[
https://issues.apache.org/jira/browse/KAFKA-12726?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17335389#comment-17335389
]
Chris Egerton commented on KAFKA-12726:
---------------------------------------
[~ryannedolan] can you confirm that this has been observed with 2.8.0? This
sounds similar to https://issues.apache.org/jira/browse/KAFKA-10792, which
should be fixed in 2.8.0.
I'm also not sure about the statement that "Workers stop Tasks sequentially"–my
understanding is that workers _trigger_ task stops sequentially (see
[Worker::stopTasks|https://github.com/apache/kafka/blob/f9de25f046452b2a6d916e6bca41e31d49bbdecf/connect/runtime/src/main/java/org/apache/kafka/connect/runtime/Worker.java#L849-L855],
which invokes
[Worker::stopTask|https://github.com/apache/kafka/blob/f9de25f046452b2a6d916e6bca41e31d49bbdecf/connect/runtime/src/main/java/org/apache/kafka/connect/runtime/Worker.java#L827-L847]
consecutively for each to-be-stopped task), but we can see under the hood in
[WorkerTask::stop and
WorkerTask::triggerStop|https://github.com/apache/kafka/blob/f9de25f046452b2a6d916e6bca41e31d49bbdecf/connect/runtime/src/main/java/org/apache/kafka/connect/runtime/WorkerTask.java#L105-L120]
that this should complete almost immediately, and neither
[WorkerSinkTask|https://github.com/apache/kafka/blob/f9de25f046452b2a6d916e6bca41e31d49bbdecf/connect/runtime/src/main/java/org/apache/kafka/connect/runtime/WorkerSinkTask.java#L157-L162]
nor
[WorkerSourceTask|https://github.com/apache/kafka/blob/f9de25f046452b2a6d916e6bca41e31d49bbdecf/connect/runtime/src/main/java/org/apache/kafka/connect/runtime/WorkerSourceTask.java#L215-L219]
override that method with anything that should block, either. The graceful
shutdown period is then enforced en masse in
[Worker::awaitStopTasks|https://github.com/apache/kafka/blob/f9de25f046452b2a6d916e6bca41e31d49bbdecf/connect/runtime/src/main/java/org/apache/kafka/connect/runtime/Worker.java#L880-L887].
Given this, I'd be very curious about a reproduction case in the event that my
misunderstanding is complete or there is an edge case that is not accounted for
currently. Perhaps we could leverage the existing
[BlockingConnectorTest|https://github.com/apache/kafka/blob/f9de25f046452b2a6d916e6bca41e31d49bbdecf/connect/runtime/src/test/java/org/apache/kafka/connect/integration/BlockingConnectorTest.java]
integration test to try to demonstrate where things break down?
> misbehaving Task.stop() can prevent other Tasks from stopping
> -------------------------------------------------------------
>
> Key: KAFKA-12726
> URL: https://issues.apache.org/jira/browse/KAFKA-12726
> Project: Kafka
> Issue Type: Bug
> Components: KafkaConnect
> Affects Versions: 2.8.0
> Reporter: Ryanne Dolan
> Assignee: Ryanne Dolan
> Priority: Minor
>
> We've observed a misbehaving Task fail to stop in a timely manner (e.g. stuck
> in a retry loop). Despite Connect supporting a property
> task.shutdown.graceful.timeout.ms, this is currently not enforced -- tasks
> can take as long as they want to stop, and the only consequence is an error
> message.
> Unfortunately, Workers stop Tasks sequentially, meaning that a stuck Task can
> prevent any further Tasks from stopping. Moreover, after a rebalance, these
> lingering tasks can persist along with their replacements. For example, we've
> seen a Worker's "task-count" metric double following a rebalance.
> While the Connector implementation is ultimately to blame here -- a Task
> probably shouldn't loop forever in stop() -- we believe the Connect runtime
> should handle this situation more gracefully.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
