shashank created CAMEL-26706:
--------------------------------
Summary: camel-core - Aggregate EIP: an AggregationStrategy whose
onCompletion throws leaves the aggregated exchange in progress, so a
recoverable repository never recovers it
Key: CAMEL-26706
URL: https://issues.apache.org/jira/browse/CAMEL-26706
Project: Camel
Issue Type: Bug
Components: camel-core
Reporter: shashank
Assignee: shashank
When a group completes, {{AggregateProcessor.onSubmitCompletion}} first books
the aggregated exchange as in progress ({{inProgressCount}} and
{{inProgressCompleteExchanges}}), then calls
{{aggregationStrategy.onCompletion(exchange)}}, and only after that adds the
{{AggregateOnCompletion}} that undoes the booking. If {{onCompletion}} throws,
nothing undoes it: the aggregated exchange is never sent, never failed and
never confirmed.
By then the group has been removed from the aggregation repository. A
recoverable repository (such as {{JdbcAggregationRepository}}) has moved it to
its completed store, but its recover task skips it on every run as "already in
progress", so the group is neither delivered nor recovered until the
CamelContext is restarted. The in progress count also stays above 0, so a stop
with {{forceCompletionOnStop}} or {{completeAllOnStop}} waits until the
shutdown timeout. With {{completionFromBatchConsumer}}, the other groups
completed in the same batch are not submitted at all and are lost the same way.
{code:java}
from("direct:start")
.aggregate(header("id"), strategyWhoseOnCompletionThrowsOnce)
.aggregationRepository(recoverableRepository).completionSize(2)
.to("mock:aggregated");
// send A1 and A2 with id A: the sender of A2 gets the exception from
onCompletion,
// and A1+A2 never reaches mock:aggregated, although it is in the completed
store
{code}
The javadoc of {{onCompletion}} says it must not throw. Even so, CAMEL-26575
made Multicast, Split and Recipient List fail the exchange in that case instead
of hanging it, and CAMEL-26309 fixed the same lost group for a rejected thread
pool task ({{onRejectedCompletion}}), but the throw happens before the on
completion is registered, so that fix does not cover it.
Expected: the aggregated exchange fails as when the thread pool rejects it (its
on completions run, so it is no longer in progress and is not confirmed, and a
recoverable repository recovers it), it is passed to the exception handler, and
the sender still gets the exception. The other groups of a batch are still sent.
_Filed with Claude Code on behalf of allthingssecurity._
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
