Following Chuck's suggestion, we now have our first successful job from an IFZ batch node. It used SLAC's CondDB replica. See http://gridui02.usatlas.bnl.gov:25880/server/pandamon/query?overview=wnlist&type=production&hours=12#PROD_SLAC The Panda ID is 299638, batch node is noma0490 and LSF job id is 132050. The trick is to let kernel(iptables) modify TCP packages going to atlasdbdev.cern.ch so that they land on atl-dbrep.slac.stanford.edu. Should this be used as a general method on all Atlas batch nodes to work around the limitation of JT-11.0.X.X ? -- Wei Yang | [log in to unmask] | 650-926-3338(O)