After upgrading to Jenkins 2.492.2 LTS and pipeline job v1508 we began seeing a memory leak requiring us to restart the controller 3 times in a 10 day period.  After some troubleshooting, and heap dump analysis, we discovered a specific job execution that had long since been aborted nearly 2 years before that was continuing to attempt to restart many times.  The build had long since been removed from the jenkins UI.

Prior to 2.492.2 release, the job log indicates that jenkins would attempt to restart the job once after a the service restarted.  After this update, Jenkins is attempting to start the job about 160x per day, eventually filling the heap.

Although we discovered this in workflow-job-plugin v1508, we are wondering if it was introduced in v1505, as some changes to the onload/reload behavior were introduced.  (#510) jglick ?