The problem occurs randomly, but fairly frequently; see for example https://ci.hibernate.org/blue/organizations/jenkins/hibernate-search/detail/master/464/pipeline/85

It seems related to how we schedule the next runnable in the executor.