[
https://issues.apache.org/jira/browse/CAMEL-26468?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18136310#comment-18136310
]
Claus Ibsen commented on CAMEL-26468:
-------------------------------------
Merged the first diagnostic step to main in
https://github.com/apache/camel/pull/28800 (commit 47459c6eeb38):
CxfProducerTest and its lifecycle methods now time out after 30 seconds with a
thread dump of every thread. The next hang on CI will fail only that test and
show where it was stuck. This issue stays open until the cause of the hang is
fixed.
_Claude Code on behalf of davsclaus_
> camel-cxf-soap - CxfProducerTest sometimes hangs on CI until the fork timeout
> -----------------------------------------------------------------------------
>
> Key: CAMEL-26468
> URL: https://issues.apache.org/jira/browse/CAMEL-26468
> Project: Camel
> Issue Type: Test
> Components: camel-cxf
> Reporter: Claus Ibsen
> Assignee: Claus Ibsen
> Priority: Major
>
> The CxfProducerTest family in camel-cxf-soap (CxfProducerTest, and
> CxfProducerContextTest / CxfProducerOperationTest which extend it) sometimes
> hangs on CI. The module has a 300 s fork timeout
> (camel.surefire.forkTimeout), so the whole fork is killed with "There was a
> timeout in the fork" and the PR build fails.
> Seen in PR builds since 2026-10-24 (the test that never logged a result):
> * 2026-10-24: dependabot plugin-tools 3.16.0 (no unfinished class visible)
> * 2026-10-28: CAMEL-26364-mcp-docs-naming - CxfProducerContextTest
> * 2026-10-28: CAMEL-25259 dsl-extensions design - CxfProducerContextTest
> * 2026-10-31: PR #28652 (CAMEL-25231) - CxfProducerContextTest
> * 2026-11-01: PR #28788 (CAMEL-26436), twice - CxfProducerTest
> The other ~68 test classes in the module take about 33 s in total, so this is
> a hang, not a slow module. It predates PR #28676 (per-test ports, merged
> 2026-10-30).
> There is no thread dump, because surefire kills the fork. As a first step,
> give the CxfProducerTest tests and lifecycle methods a timeout in a separate
> thread and turn on JUnit's thread dump on timeout
> (junit.jupiter.execution.timeout.threaddump.enabled). A hang then fails only
> that test (surefire retries it), and the log shows where it is stuck, so the
> real cause can be fixed.
> _Claude Code on behalf of Claus Ibsen_
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
