[
https://issues.apache.org/jira/browse/KAFKA-6661?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16401254#comment-16401254
]
Ewen Cheslack-Postava commented on KAFKA-6661:
----------------------------------------------
[~rhauch] Got this cherry-picked back to 0.10.0. If we want 0.9.0 as well, we
probably need a separate PR since there was enough code movement to make it
non-trivial.
On a related note, it wasn't too bad in this case, but for things we want to
backport its better to separate the must-have stuff from the nice-to-have
improvements. It's not so much the risk (in this case it was just a toString
and some log statements), but that the larger the patch, the less likely we get
a clean cherry-pick and that's a lot more time consuming in cases where we want
to cherry-pick through a bunch of release branches. In this case the pain was
really caused by a different commit that mostly made cosmetic improvements that
make cherry-picking encounter conflicts, but something to keep in mind in the
future.
> Sink connectors that explicitly 'resume' topic partitions can resume a paused
> task
> ----------------------------------------------------------------------------------
>
> Key: KAFKA-6661
> URL: https://issues.apache.org/jira/browse/KAFKA-6661
> Project: Kafka
> Issue Type: Bug
> Components: KafkaConnect
> Affects Versions: 0.9.0.0, 0.10.0.0, 0.11.0.0, 1.0.0
> Reporter: Randall Hauch
> Assignee: Randall Hauch
> Priority: Critical
> Fix For: 0.10.0.2, 0.10.1.2, 0.10.2.2, 0.11.0.3, 1.0.2, 1.1.1
>
>
> Sink connectors are allowed to use the {{SinkTaskContext}}'s methods to
> explicitly pause and resume topic partitions. This is useful when connectors
> need additional time processing the records for specific topic partitions
> (e.g., the external system has an outage).
> However, when the sink connector has been paused via the REST API, the worker
> for the sink tasks pause the consumer. When the connector is polled, the poll
> request might timeout and return no records. Connect then calls the task's
> {{put(...)}} method (with no records), and this allows the task to optionally
> call any of the {{SinkTaskContext}}'s pause or resume methods. If it calls
> resume, this will unexpectedly resume the paused consumer, causing the
> consumer to return messages and the connector to process those messages --
> despite the connector still being paused.
> This is reported against 1.0, but the affected code has not been changed
> since at least 0.9.0.0.
> A workaround is to remove rather than pause a connector. It's inconvenient,
> but it works.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
