>> The idea behind "--load" is great, but I think it's not working that >> good. I'd vote for a mechanism of delayed job-starts when "--load" >> is in use. > > I can see a delay mechanism can be useful elsewhere, but it is not > that hard for you to do yourself:
That was not what I wanted. Yes, an experienced user can easily create a script that does this, but my intension is something quite different: GNU parallel does a great job in putting high load on a system. And my question was: Shouldn't we take more care that not-so-experienced users do not overload their machines by accident. (Also see my to-be-written mail about "transfer" and NFS, and how I hit my cluster in a way that I needed to hard-reset the compute nodes to bring the head node back to life. :-)) Thomas
