> It's true that it would let you do that, but I very much doubt that it's
> important for the LHC VOs - you could get a similar effect just by
> having high-priority queues, and I don't think there is any use case for
> jobs to really start immediately (although there is for some other VOs).
Andrei would disagree with you. He storngly advocated this last second
priotirisation at the TCG last year. This is why he didn't even want to
hear to have a sort of VO-superscheduler. He said that since in my
scheme the pilot were still user pilots if that user jobs ended the
pilot place holder was wasted and dismissed the whole idea as non proposal.
I can dig out the email if you want.
cheers
alessandra
--
***********************************
* Alessandra Forti *
* NorthGrid Technical Coordinator *
* University of Manchester *
***********************************
|