[
https://issues.apache.org/jira/browse/FLINK-25690?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17477795#comment-17477795
]
Etienne Chauchot commented on FLINK-25690:
------------------------------------------
[~trohrmann] [~chesnay] That being said I've already done the _RetryRule_ code
because I needed it for cassandra flakiness fixes but I can comply to what you
think is best guys !
You both mentioned @after methods. So rather than allowing
_RetryRule#setExceptionHandler(Runnable)_ you prefer that we make tests author
define @after methods and call them before retrials to stay 100% junit standard
?
I checked RetryExtension, it does not supports this use case but it could be
extended very easily by implementing _AfterEachCallback_
I guess you prefer that rather than changing _RetryRule_. So I can do the
_RetryExtension_ change and update Cassandra tests to use the new
_RetryExtension_. WDYT?
> RetryRule should allow a callback to be called on exception and failure cases
> -----------------------------------------------------------------------------
>
> Key: FLINK-25690
> URL: https://issues.apache.org/jira/browse/FLINK-25690
> Project: Flink
> Issue Type: Improvement
> Reporter: Etienne Chauchot
> Assignee: Etienne Chauchot
> Priority: Major
> Labels: test
>
> RetryRule is a Flink junit rule that allows to retry tests when they
> temporary fail (either by throwing an exception or by failing an assertion)
> usually in case of load. In some cases, there is some treatment to be done
> for the test to be idempotent when retried for example cleaning of created
> data etc...
> Of course, this callback will be optional.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
