This is Atlas using our prototype Grid as production site. They also see failing jobs. I redirected those output to /scratch. I think Booker's suggestion (wipe it out) is the only thing we can do. Wei Yang | [log in to unmask] | 650-926-3338(O) Booker Bense wrote: > On Sun, 23 Jul 2006, Renata Maria Dart wrote: > >> Hi, the monitoring queue has a couple hundred "/tmp 100%" messages for >> various nomas. The problem seems to be caused in each case by user >> osgusatlas000007 who has written a 900+mb directory named >> Panda_Pilot_xxxxxxx. Is there a way to determine who >> osgusatlas000007 really is and get them to copy what they need >> elsewhere? > > Yes, Wie can provide the answer to this and I can with a little > poking around. > >> Or should we just wipe out their stuff? > > Wipe it out if it's causing problems. The Panda stuff needs to > be smart enough to pick the correct tmp directory. It is not > anything that needs to be preserved. > > _ Booker C. Bense >