Hadoop streaming uses STDIN/STDOUT for passing the key/value pairs between the mappers and reducers, so the log messages have to be written to a specific log file - check the sample code and the python logging documentation for more details. This Query might also help.