Testbed Support for GridPP member institutes > [mailto:[log in to unmask]] On Behalf Of Gustav Wikström said: > A more general question: when a job exceeds memory limits, shouldn't > it fail instead of being cancelled? At least then the user could get > the reason for the failure. It looks like the EMI version of CREAM should have somewhat better messages when the job gets some signals: https://savannah.cern.ch/bugs/?51703 Stephen