As more jobs run on a Hudson installation, it gets increasingly painful to find
the right moment to redeploy Hudson. This is particularly a problem for myself,
because I run a very large Hudson farm, and I need to frequently deploy a new
version of Hudson to do the field testing.
Hudson should be able to enter the ordinary shutdown mode, where it queues up
jobs but never start executing the new one. It should also remember the list of
jobs that need to be rescheduled when Hudson comes back next time.
Actually stopping the service and redeploying is container/environment specific,
so some part of this has to be done outside the container.