Hoi, I (finally) have jobs running on B+ and B0 cocktail, each 5k events in bfobjy. I hope they make it within the CPU limit despite objy (they are now using about 3 seconds per event on a bronco). The output has started to appear in ~ursl/d/log/sx-071802/*.log ~ursl/d/output/sx-071802/*.hbook If the information of Yury is correct (and he should know, I presume), I think we cannot run on Stream6 in MC. It's not clear to me why we can do so on data. If this is true, my proposal would be to validate the objy production of cocktail against a kanga production on identical runs, and then go on with the full production in Kanga Stream6. The Kanga validation will be quick, because all we need for that is allevents (still present at SLAC). Speak up if you have a better idea. Duplicate killing does not count (imvho). Cheers, --U.