LHC Computer Grid - Rollout
> [mailto:[log in to unmask]] On Behalf Of Mattias Wadenstein
said:
> First a question: Does jobs in general provide a walltime for
> LRMS, or is
> it assumed that the queue default time given is large enough to handle
> long-running production jobs?
Jobs themselves don't request any limits, they take the defaults for the
queue. Usually sites configure wall time limits somewhat larger than the cpu
limit - a factor 2 might be appropriate, although some sites use less. The
actual time limits are up to you, but you should probably allow at least a
24-CPU-hour job to run.
Stephen
|