The CI (running now mostly for main branch and for 5.1) recently started showing intermittent failures during in-container runs with RequestScopeEventMessageDeliveryTest - in what could be 30-50% of runs on either JDKs.
This is no new thing and has happened before, although much more rarely.
If memory serves, the test passes locally just fine so it's some timining/timeout issue.
We should try to tackle the test or perhaps re-write it to have the CI green; we could also temporarily disable the test and keep this issue open as a tracker if the CI gets too annoying.