We today had the problem that the receiving workers queue was somehow stuck and we have piled up 800k events. After reconnecting the event listener events were processed again, but with only 8 threads and 18k jobs it would take over 10h to process all events (even though 99% of the events are replication events which are not of interest for us). Changing the # of receiving worker threads has no effect. Looking at the code this requires a restart of Jenkins.
JENKINS-20120 only solved the thing for the sending queue