[
https://issues.apache.org/jira/browse/IGNITE-29149?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Nikita Amelchev updated IGNITE-29149:
-------------------------------------
Component/s: extensions
> [Extensions] Fix flaky testWithExpiryPolicy and metadata consumer leak in
> Kafka to Ignite streamer
> --------------------------------------------------------------------------------------------------
>
> Key: IGNITE-29149
> URL: https://issues.apache.org/jira/browse/IGNITE-29149
> Project: Ignite
> Issue Type: Bug
> Components: extensions
> Reporter: Nikita Amelchev
> Assignee: Nikita Amelchev
> Priority: Minor
>
> Kafka to Ignite replication in tests can take longer than the entry TTL in
> testWithExpiryPolicy because applier threads poll their partitions in turn
> with the default consumer poll timeout, so the entry expires before it
> reaches the destination cluster and the test hangs until the timeout.
> Besides, KafkaToIgniteMetadataUpdater never closes its Kafka consumer, so
> each stop of the streamer inside a running JVM leaves the consumer with its
> connection and heartbeat thread alive, and such consumers pile up from test
> to test.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
