[
https://issues.apache.org/jira/browse/CAMEL-25275?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Claus Ibsen resolved CAMEL-25275.
---------------------------------
Resolution: Fixed
> camel-consul - the key/value and event consumers stop watching for good after
> one failed query, while the route stays started
> -----------------------------------------------------------------------------------------------------------------------------
>
> Key: CAMEL-25275
> URL: https://issues.apache.org/jira/browse/CAMEL-25275
> Project: Camel
> Issue Type: Bug
> Components: camel-consul
> Reporter: shashank
> Assignee: shashank
> Priority: Major
> Fix For: 4.23.0
>
>
> The {{consul:kv}} and {{consul:event}} consumers watch with a chain of
> blocking queries: the callback of each query
> ({{ConsulResponseCallback.onComplete}} /
> {{EventResponseCallback.onComplete}}) processes the answer and starts the
> next query. The failure callback only reports the exception:
> {code:java}
> @Override
> public void onFailure(Throwable throwable) {
> onError(throwable); // exception handler, nothing else
> }
> {code}
> So when a single query fails (the Consul agent restarts, a network error, a
> 500 while the Consul cluster has no leader, a read timeout), no further query
> is started: the consumer receives no change anymore, for good, while the
> route stays started. Only one WARN line from the exception handler shows it.
> h3. Reproduction
> Unit tests with a mocked {{Consul}} client ({{consulClient=#consul}}):
> * {{ConsulKeyValueWatchTest}}: the first {{getValue}} fails with
> {{ConsulException}}: main never calls {{getValue}} again ({{Wanted 2 times
> ... But was 1 time}}); with the fix the second query answers and the value
> reaches the route.
> * {{ConsulEventWatchTest}}: the same for {{listEvents}}.
> h3. Proposed fix
> * {{ConsulKeyValueConsumer}}: after a failure, query the key again after
> {{blockSeconds}} (at least one second, so that an agent that is down is not
> queried in a loop), on a single-thread scheduler of the consumer (started and
> shut down with it, as {{ConsulEventConsumer}} already does).
> * {{ConsulEventConsumer}}: after a failure, schedule the next query as after
> an answer (its {{watch()}} already waits {{blockSeconds}}, see CAMEL-12418).
> Both only while the consumer runs. With the fix the two tests pass and the
> camel-consul unit tests pass (6; the 27 ITs need Docker). {{mockito-core}} is
> added as a test dependency.
> Affected: 4.14.x, 4.18.x and main (same code since 2017).
> Duplicate check (2026-10-02): JIRA component camel-consul and text "consul"
> (90 issues): watch related are CAMEL-12418 (CPU load, the scheduling delay),
> CAMEL-12511 (NPE on start), CAMEL-10073; none about failures. GitHub pull
> requests "consul consumer": none.
> _Filed with Claude Code on behalf of allthingssecurity._
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
