Probably the case since it was implemented. The task updates the status for 30 jobs at the time using this condition:
pr.codeRan is false and not(buildStatus = 'None')

This is problematic, because a job may have codeRan set to false, while buildStatus set to CompilationFailure, RunningFailure, Killed or Canceled, which are valid states for a terminated job. I.e., it tries to update the same 30 jobs over and over again, which already have been terminated.

Submitted by Elmer van Chastelet on 18 March 2015 at 09:52

Also added statistics about how many of these jobs are scheduled in administration page.