[
https://issues.apache.org/jira/browse/KAFKA-5075?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15971639#comment-15971639
]
Jiangjie Qin commented on KAFKA-5075:
-------------------------------------
[~hachikuji] I also think it is worth fixing. [~gwenshap] I'll cherry-pick the
fix to 0.10.2 branch if you don't mind having another RC.
> Defer exception to the next pollOnce() if consumer's fetch position has
> already increased
> -----------------------------------------------------------------------------------------
>
> Key: KAFKA-5075
> URL: https://issues.apache.org/jira/browse/KAFKA-5075
> Project: Kafka
> Issue Type: Bug
> Components: clients, consumer
> Affects Versions: 0.10.2.0
> Reporter: Jiangjie Qin
> Assignee: Dong Lin
> Fix For: 0.11.0.0
>
>
> In Fetcher.fetchRecords() we iterate over the partition data to collect the
> ConsumerRecords, after we collect some consumer records from a partition, we
> advance the position of that partition then move on to the next partition. If
> the next partition throws exceptions (e.g. OffsetOutOfRangeException), the
> messages that have already been read out of the buffer will not be delivered
> to the users. Since the positions of the previous partitions have been be
> updated, those messages will not be consumed again either.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
