[
https://issues.apache.org/jira/browse/CAMEL-24994?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Claus Ibsen resolved CAMEL-24994.
---------------------------------
Fix Version/s: 4.23.0
Resolution: Fixed
Done in https://github.com/apache/camel/pull/26850 (merged 2026-09-26), in
Camel 4.23.0.
_Claude Code on behalf of Claus Ibsen_
> camel-core - Enrich and Poll Enrich EIPs: the resource on completions are
> lost when the enrich fails
> ----------------------------------------------------------------------------------------------------
>
> Key: CAMEL-24994
> URL: https://issues.apache.org/jira/browse/CAMEL-24994
> Project: Camel
> Issue Type: Bug
> Components: camel-core
> Reporter: Claus Ibsen
> Assignee: Claus Ibsen
> Priority: Major
> Fix For: 4.23.0
>
>
> A deep review of the Content Enricher EIPs (enrich / Enricher and pollEnrich
> / PollEnricher) found the bug below in both. It was reproduced against
> 4.23.0-SNAPSHOT.
> *The on completions of the resource exchange are lost when the enrich fails.*
> The on completions (synchronizations) registered on the resource exchange
> only ran when the enrichment succeeded. When the resource failed, or when the
> aggregation strategy threw an exception, they never ran.
> * pollEnrich from a file: when the aggregation failed, the polled file's on
> completion (commit/rollback, and removal from the in-progress repository) did
> not run. So the file stayed marked as in progress, and it was not polled
> again until the application was restarted.
> * enrich: when the enrich failed, the on completion that a producer registers
> to release a resource did not run. The http producer, for example, closes the
> response stream this way.
> When the resource was used (aggregated, or discarded by the aggregation
> strategy), its on completions are handed over to run together with the
> original exchange. When the resource or the aggregation failed, they now run
> right away as failed, so a polled file is rolled back and can be polled
> again. They are not handed over on the failure paths, as the exchange may
> still recover (for example by redelivery), and a file whose content was never
> used would then be committed.
> _Claude Code on behalf of Claus Ibsen_
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
