[
https://issues.apache.org/jira/browse/KAFKA-1062?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Neha Narkhede closed KAFKA-1062.
--------------------------------
> Reading topic metadata from zookeeper leads to incompatible ordering of
> partition list
> --------------------------------------------------------------------------------------
>
> Key: KAFKA-1062
> URL: https://issues.apache.org/jira/browse/KAFKA-1062
> Project: Kafka
> Issue Type: Bug
> Components: consumer
> Affects Versions: 0.8
> Reporter: Sam Meder
> Assignee: Neha Narkhede
> Attachments: sorted.patch
>
>
> The latest consumer changes to read data from Zookeeper during rebalance have
> made the consumer rebalance code incompatible with older versions (making
> rolling upgrades without downtime hard). The problem relates to how
> partitions are ordered. The old code seems to have returned the partitions
> sorted:
> ... rebalancing the following partitions: ArrayBuffer(0, 1, 2, 3, 4, 5, 6, 7,
> 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19) for topic
> produce-indexable-views with consumers: ...
> the new code instead uses:
> ... rebalancing the following partitions: List(0, 5, 10, 14, 1, 6, 9, 13, 2,
> 17, 12, 7, 3, 18, 16, 11, 8, 19, 4, 15) for topic produce-indexable-views
> with consumers: ...
> This causes new consumers and old consumers to claim the same partitions. I
> realize that this may not be a big deal (although painful for us since it
> disagrees with our deployment automation) since the code wasn't officially
> released, but it seems simple enough to sort the partitions if you'd take
> such a patch.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
