because the sorting order will change as the list grows with newly discovered partitions.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---