[
https://issues.apache.org/jira/browse/CAMEL-26791?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Claus Ibsen reassigned CAMEL-26791:
-----------------------------------
Fix Version/s: 4.24.0
Assignee: Claus Ibsen
> camel-chaos - an expectation that can judge a duplicate fault: were the
> duplicates processed twice
> --------------------------------------------------------------------------------------------------
>
> Key: CAMEL-26791
> URL: https://issues.apache.org/jira/browse/CAMEL-26791
> Project: Camel
> Issue Type: Improvement
> Components: camel-jbang
> Reporter: Claus Ibsen
> Assignee: Claus Ibsen
> Priority: Major
> Fix For: 4.24.0
>
>
> Found in the recorded runs of {{camel cmd chaos-explore}} (CAMEL-26760). The
> AI tried the hypothesis "when payment receives duplicates, refunds should
> handle them safely without double-processing" with
> {{endpoint=http://localhost:8080/payment duplicate rate=0.1
> expect=refunds:failed<=0}}. It passed, but 8 duplicate calls went through:
> the payment route completed 69 exchanges for 41 orders and 20 refunds, so 8
> payments or refunds were done twice. No expectation of camel-chaos can see
> that: {{failed}}, {{completed}}, {{handled}} and the times all look fine when
> a duplicate is processed.
> A design question rather than a bug. Chaos knows which sends it duplicated
> (it fired 8 times). When the receiver is a route of the same integration
> (here the payment route, or a stubbed endpoint consumed by a route), chaos
> could count how many of the duplicates that route completed, for an
> expectation such as {{expect=payment:duplicates<=0}} (processed twice: none),
> met when an idempotent consumer, or a check of the route, dropped them. When
> the receiver is outside the integration, chaos cannot tell, and the rule
> could say so.
> Open points: how to recognise the duplicate in the receiving route (an
> exchange property or header chaos sets on the copy, which crosses HTTP only
> as a header), and whether the suggestions of {{chaos-suggest}} should pair a
> duplicate fault with this expectation.
> _Claude Code on behalf of Claus Ibsen_
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
