[
https://issues.apache.org/jira/browse/CAMEL-26709?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Work on CAMEL-26709 started by shashank.
----------------------------------------
> camel-kafka - SingleNodeKafkaResumeStrategy does not work after a route
> restart, and one bad record stops loading offsets
> -------------------------------------------------------------------------------------------------------------------------
>
> Key: CAMEL-26709
> URL: https://issues.apache.org/jira/browse/CAMEL-26709
> Project: Camel
> Issue Type: Bug
> Components: camel-kafka
> Reporter: shashank
> Assignee: shashank
> Priority: Major
>
> A route stop/start (JMX, the route controller, a route policy) stops the
> resume strategy of the route and starts it again ({{ResumableProcessor}}, and
> {{Splitter}} with a resume strategy). {{SingleNodeKafkaResumeStrategy}} does
> not support this:
> * {{stop()}} closes the producer but keeps it, and {{start()}} only creates a
> producer when there is none. Every later {{updateLastOffset}} fails with
> "Cannot perform operation after producer has been closed" (only logged by the
> resumable completion, so the route keeps running and the offsets are lost;
> after the next application restart the records are processed again).
> * {{stop()}} shuts down the executor but keeps it, and {{loadCache()}}
> submits the refresh task to it again, so the offsets are not loaded any more.
> * {{subscribed}} stays true, so the consumer of the next start does not
> subscribe to the topic.
> Also, one record of the resume topic that cannot be deserialized, for example
> one without a key or value (a tombstone on a compacted topic), makes
> {{poll()}} throw ({{ByteBuffer.wrap(null)}}). {{refresh()}} logs it once and
> the refresh thread ends, so every later offset is no longer loaded.
> CAMEL-23996 fixed the unlock in {{stop()}}, but not the restart.
> Expected: after a stop/start the strategy uses a new producer, executor and
> subscribed consumer, and a record that cannot be deserialized is skipped with
> a WARN.
> _Filed with Claude Code on behalf of allthingssecurity._
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
