LHC Computer Grid - Rollout
> [mailto:[log in to unmask]] On Behalf Of Dr D J Colling
said:
> The point that you make is a very interesting one, if obvious once you
> think about it. Jobs that run for a long time and then fail
> really can
> waste a lot of (valuable?) cpu time.
This is of course one of the arguments for passing the expected cpu time
through to the LRMS, at the moment if a job gets stuck in some way it
blocks the CPU until the maximum time for the queue runs out, which may
be far longer than the time the successful jobs take.
Stephen
|