Major Description
In OAK-4368 a test LucenePropertyIndexTest#longRepExcerpt for measuring improved efficiency of excerpt extraction was introduced, however it's too aggressive as it tests rep:excerpt on a property whose text has a very high number of excerpt matches (500) over 6k nodes.
While this has always succeeded on my local machine it seems that Jenkins CPU capabilities are more limited and therefore it tends to fail every now and then.