Our failure rate on recent Hudson runs has been pretty bad, but I believe that all of these failures are because our jobs are running out of memory. I've modified the job configurations to increase the memory, so hopefully this will fix the problem. Best regards, Randall