Several `BindException` when executing in GitLab

[namero999]

Description

We hve recently upgraded from 8.4.6 to 8.5.9 and we are now experiencing some flakyness in our test execution in a GitLab worker. We didn't manage to reproduce the issue locally.

WorkerPlanEndedTest > If no variables are present, the plan is marked as failed (invalid state) FAILED
    java.lang.RuntimeException at InMemoryEngine.java:73
        Caused by: java.io.IOException at NettyServer.java:333
            Caused by: java.net.BindException at Net.java:-2

This happens both with parallel test execution on or off. Before the upgrade we always ran with parallelism on and never encountered this issue.

This behaviour seems random as retrying the pipeline usually works.

Would it be possible to get an idea of what could cause this, so that we can do some research? Looking at the code, it seems like the library is requesting a random port, however if the issue was a port already in use, the error message should be explicit about this. This stack trace does not make it clear if the issue is a port already in use or something else.

To be fair I'm not exactly sure if this issue is in this library or in camunda-spring-starter-test.

[megglos]

Hey @namero999 !

Thanks for reaching out on this, do you have the full stacktrace available? Was the cause Address already in use like we encountered as well with #1661 ?