On 2016-11-17 15:24, Alessandra Forti wrote: > VAC (and other cloud resources) are a problem that needs to be solved. > It is not clear to me why you are not running ATLAS jobs though. Let's ask Andrew about that - I'll take it to him offline. > It might also account why Liverpool in October had a bigger discrepancy > than in the previous two months (you run atlas jobs) I'v sent you my theory on that - it's the same thing (algorithm that makes invalid use of composite/average figure that does not match HS06 actually delivered). Let's see if it goes right now that ATLAS have resumed normal service. If it doesn't improve (at least) I've promised to eat my hat... > It seems also a bit of a waste that you put your most powerful nodes in > VAC if it > isn't used by all the experiments. It is the amount of HS06 that I consider important, not where that HS06 is concentrated. Providing the same HS06 on different nodes would not get more work through. We are obliged by the PMB to dedicate a significant amount of HS06 as a VAC scale test. I have no say in it; Dave Colling and Andrew McNab made the request and that's that, I'm afraid. So, I'll try to find out from Andrew why so few ATLAS jobs come through VAC. Cheers, Ste