[
https://issues.apache.org/jira/browse/CAMEL-24981?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18118595#comment-18118595
]
Claus Ibsen commented on CAMEL-24981:
-------------------------------------
PR: https://github.com/apache/camel/pull/26812
_Claude Code on behalf of Claus Ibsen_
> camel-core - Error handler keeps the onException of an earlier redelivery
> attempt when the exception type changes
> -----------------------------------------------------------------------------------------------------------------
>
> Key: CAMEL-24981
> URL: https://issues.apache.org/jira/browse/CAMEL-24981
> Project: Camel
> Issue Type: Bug
> Components: camel-core
> Reporter: Claus Ibsen
> Assignee: Claus Ibsen
> Priority: Minor
>
> If the exception changes type between redelivery attempts, the error handler
> keeps using the onException settings chosen for the first exception.
> h3. Reproducer
> {code:java}
> errorHandler(deadLetterChannel("mock:dead"));
> onException(IOException.class).maximumRedeliveries(1).handled(true).to("mock:io");
> from("direct:b").process(e -> {
> if (attempt.getAndIncrement() == 0) {
> throw new IOException("first");
> }
> throw new IllegalArgumentException("second");
> });
> {code}
> The first attempt fails with IOException and is redelivered. The redelivery
> fails with IllegalArgumentException, which no onException matches, so it
> should go to the dead letter channel. Instead it is routed to mock:io and
> marked handled by the IOException onException.
> h3. Cause
> RedeliveryErrorHandler.RedeliveryTask.handleException only overwrites the
> failure processor, handled/continued/retryWhile predicates,
> useOriginalMessage/Body, onRedelivery/onExceptionOccurred processors and
> redelivery policy when an exception policy matches. It never resets them to
> the error handler defaults first, so values from an earlier attempt's policy
> carry over. The redelivery policy is also derived from the previous attempt's
> derived policy instead of the error handler's own policy. This behaviour
> dates from 2009, and nothing in history or JIRA says it was intended.
> h3. Fix
> Reset the per-attempt state to the error handler defaults before applying the
> policy that matches the current exception.
> _Claude Code on behalf of Claus Ibsen_
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
