Question: how to prevent hadoop job to fail on corrupted input file