[
https://issues.apache.org/jira/browse/CAMEL-20680?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17851357#comment-17851357
]
Jang-Vijay Singh commented on CAMEL-20680:
------------------------------------------
Initially, I had narrowed down the cause of slowness to send to partition 1, so
had removed the loop that publishes to it.
Root cause of this slowness was that although this test requested 2 partitions
for the topic, only 1 (partition 0) would actually get created before it was
starting to get used by the producer _(as this is an asynchronous step)._
Proposed change with the root cause fix are here:
[https://github.com/apache/camel/pull/14298]
* Added a wait in {{setupTopics ()}} to ensure that required number of
partitions are actually created & verified
* Restored the loop in {{publishMessagesToKafka()}} that publishes to
partition 1 (but with no significant addition to test execution time)
* Change the position of errored record from 5 to 6 _(to ensure that something
is actually read from partition 1 as well)_
After these, the test continues to execute in under 45 sec even after adding
back the loop that sends to partition 1 _(the long delay in
{{publishMessagesToKafka}} when CAMEL-20680 was created was because partition 1
was being used before it was even created!)_
As a side effect, I expect the flakiness observed till build 96 to also go away
(ref: CAMEL-20722)
> camel-kafka: reduce KafkaBreakOnFirstErrorSeekIssueIT test duration
> -------------------------------------------------------------------
>
> Key: CAMEL-20680
> URL: https://issues.apache.org/jira/browse/CAMEL-20680
> Project: Camel
> Issue Type: Task
> Components: camel-kafka
> Affects Versions: 4.4.1, 4.5.0
> Reporter: Otavio Rodolfo Piske
> Assignee: Otavio Rodolfo Piske
> Priority: Major
> Labels: help-wanted
> Fix For: 4.x
>
> Attachments: image-2024-05-01-08-13-28-056.png
>
>
> The test KafkaBreakOnFirstErrorSeekIssueIT takes a very long time to run.
> This consumes unnecessary resources from ASF CI. We should investigate how to
> improve it to it runs quickly.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)