Last modified: 2014-05-09 07:48:48 UTC
Doing so could help identify machines that are struggling for whatever reason.
It may be interesting to track and call out in the logs not the "last" but the most deviant times (> N standard deviations from mean or Nth centile). Depending on how hosts get added to the run queue the last one or two may seem to be taking a longer time to complete but really not be deviant.