> If it's a shared partition it may not be lhcb files which are the problem. > My money would be on atlas, who seem quite good at it, they apparently > filled 1.7 Tb at NIKHEF ... Sorry for the delay - I was in a meeting earlier. Correct, it's 99.7% Atlas. On an admittedly small SE (I'm working on it). I'm not clear that it's my responsibility to clean up after experiments, but since it's turned into a talking point I've just done a find atlas -atime +31 | xargs rm which has had the desired effect! What's the correct reporting route for this sort of experimental (mis)use? Does each experiment's administrator have an email address tucked away in a file somewhere? Lawrie Lowe.