gnodet opened a new pull request, #30148:
URL: https://github.com/apache/camel/pull/30148
## Summary
Fix `OAuthCodeFlowServletTest` and `OAuthClientCredentialsServletTest`
failing with
`BindException` and `TestAbortedException` (skip instead of proper failure)
when both
tests run in the same JVM fork.
## Root Cause
Two bugs in `AbstractKeycloakTest`:
1. **Shared static port**: `AvailablePortFinder.find()` was called once as a
static
field initializer in the _base class_ (`AbstractKeycloakTest`). Static
initializers
execute once per class-load in the JVM — so all subclasses shared the
same `port`
value. When `OAuthClientCredentialsServletTest` and
`OAuthCodeFlowServletTest` ran
in the same Surefire fork, the first class to call `@BeforeAll`
successfully bound
Undertow to that port; the second class then failed with `BindException`
because the
same port was still occupied.
2. **Undertow started before Keycloak assumption**: Both servlet test
classes started
the Undertow server unconditionally in `@BeforeAll`, then checked
`Assumptions.assumeTrue(admin.isKeycloakRunning())` only inside
individual test
methods. When Keycloak was absent this caused a hard `RuntimeException`
(`BindException`) instead of a clean `TestAbortedException` skip.
## Fix
- Remove the shared `SERVER_PORT` / `port` / `APP_BASE_URL` static fields
from
`AbstractKeycloakTest`.
- Add `protected abstract int getPort()` to the base class, with
`getAppBaseUrl()`
computed from it.
- Each concrete test class (`OAuthCodeFlowServletTest`,
`OAuthClientCredentialsServletTest`, `OAuthCodeFlowVertxTest`,
`OAuthClientCredentialsVertxTest`, `SSLCertTrustTest`) now declares its own
`static final AvailablePortFinder.Port SERVER_PORT =
AvailablePortFinder.find()`
and implements `getPort()`. Each class gets a distinct port even when
running in
the same JVM fork.
- Change `createUndertowServer()` to accept an explicit port parameter.
- For servlet tests: check
`Assumptions.assumeTrue(admin.isKeycloakRunning())` in
`@BeforeAll` _before_ calling `createUndertowServer()`, so Undertow is
never started
when Keycloak is absent and skips are clean.
## Tests
No new tests added — the fix is in the test infrastructure itself. The issue
was that
the existing tests failed with `BindException`; after this fix each class
binds to its
own distinct port and the Keycloak skip propagates cleanly via JUnit's
`@BeforeAll`
assumption mechanism.
Fixes CAMEL-23480
---
_Hermes Agent (Claude Sonnet 4.6) on behalf of Guillaume Nodet_
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]