[
https://issues.apache.org/jira/browse/KAFKA-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17072032#comment-17072032
]
ASF GitHub Bot commented on KAFKA-9754:
---------------------------------------
scott-hendricks commented on pull request #8396: KAFKA-9754 - Trogdor - Ignore
produce errors, better error statistics.
URL: https://github.com/apache/kafka/pull/8396
This change adds the following features to the Trogdor ProduceBenchWorker:
Ability to ignore errors caused by SendRecords.
Tracking of the number of errors ignored.
Additional tracking of the number of callbacks with exceptions.
In order to maintain backward compatibility, ignoring errors is disabled by
default.
All build-time integration tests passed:
BUILD SUCCESSFUL in 42m 8s
334 actionable tasks: 255 executed, 79 up-to-date
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
> Trogdor workloads fail because topics are not fully created when workloads
> start.
> ---------------------------------------------------------------------------------
>
> Key: KAFKA-9754
> URL: https://issues.apache.org/jira/browse/KAFKA-9754
> Project: Kafka
> Issue Type: Bug
> Components: system tests
> Reporter: Scott Hendricks
> Assignee: Scott Hendricks
> Priority: Major
>
> In many instances, when running with high partition counts, trogdor
> ProduceBench workloads have been failing with the following error:
> {noformat}
> This server does not host this topic-partition. {noformat}
> This is caused by the test not checking that the topic is fully created
> before starting to produce to it.
> As far as I see it, there are 2 options here:
> # Check the topic status at the beginning of the test and wait for the topic
> to be in a healthy state.
> # Add exception handling around the SendRecords call to allow for ignoring
> these failures and continuing.
> #2 would be easier than #1, however they both have trade-offs.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
